1. Introduction
The application of image processing for mineral detection is crucial for the development of the mining field throughout the world [1]. It is not only important for the automation of mining processes, but also provides the basis for subsequent processes for improving ore quality [2]. Ore-production capacity can be improved by applying image-processing technology to each link of the mining process. Therefore, applications of image processing to the mining field have theoretical significance and practical application value. Some scholars have applied image-processing technology to various research and experiments [3], developed algorithms for mining research, and achieved good results, proving the feasibility and effectiveness of the technology in the mining field [4,5]. In this paper, we present a comprehensive literature review on image-processing methods and their applications in the mining field. This paper focuses on the current state and future trends of research on image processing in the mining field. It can serve as a reference for researchers working in this field.
At present, the most advanced image-processing methods for mining research include convolutional neural networks [6] and deep learning [7], which have been popular subjects for research in recent years. Image processing is primarily a data-based modeling method that uses images or videos, rather than process measurements, as input to predict other measurements. Although TensorFlow [8] and OpenCV [9] are currently the most popular research tools and methods for image processing, they have not been applied in mining research. They can be combined with mining research to achieve ore sorting, classification, particle filtering [10], etc. They can not only improve the efficiency of scientific research but also improve industrial production capacity.
Although widespread applications of machine learning techniques appear in the mineral processing literature, a key element that has not been explicitly discussed is data quality. The quality of the data used for machine learning techniques is critical to the validity of the methods’ applications, and a large quantity of mineral resource data are not used as data sources for machine learning techniques in mineral processing.
At present, image processing has not been applied in the relatively cutting-edge area of biomineral research. Image processing has the characteristics of multidisciplinary theory and technology integration. It has the advantages of low cost, low pollution, and applicability to a wide range of processing objects (not limited to specific minerals). It is the future of mineral processing and one of the important directions for development. The application of image-processing technology in the process of bio-beneficiation can improve the research efficiency.
2. Data and Methods
2.1. Data
Software helps scientists identify research problems, analyze data, visualize results, and disseminate knowledge [11]. However, the academic value of software has long been underestimated, or even neglected, by the current publication-oriented reward system. Recently, a huge increase has been seen in the software that is freely available for academic use [12]. With increasing recognition of the value of data and such software packages [13,14], some academics have argued that software should also be considered as an academic contribution [15,16]. Since 2013, the National Science Foundation has recognized software as a valid research tool. Software was also listed as an academic contribution by the 2014 UK Research Excellence Framework. However, several funding agencies, policymakers, and managers do not regard software as an effective research tool [16].
Chen (2006) used CiteSpace to analyze and summarize related works in the literature with the aim of providing a reference for scholars and engineers. CiteSpace is a software developed for citation visualization and analysis that uses data visualization and metrology to analyze research literature for potentially useful knowledge. CiteSpace is based on co-citation analysis and the Pathfinder network algorithm. It measures the literature (i.e., the set) for a specific field to find the critical path and knowledge inflection points in the evolution of subject fields [17]. A series of visualization maps can be generated to analyze the dynamics for the evolution of the field and explore the frontiers of recent advances. The structure, rules, and distribution of the scientific knowledge are presented via visualization; therefore, the obtained results are called a scientific knowledge map. This map is a graph that shows the development, the status quo, the emerging disciplines, and the internal and external structures of the scientific knowledge for a specific field.
Scientific knowledge is based on scientometrics and involves the interdisciplinary fields of science, applied mathematics, information science, and information metrology [18]. In contrast to a traditional literature review, a scientific knowledge map uses visualization analysis to present scientific knowledge more intuitively and systematically, making it easier to understand and grasp the direction of future development. This method has been widely used in medicine, engineering, psychology, management, education, economics, and many other disciplines to achieve remarkable results [19]. However, it has rarely been used in the mining field.
We generated statistics on the studies that have used CiteSpace to prove its effectiveness. We selected the Web of Science (WoS) database and three Chinese full-text journal article databases: CNKI, Wanfang, and CQVIP. We used the search term “CiteSpace” with different letters capitalized (e.g., “citespace” and “Citespace”). The time span was limited to 2004–2021 to coincide with the development and spread of CiteSpace. We then retrieved and downloaded the full text of the obtained articles. We carefully screened the articles to ensure that they were CiteSpace-based and found 9710 articles. We found that the number of papers using CiteSpace has continued to increase, especially in recent years. This helped us achieve the purpose of our research and obtain convincing results. Table 1 presents research fields containing ≥ 20 papers that used CiteSpace. It is expected that CiteSpace will be increasingly applied in the mining field [20].
2.2. Method
In this study, the data included literature collected from the Science Citation Index Expanded (SCIE) and Scopus. Data were retrieved on 1 January 2022 from the WoS (using only SCIE and CPCI-S) database and Scopus, which was created by Clarivate Analytics (USA) in 1988. The main search terms were “image processing” and “minerals”. Related keywords, such as “machine vision”, “rock”, and “flotation foam”, were also used to ensure the integrity of the analysis data. The search period was from 1 January 1988 to 1 January 2022. In total, 6135 items were collected. To ensure the accuracy of the results, the details of each paper were carefully filtered: each paper needed to contain two keywords and had to be about the application of image processing in mining. Finally, 2173 records were selected to establish a database for this study.
3. Image Processing and Applications in the Mining Field
In the late 1970s, Marr proposed using computer vision to analyze and calculate images [21]. Image processing refers to converting an image signal into a digital signal to improve the visualization for humans. Common processing methods include image enhancement, image segmentation, edge extraction, shape analysis, and image compression. The basic process is to use an image acquisition card and camera to convert an external image into a digital image represented by red (R), green (G), and blue (B) channels. Then, the data are analyzed, processed, and output using an appropriate software. Table 2 lists major image processing methods and algorithms. As shown in Table 2, the main commonly used methods of image processing include image transformation, image enhancement and restoration, image segmentation, image description, and image classification (recognition). These methods can complete a series of image processing and applications through different algorithms, whether in scientific research or in an industrial field. We used these methods to achieve the purpose of our research. The specific meaning, methods, and main features of image processing are described in Table 2.
3.1. Applications of Image Processing in the Mining Field
Although machine vision technology has long received the attention of scholars at home and abroad, its development did not start until the 1990s [28]. With the rapid development of electronic and computer hardware in the past two decades, machine vision technology is gradually being applied in various fields [29]. Many foreign research institutions began to invest money and resources into the application of machine vision to the mining process in the 1970s and 1980s [30], which allowed them to realize technological advances and to accumulate technical experience. Currently, machine vision technology is used in many production processes in the mining industry, including crushing and grinding, flotation of minerals [31] and coal mud, and online detection of product quality [32], product granularity, and coal-dust particles [33]. Ore sorting, particle-size detection, and mineral flotation are three major aspects of mining. Thus, we focused on the applications of image processing to these three activities.
3.2. Image Processing for Ore Sorting
Most mines, both at home and abroad, separate ore via the traditional manual sorting. Although the research on automating the separation process is increasing, current methods have low efficiency, low accuracy, and high cost. To address these issues, researchers have considered applying image-processing techniques for the automation of ore-sorting equipment, such as light-based separators. Some degree of automation has been realized regarding ore separation. Table 3 presents relevant research that focused on the applications of image processing for ore sorting. It can be concluded that the ore-sorting equipment in the early 20th century is the optical sorting machine from the UK. In the 1970s, the mineral concentrator was developing the M16 type photoelectric sorting machine. After 2000, the era of automation developed rapidly. Sorting equipment based on machine vision and deep learning has become the mainstream method of the times; sorting efficiency and accuracy have reached certain standards, effectively improving the industrial and scientific research value of image processing.
3.3. Image Processing for Particle-Size Detection
In mining, the granularity of the rock and ore is important for determining the effectiveness of the crushing processes. Thus, obtaining the granularity of rock and ore in real time and with high accuracy is important for automating the process. Digital image processing has been applied to particle-size detection in industrial processes [42]. This detection method can adapt to various environments, does not need manual contact, and reduces the need for labor, because most of the work is performed using a computer. It can achieve real-time detection and, thereby, improve the efficiency of crushing processes and promote automation. Table 4 presents several studies that focused on the application of image processing to particle-size detection in the mining industry. It can be noted that Australian scientist Julius Kruttschnitt used optical instruments to measure the string length of ore particle size to obtain particle-size information in 1976. Although there were major defects, this was a breakthrough for online particle-size detection. In 1988, William Petruk et al. analyzed the particle-size composition of lead–zinc ore before and after crushing,. This was the earliest time that image processing technology was formally applied in the field of mineral processing, setting a model for the subsequent application of image processing in the field of mineral processing. In 1996, a system was developed to describe the shape of concave–convex particles based on image analysis. In 2005, the process of predicting the grain size distribution of rocks by computer vision was realized. In 2015, Lalit et al. used image processing technology and random distribution technology to analyze 3D images of mineral particles, indicating that image-processing technology has gradually moved from a macro-size study of coal to a micro-size study. The appearance of image-processing technology is the turning point of mineral particle-size detection.
3.4. Image Processing for Mineral Flotation
Froth flotation is a widely used method to perform ore-dressing operation for mineral extraction. The dosage of flotation reagents in the flotation tank can be controlled to cause different types of minerals to selectively adhere to bubbles, based on mineral surface wettability differences for easy separation [56]. Image processing can be used to monitor the flotation process, and a key point is accurately correlating the visual characteristics of the foam surface to the process conditions [57]. This requires an appropriate algorithm for extracting the visual features of the foam surface [58]. The obtained foam image is processed using an appropriate method to extract the visual characteristic parameters of the foam surface that are closely related to the flotation condition [59]. Realization of the condition-perception function, similar to human observation of the foam surface, is the prerequisite for the successful implementation of the flotation process monitoring system based on machine vision. Researchers have extensively studied the application of image processing to extract visual features from the foam surface, such as the foam color, bubble morphology, foam texture [60], and foam dynamic characteristics [61] (e.g., flow rate and stability). Table 5 lists some specific algorithms used for this purpose, together with their characteristics. It can be concluded that each feature shows different properties of foam, each feature has its own algorithm, and the expression accuracy and characterization points of different algorithms are different. At present, a good deal of research is based on the improvement of a variety of algorithms, and the research into algorithms is one aspect of the current research on flotation foam.
In general, feature extraction methods are mainly used in the study of ore and the classification of particle size. The most common techniques are the gray level co-occurrence matrix (GLCM), the Gabor filter, and wavelet texture analysis. These methods are more widely used in flotation foam under monitoring [70]. The analysis and monitoring of flotation froth are the topics that receive the most attention, mainly due to the establishment of a performance link between grade and recovery between the appearance of flotation froth and flotation [71], the most commonly used method of which is the extraction of texture features. This method is closely related to parameters such as bubble size distribution, solid attachment and foam stability, as well as to key properties such as grade and recovery. However, after reading a large amount of literature and reviewing relevant statistics, we found that most of the applications consist of research based on experimental data, and only small parts of the applications are used in the trial operation stage in factories.
Figure 1 shows the main typical input and output process (information flow) of image-processing research in the mining field. In addition, Table 6 shows the current problems of image processing in the study of ore sorting, ore particle size, and flotation froth. It is hoped that this article’s reference methods and recommendations for the application of image processing in the mining sector will be useful to plant engineers and researchers, so they can quickly identify the methods and processes that have been applied and, thus, determine which are the most promising. technologies to face the challenges.
4. Scientific Measurement and Analysis
4.1. Rate of Publication
Figure 2 shows the statistical results for image-processing applications in the mining field from 1988 to 2021. The earliest formal application of image-processing technology to the mining process was [72], which analyzed the particle-size composition of lead–zinc ore before and after crushing. After 2010, neural networks were combined with image processing for application in the mining field. For example, Ko and Shang (2010, 2011) considered using a delayed neural network to predict the feed quantity of a semiautomatic rolling mill. The number of papers published on image-processing applications in the mining field was below 100 per year before 2014 and increased steadily from 2014 to 2021. In 2019, the number of papers reached more than 200. Advances in machine vision led to its increasing application in mineral recognition in the mining field. We used a convolutional neural network [73] to predict future publications on image processing in the mining industry.
4.2. National and Regional Heat Analysis
Figure 3 shows the research published on image processing in mining. The lines represent the cooperative relationships between different countries, and their thicknesses indicate the closeness of the cooperative relationships. A thicker line indicates closer cooperation. The knowledge map has 135 nodes and 482 links in total, with a connection density of 0.0533, where LRF = 3.0, LBY = 5, and e = 1.0. Different nodes represent different countries. The map clearly displays the distribution of research on image processing in the mining field around the world. The thickness and density of connections between different nodes are not high, indicating that moderate cooperation exists between different countries and that researchers are generally working independently from each other. Table 6 lists the 10 most effective countries and regions in terms of publications on image processing in the mining area. Among the regions, 33.63% of publications came from Europe, 26.04% came from Asia, and 40.33% came from other regions. The three countries with the most publications were China (421), the US (348), and Australia (261).
4.3. Distribution of Articles in Journals
Table 6 lists the number of publications on image processing in the mining field in various journals. Most papers have been published in Am Mineral (474; 12.74%), followed by Miner Eng (472; 12.69%), Geophysics (401; 10.78%), and Int J Miner Process (390; 10.48%). Other notable journals include Geochim Cosmochim AC, Nature, Geology, Contrib Mineral Petr, Science, and Chem Geol. These results suggest that relevant research has focused on not only the mining and machine vision fields but also on interdisciplinary fields. This is in line with the current needs of society. In Figure 4, some journals are marked with red center points, indicating high centrality scores. The intermediate centrality is a measure associated with the transformational potential of specific SCI contributions [74]. These nodes are often key to connecting networks and different stages of development in various fields. Notably, some of the key subject lines were not in the top 10 disciplines. This suggests that the importance of a given subject cannot be determined solely by its number of publications; however, it cannot be denied that the number of publications is a good indicator for a given subject.
4.4. Time–Frequency Analysis of Keywords
Keywords are prompts for article content [75] and advanced centralized descriptions that can be utilized by CiteSpace V (R.5.8.2). Figure 5 shows the results of a co-occurrence analysis of keywords [76]. The figure comprises a network with 988 nodes and 3814 connections. The network density is 0.0078 and the modularity Q value is 0.5942 (>0.3), indicating a good clustering effect. Further, the average silhouette value is 0.7992 (>0.5), indicating that the clustering results are reasonable. Among the keywords, “image processing”, “mineral”, “image analysis”, “remote sensing”, “mineral exploration”, “algorithm”, “rock”, “froth flotation”, and “ore” appeared with high frequency. Figure 6 shows the variation word list. Prominent keywords for image analysis included “mineral exploration”, “image analysis”, “process control”, “flotation froth”, “silicate mineral”, “mineralogy”, “imaging system”, “machine learning”, “CCN”, and “deep learning”. In the first few years of the study period, the most common keywords were “image analysis”, “flotation froth”, and “model”. Therefore, early research on image processing in the mining field mainly focused on flotation froth. Later keywords included “CCN”, “machine learning”, and “deep learning”. This shows that researchers focused on processing images at the micro scale. In conclusion, the application of intelligent image-processing systems to the mining field attracted research interest, due to the high labor costs of physical and chemical technologies associated with mining separation processes. The present applications of image-processing technologies to the mining field include the online monitoring of flotation froth, combining neural networks with other algorithms to build models, and constructing a comprehensive ore-grading system. Further, image-processing algorithms continue to evolve. The major challenges to realizing the commercial application of image processing in the mining field include technical bottlenecks, difficulty with market development, and low investment benefits.
4.5. Mapping and Clustering of Productive Institutions
The contributions of different institutions can be estimated by their association with at least one published author. The most-cited institution was Chinese Acad Sci (2006) (cited 67 times), followed by China Univ Geosci (2007) (cited 51 times), Cent South Univ (2013) (cited 33 times), Cent S Univ (2008) (cited 29 times), and Univ Queensland (2002) (cited 28 times). Figure 7 and Table 6 indicate that among the top 10 research institutions, seven were in China, two were in Australia, and one was in Poland. These results indicate that the application of image processing in mining has received more attention in these three countries than in other countries. Figure 7 shows that the Chinese Acad Sci has close cooperative relationships with the China University of Mining and Technology, China Uni Geosci, and other Chinese universities, as well as with the University of Queensland. There are several benefits of working with other institutions, such as increased funding, improved equipment, greater knowledge and experience, and better ability to extend technology from the laboratory to practical applications.
4.6. Authors
The knowledge map was used to identify the core authors and their coworkers in this field. Table 7 presents the top 10 authors in terms of the number of published papers. The map includes 1096 nodes with 1303 links and a connection density of 0.0022. The results showed that 63 people published more than five papers as the first author and accounted for 678 papers, or 24.80% of the total number of publications. This illustrates the great contribution of these authors to research regarding the mining field. Cluster analysis showed that the authors had a low concentration and a high dispersion. The connection between individual teams was weak, and academic exchanges were sparse.
5. Conclusions
Machine vision is based on computer vision theory. It involves related technologies, such as optical imaging, visual information processing, artificial intelligence, and electromechanical integration. Currently, many mining enterprises use manual labor to distinguish between ore and waste stone according to visual features, such as the color, grayscale, texture, and luster of the ore. Manual separation is vulnerable to subjective factors, such as individual standards, visual fatigue, thoughts, and emotions, and it is difficult to ensure the consistency of primary standards. In addition, manual separation increases labor costs. Therefore, the practical application of image processing has great implications for future developments in the mining field.
Future research directions for image processing in the mining field may involve the integration of software and hardware on processing chips to greatly improve the efficiency of practical applications, such as through the combination of neural networks with other algorithms. Future research on particle-size detection will need to consider how to use images to reflect the three-dimensional distribution of pellets, and how to use the detected particle-size information to adjust the speed of the pelletizing machine, the amount of water added, and other operating parameters to optimize the pelletizing production process. Future research on froth flotation should focus on identifying multiple features to better predict process parameters and how to use image information to adjust the feed rate, dosage, and operating parameters to optimize the flotation process. This can reduce the workload on workers and improve the mineral recovery-by-flotation processes. In addition, many studies only considered laboratory conditions; therefore, future research will need to consider industrial tests and practical applications.
6. Future Outlook
-
1.. In this study, we first analyzed specific image-processing methods and their applications in the mining field with a focus on three major aspects: ore sorting, particle-size detection, and mineral flotation. Then, bibliometric analysis and CiteSpace V (R.5.8.2) were used to generate a knowledge map regarding the applications of image processing in the mining field from 1988 to 2021. The results showed that image processing plays an important role in mining, with intensive research and worldwide studies on this topic that vary widely in scale (macro to micro) and can cover a single field or multiple disciplines. From the perspective of bibliometrics, we determined from the keywords that the words’ models, systems, classifications, and algorithms appear more frequently, indicating that the most common method for image processing is to establish a system model and use different algorithms for classification to achieve the purpose of classifying different levels of ores. Through the publications of journals and institutions, we determined that research on image processing in the mining field has been published in more leading journals and by more institutions, indicating that the subject has been explored in depth, but there are still many journals and institutions that have not had research in this direction. The research shows that many institutions and scholars have not yet been involved in this field, and in the future this field can be explored.
-
2.. In addition, the research directions in the mining field mainly include geological engineering, prospecting engineering, mining engineering, mineral processing engineering, oil and gas well engineering, oil and gas field development engineering, mine electrical engineering, mining environmental engineering, mine comprehensive utilization engineering [77], etc. The ore sorting, mineral particle size, and flotation foam treatment discussed in this article are all involved in the research directions in the mining field and can be used in other mineral-processing processes, as follows:
Image processing can be used to analyze the classical characteristic information (such as pore size, quantity, morphology, distribution, etc.) of the microstructure formed in synthetic minerals, including pore defects and characteristic mineral phases, so that synthetic minerals can be automatically obtained intuitively, qualitatively, and quantitatively [78]. It provides a reference for the scientific determination of the quality, density, and structural uniformity of synthetic minerals for porosity, pore size, and other data, and can analyze the composition effect of synthetic minerals and the factors affecting mineral strength. In the future, it can be widely used in automated laboratories or in guiding project construction, and can be used for the quality detection, monitoring, and defect analysis of synthetic minerals, which will provide scientific, rapid, and accurate analysis and detection methods for mineralogy and other disciplines. With the development of image processing and pattern recognition, their applications in the microstructure analysis of synthetic minerals will have broader prospects.
Image processing high-resolution spectral data containing information from the visible wavelength region to the near-infrared region can be obtained by hyperspectral imaging. Features of the hyperspectral data are then extracted and learned using deep learning, allowing spectral patterns unique to each mineral to be identified and analyzed. By combining hyperspectral imaging and deep learning, mineral types can be identified prior to the mineral-processing stage. This automatic mineral identification system can determine not only the types of minerals, but also the size of mineral crystals at the same time. Thus, the combined method of deep learning and hyperspectral imaging is effective in identifying mineral species and features with high accuracy, high speed, and low cost.
Some scholars have designed suitable algorithms for image-processing research in the mining field to achieve the purpose of separating different ores and gangue minerals—for example, applying image-processing and computer-vision techniques combined with multi-criteria decision-making (MCDM) and analytical hierarchy process (AHP) methods to detect different types of ores. There are also many proposed methods based on improving grayscale, texture, and color-image features to achieve the sorting of ore types.
-
3.. Our analysis presents the overall state of research on this topic and provides fellow researchers with a theoretical reference for accurately grasping the current state, direction, and hotspots in this field. To the best of our knowledge, this is the first application of bibliometric analysis to image processing in the mining field. Our data analysis process was relatively objective. However, our study still had some shortcomings. For example, not all data could be included, so some minor details were ignored. Most of the publications retrieved from the SCIE database were written in English; therefore, the analysis was incomplete because it did not include publications written in other languages. In addition, publications outside the SCIE- WOS and Scopus -registered journals were not included in the analysis. Although the collected literature described the current situation well, we lacked the opportunity to consider such papers directly.
Not applicable.
Not applicable.
No new data were created or analyzed in this study. Data sharing is not applicable to this article.
The authors declare no conflict of interest.
Footnotes
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Enlarge this image.Figure 1. Typical input-output process (information flow) of image processing in the mining field.
Enlarge this image.Figure 2. Number of publications each year and the rate of increase from 1988 to 2021.
Enlarge this image.Figure 3. Knowledge graph analysis of national and regional heat on image-processing applications in mining from 1988 to 2021.
Enlarge this image.Figure 4. Knowledge graph analysis of journal-published research on image-processing applications in mining from 1988–2021.
Enlarge this image.Figure 5. Knowledge graph analysis of keywords co-occurrence on image-processing applications in mining from 1988–2021.
Enlarge this image.Figure 6. Top 15 keywords with the strongest citation bursts. (The red line represents the time when keywords appeared, and the blue line represents the 1988–2021 timeline.)
Enlarge this image.Figure 7. Knowledge graph analysis of productive institutions’ analysis on image-processing applications in mining from 1988–2021. (The red circle indicates high frequency).
Disciplines containing ≥ 20 articles using CiteSpace.
| Disciplines | Number of Articles Using CiteSpace |
|---|---|
| Information and library sciences | 4576 |
| Education | 806 |
| Management | 684 |
| Computer science | 625 |
| Sports science | 456 |
| Multidisciplinary social sciences | 424 |
| Geography | 402 |
| Building science and engineering | 382 |
| Economics | 356 |
| Communication | 342 |
| Environmental sciences | 332 |
| Medicine | 325 |
Common image processing methods and their characteristics.
| Method | Definition | Algorithm | Features |
|---|---|---|---|
| Image transformation [ |
An image originally defined in the image space is converted into another space in a certain form. The properties of the space are used to facilitate certain processes and the image is then converted back into the image space to achieve the desired effect. | Fourier transform [ |
Efficient processing and analysis of images |
| Image enhancement and restoration [ |
Digital image processing technology can be used to emphasize one region of interest in an image and suppress other regions. | Image enhancement: histogram equalization, image smoothing, image sharpening, false-color method, etc. |
Regions of interest are highlighted for clarity for the user and can be represented by different colors. |
| Image segmentation | An image is divided into specific and unique areas to identify the objects of interest. | Gray threshold segmentation, region segmentation, edge segmentation, histogram method, segmentation based on a specific theory, etc. | A digital image is divided into nonintersecting regions, and it is also a marking process. Most segmentation algorithms are aimed at solving specific problems. |
| Image description | After segmentation, data, symbols and formal language are used to represent the cells with different characteristics. | Curve fitting, chain code, Fourier description based on arc length polar radius, moment description, etc. | Descriptions can involve a region, the relationship between regions, and the structure of the description. The description uses lines, curves, regions, and other geometric features. |
| Image classification (recognition) | Image processing method to distinguish different categories of objects according to their varying characteristics reflected in the image information. | Index technology based on color features and image classification technology based on texture, shape, spatial relations, etc. | Each pixel or region in an image is quantitatively analyzed by a computer for classification, rather than being interpreted by a human. |
Important developments in research on image-processing applications for ore sorting.
| Time | Research Results | Main Characteristics | Source |
|---|---|---|---|
| Early 20th century | Optical sorter | Various ore images were used to design an optimized Faster R-CNN algorithm [ |
Sortex UK |
| 1970s | M16 photoelectric sorting machine | It can quickly sort magnesite ore with an accuracy of 92–93%. Rescreening allows it to reach an accuracy of 98–99% [ |
Ore sorters |
| Early 21st century | Excited light pickers (i.e., X-ray pickers) | Used to sort metal, nonmetal, precious metal ores, and other rare ores [ |
Redwave, Austria and Mogensen, Germany |
| 1980 | RM161-50 Radioactive Sorter | It relies on the inherent radioactivity of a mineral or an external radiation source for ore sorting [ |
Sortex UK |
| 2000 | Automatic sorting system for ore based on machine vision | Uses traditional machine vision with machine learning. An optimized support vector machine is used for identification. The centroid method is used to locate the specific orientation of coal mine gangue relative to the camera. | [ |
| 2018 | Deep learning–based method for automatic feature extraction | The CNN U-net is used with a successful detection rate of 90%; features are automatically extracted to complete the detection. | [ |
| 2019 | Wolframite sorting system designed using machine vision and machine learning | Algorithm is designed based on the combination of a genetic algorithm with a neural network. | [ |
| 2021 | Optimized Faster R-CNN algorithm is designed using various ore images as research samples | The algorithm finds local and global optimal solutions. The algorithm parameters were optimized. The algorithm requires a long training time but has a detection accuracy of ~98%. | [ |
Important developments in research on image-processing applications for particle-size detection.
| Time | Development | Main Characteristics | Source |
|---|---|---|---|
| 1976 | Optical instruments are used to measure the chord length of an ore particle. | The description of the ore particle size has major defects and errors, and this approach has limited applicability to mining. However, this is a breakthrough in online particle-size detection. | [ |
| 1988 | Analysis of the particle-size composition of lead–zinc ore before and after crushing | Earliest formal application of image processing technology to the mining process and serves as a model for subsequent applications of this technology to mineral processing. | [ |
| 1992 | Two algorithms are designed for classifying rock fragments produced by blasting. | They solve the problem of multilink mineral-image segmentation. | [ |
| 1994 | Online coarse particle-size analysis method is designed using image processing technology. | This method accurately describes the particle-size distribution on a static belt and when the coal blocks are not stacked. However, the prediction accuracy is not ideal for a moving belt. | [ |
| 1996 | Particle-size image analysis system | Although this system is proposed to solve the problem of particle-size distribution with image processing technology, it is only applicable to coarse particles, and the accuracy needs improvement. | [ |
| 1997 | Image-analysis method for effectively describing the shapes of concave and convex particles | This method has general applicability and provides a novel strategy for characterizing the particle-size distribution of coal particles. | [ |
| 2002 | Rock-image segmentation-and-recognition technology based on three-dimensional information | This technology can effectively identify surface rock fragments and improve the segmentation accuracy of stacked rock fragments. However, it does not address the classification of covered rock particles. | [ |
| 2003 | Predicting particle thickness using area | This novel idea used two-dimensional information to supplement three-dimensional information. | [ |
| 2005 | Using the area of a region to predict the grain size and volume of rock fragments | This method uses computer vision to predict the grain-size distribution of rock. | [ |
| 2006 | Measurement of mineral particle–size parameters in nonoverlapping tiles using image analysis | This method uses multiple parameters, rather than a single parameter, to improve the prediction accuracy. | [ |
| 2009 | Repeated flash imaging to obtain high-quality images | Repeated flash imaging improves the segmentation effectiveness to better fit the true shape of a particle, and measure it more accurately. | [ |
| 2010 | Predictive estimation of coal particle–size distribution using a kernel method in machine learning | The kernel method is used to predict the particle-size distribution of coal samples with an accuracy of 84.2%. | [ |
| 2015 | Microscopic three-dimensional analysis of coal particles using image processing and random analysis | This method applies image-processing technology to microscale analysis rather than to macroscale analysis. | [ |
Important developments in research on image-processing applications for mineral flotation.
| Main Visual Features | Algorithms | Main Characteristics | References |
|---|---|---|---|
| Foam color | RGB space conversion, HSV space, and multiresolution multivariate variable image analysis | The surface color of the flotation froth is directly related to the mineral composition and grade. | [ |
| Bubble morphology | Watershed segmentation, valley edge segmentation, etc. | The bubble size and shape characterize the foam structure and are closely related to key working parameters, such as the concentration of the foaming agent and the aeration rate. | [ |
| Foam texture | GLCM, color co-occurrence matrix, WTA, Gabor wavelet, etc. | Changes in the flotation conditions result in unique textural features for the froth surface and different shades of color, depending on the mineral. | [ |
| Foam dynamic characteristics | Clustering algorithm, RK algorithm, feature point registration based on sifting, etc. | Dynamic features are mainly used to describe the characteristic behavior of froth moving in the flotation machine, including the froth flow rate, froth stability, etc. | [ |
Current problems of image processing in ore sorting, ore particle size, flotation foam, and machine learning research.
| Research Direction of Image Processing in Mining Field | Problems in the Current Research |
|---|---|
| The mineral sorting | It is difficult to maintain the system. The stability of the system is not sufficient, the ore cannot be shot without dead angle, the hardware assembly is difficult, the connection between each module is poor, and so on. |
| The mineral particle-size detection | The particle size of multi-layer stacked ore is not sufficiently accurate and the experiment is only carried out in the laboratory, without further study in industrial and practical applications. Multidimensional analysis has not been carried out. |
| Mineral flotation foam | The illumination intensity is too high or too low, the illumination is not uniform, the illumination is not stable, the illumination area is not sufficient, the illumination angle is not sufficient, the algorithm accuracy is not sufficient, and so on. |
| Machine learning | The quality of data is uneven, the amount of data is not sufficient, the effect of training model is not sufficiently accurate, and there is no further research regarding industrial and practical applications. |
Top 10 countries, journals, and institutes according to the number of publications on image-processing applications in the mining field.
| Rank | Country | Number |
Journals | Number |
Institutes | Number |
Author | Number |
|---|---|---|---|---|---|---|---|---|
| 1 | China | 421 |
Am Mineral | 474 |
Chinese Acad Sci | 67 |
Weihua Gui | 29 |
| 2 | USA | 348 |
Miner Engineering | 472 |
China Univ Geosci | 51 |
C. Aldrich | 27 |
| 3 | Australia | 261 |
Geophysics | 401 |
Cent South Univ | 30 |
J. Miller | 27 |
| 4 | Canada | 198 |
Int J Miner Process | 390 |
Cent S Univ | 29 |
A. Jaedsaravani | 24 |
| 5 | Germany | 184 |
Geochim Cosmochim AC | 383 |
Univ Queensland | 28 |
M. Massinaei | 23 |
| 6 | France | 138 |
Nature | 337 |
China Univ Min & Techol | 27 |
E. Donskoi | 22 |
| 7 | England | 124 |
Geology | 326 |
Univ Chinese Acad Sci | 23 |
Nigel J Cook | 21 |
| 8 | Italy | 115 |
Contrib mineral Petr | 324 |
Curtin Univ | 21 |
Cristian L Ciobanu | 21 |
| 9 | Iran | 112 |
Science | 308 |
Polish Acad Sci | 20 |
G. Bonifazi | 21 |
| 10 | India | 100 |
Chem Geol | 305 |
Univ Sci & Technol Beijing | 19 |
Liu J. | 19 |
References
1. Caggiano, A.; Zhang, J.; Alfieri, V.; Caiazzo, F.; Gao, R.; Teti, R. Machine learning-based image processing for on-line defect recognition in additive manufacturing. CIRP Ann.; 2019; 68, pp. 451-454.
2. Okada, N.; Maekawa, Y.; Owada, N.; Haga, K.; Shibayama, A.; Kawamura, Y. Automated identification of mineral types and grain size using hyperspectral imaging and deep learning for mineral processing. Minerals; 2020; 10, 809. [DOI: https://dx.doi.org/10.3390/min10090809]
3. Ren, Z.; Sun, L.; Zhai, Q.; Liu, X. Mineral mapping with hyperspectral image based on an improved k-means clustering algorithm. Proceedings of the 2019 IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2019); Yokohama, Japan, 28 July–2 August 2019.
4. Ebrahimi, M.; Abdolshah, M.; Abdolshah, S. Developing a computer vision method based on AHP and feature ranking for ores type detection. Appl. Soft Comput.; 2016; 49, pp. 179-188.
5. Fan, L.F.; Gao, J.W.; Wu, Z.J.; Yang, S.Q.; Ma, G.W. An investigation of thermal effects on micro-properties of granite by X-ray CT technique. Appl. Therm. Eng.; 2018; 140, pp. 505-519. [DOI: https://dx.doi.org/10.1016/j.applthermaleng.2018.05.074]
6. Fujiwara, E.; dos Santos, M.F.M.; Schenkel, E.A.; Ono, E.; Suzuki, C.K. Optical classification of quartz lascas by artificial neural networks. Miner. Process. Extr. Met. Rev.; 2015; 36, pp. 281-287. [DOI: https://dx.doi.org/10.1080/08827508.2014.978315]
7. Wang, J.; Li, L.; Yang, S. Image-based rock mixing ratio estimation by using illuminance analysis in underground mining. Int. J. Coal Prep. Util.; 2021; 42, pp. 3745-3762. [DOI: https://dx.doi.org/10.1080/19392699.2021.1995375]
8. Nagrath, P.; Jain, R.; Madan, A.; Arora, R.; Kataria, P.; Hemanth, J. SSDMNV2: A real time DNN-based face mask detection system using single shot multibox detector and MobileNetV2. Sustain. Cities Soc.; 2021; 66, 102692. [DOI: https://dx.doi.org/10.1016/j.scs.2020.102692]
9. St-Charles, P.L.; Bilodeau, G.A.; Bergevin, R. Subsense: A universal change detection method with local adaptive sensitivity. IEEE Trans. Image Process.; 2015; 24, pp. 359-373. [DOI: https://dx.doi.org/10.1109/TIP.2014.2378053]
10. Kursun, I. Particle size and shape characteristics of kemerburgaz quartz sands obtained by sieving, laser diffraction, and digital image processing methods. Miner. Process. Extr. Metall. Rev.; 2009; 30, pp. 346-360. [DOI: https://dx.doi.org/10.1080/08827500903149659]
11. Howison, J.; Deelman, E.; McLennan, M.J.; da Silva, R.F.; Herbsleb, J.D. Understanding the scientific software ecosystem and its impact: Current and future measures. Res. Eval.; 2015; 24, pp. 454-470. [DOI: https://dx.doi.org/10.1093/reseval/rvv014]
12. Hannay, J.E.; MacLeod, C.; Singer, J.; Langtangen, H.P.; Pfahl, D.; Wilson, G. How do scientists develop and use scientific software?. Proceedings of the 2009 ICSE Workshop on Software Engineering for Computational Science and Engineering; Vancouver, BC, Canada, 23 May 2009.
13. Howison, J.; Bullard, J. Software in the scientific literature: Problems with seeing, finding, and using software mentioned in the biology literature. J. Assoc. Inf. Sci. Technol.; 2016; 67, pp. 2137-2155. [DOI: https://dx.doi.org/10.1002/asi.23538]
14. Thelwall, M.; Kousha, K. Academic software downloads from Google Code: Useful usage indicators?. Inf. Res.; 2016; 21, n1.
15. Hafer, L.; Kirkpatrick, A.E. Assessing open source software as a scholarly contribution. Commun. ACM; 2009; 52, pp. 126-129. [DOI: https://dx.doi.org/10.1145/1610252.1610285]
16. Piwowar, H. Value all research products. Nature; 2013; 493, 159. [DOI: https://dx.doi.org/10.1038/493159a] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23302843]
17. Chen, Y.; Chen, C.; Liu, Z.; Hu, Z.; Wang, X. Methodological function of CiteSpace knowledge graph. Stud. Sci. Sci.; 2015; 33, pp. 242-253.
18. Chen, Y.; Liu, Z.; Chen, J.; Hou, J. Development history of scientific knowledge graph. Stud. Sci. Sci.; 2008; 26, 12.
19. Onan, A.; Korukoğlu, S.; Bulut, H. A hybrid ensemble pruning approach based on consensus clustering and multi-objective evolutionary algorithm for sentiment classification. Inf. Process. Manag.; 2017; 53, pp. 814-833. [DOI: https://dx.doi.org/10.1016/j.ipm.2017.02.008]
20. Onan, A. Two-stage topic extraction model for bibliometric data analysis based on word embeddings and clustering. IEEE Access; 2019; 7, pp. 145614-145633. [DOI: https://dx.doi.org/10.1109/ACCESS.2019.2945911]
21. Ouyang, W.L.; Cham, W.K. Fast algorithm for walsh hadamard transform on sliding windows. IEEE Trans. Pattern Anal. Mach. Intell.; 2010; 32, pp. 165-175. [DOI: https://dx.doi.org/10.1109/TPAMI.2009.104]
22. Wang, G.; Garcia, D.; Liu, Y.; De Jeu, R.; Dolman, A.J. A three-dimensional gap filling method for large geophysical datasets: Application to global satellite soil moisture observations. Environ. Model. Softw.; 2012; 30, pp. 139-142. [DOI: https://dx.doi.org/10.1016/j.envsoft.2011.10.015]
23. Duarte, M.F.; Parente, M. Non-homogeneous hidden markov chain models for wavelet-based hyperspectral image processing. Proceedings of the 2013 51st Annual Allerton Conference on Communication, Control and Computing (AL-LERTON); Monticello, IL, USA, 2–4 October 2013.
24. Bai, X.Z.; Zhang, Y. Enhancement of microscopy mineral images through constructing alternating operators using opening and closing based toggle operator. J. Opt.; 2014; 16, 125407. [DOI: https://dx.doi.org/10.1088/2040-8978/16/12/125407]
25. Wang, S. New progress in ore picking. Uranium Min. Metall.; 2010; 6, pp. 162-178. [DOI: https://dx.doi.org/10.1504/IJLSM.2010.030958]
26. Cui, G. Current status and future development of sorting technology at home and abroad in China. Non-Met. Miner.; 1983; 1384, pp. 351-358. [DOI: https://dx.doi.org/10.1007/978-3-030-74811-1_51]
27. Liu, G. Experiment and study of X-ray sorting machine in molybdenum ore preselection. Mod. Min.; 2009; 000, pp. 75-77.
28. Luan, L.; Guo, L.; Zhang, M. Edge detection algorithm for ore block distribution image processing. J. Northeast. Univ. Nat. Sci. Ed.; 2004; 4, pp. 35-38.
29. King, R.P. Determination of the distribution of size of irregularly shaped particles from measurements on sections or projected areas. Powder Technol.; 1982; 32, pp. 87-100. [DOI: https://dx.doi.org/10.1016/0032-5910(82)85009-2]
30. Voulodimos, A.; Doulamis, N.; Doulamis, A.; Protopapadakis, E. Deep learning for computer vision: A brief review. Comput. Intell. Neurosci.; 2018; 2018, 7068349. [DOI: https://dx.doi.org/10.1155/2018/7068349]
31. Mccoy, J.T.; Auret, L. Machine learning applications in minerals processing: A review. Miner. Eng.; 2019; 132, pp. 95-109. [DOI: https://dx.doi.org/10.1016/j.mineng.2018.12.004]
32. Liu, X.; Aldrich, C. IEEE. Monitoring of froth flotation with transfer learning and principal component models. Proceedings of the 2021 Australian & New Zealand Control Conference (ANZCC); Gold Coast, QL, Australia, 25–26 November 2022.
33. Pei, Z.; Chang, L.; Xue, P.; Harrison, R.J. MagNet: Automated magnetic mineral grain morphometry using convolutional neural network. Geophys. Res. Lett.; 2022; 49, e2022GL099118. [DOI: https://dx.doi.org/10.1029/2022GL099118]
34. Li, X.; Chen, N.; Dong, M.; Xie, Y.; Huang, Y. Research progress of chromite separation technology. Guangzhou Chem. Ind.; 2014; 42, pp. 32-34.
35. Liu, F.; Qian, J.; Wang, X.; Song, J. Automatic sorting of gangue from coal mine based on image processing and recognition technology. J. China Coal Soc.; 2000; pp. 534-537. [DOI: https://dx.doi.org/10.3321/j.issn:0253-9993.2000.05.020]
36. Xu, S.; Zhou, Y. Experimental study on intelligent identification of ore minerals under microscopy based on deep learning. J. Petrol.; 2018; 34, pp. 3244-3252.
37. Xiao, J. Research and Design of Intelligent Sorting System of Wolfragsten Ore Based on Machine Vision. Ph.D. Thesis; Hunan University: Changsha, China, 2019.
38. Deng, T.; Yu, Y. Research on ore identification and classification based on PSO-faster R-CNN improved algorithm. Min. Res. Dev.; 2021; 41, pp. 178-182.
39. Yen, Y.K.; Lin, C.L.; Miller, J.D. Particle overlap and segregation problems in on-line coarse particle size measurement. Powder Technol.; 1998; 98, pp. 1-12. [DOI: https://dx.doi.org/10.1016/S0032-5910(97)03405-0]
40. Petruk, W. Automatic image analysis for mineral beneficiation. JOM J. Miner. Met. Mater. Soc.; 1988; 40, pp. 29-31. [DOI: https://dx.doi.org/10.1007/BF03259018]
41. Li, H. Review of particle size detection technology. J. Liaoning Univ. Sci. Technol.; 2007; pp. 6-7. [DOI: https://dx.doi.org/10.3969/j.issn.1008-3723.2007.03.003]
42. Kupka, N.; Rudolph, M. Froth flotation of scheelite—A review. Int. J. Min. Sci. Technol.; 2018; 28, pp. 373-384. [DOI: https://dx.doi.org/10.1016/j.ijmst.2017.12.001]
43. Wu, X.; Kemeny, J.M. A segmentation method for multi-connected particle delineation. Proceedings of the IEEE Workshop on Applications of Computer Vision; Palm Springs, CA, USA, 30 November–2 December 1992.
44. Nakajima, Y.; Gotoh, K.; Tanaka, T. On-line particle size analyzer. Ind. Eng. Chem. Fundam.; 2002; 6, pp. 587-592. [DOI: https://dx.doi.org/10.1021/i160024a015]
45. Downs, D.C.; Kettunen, B.E. On-line fragmentation measurement utilizing the CIAS(R) system. Measurement of Blast Frag-mentation; Routledge: Abingdon, UK, 2018.
46. Hundal, H.S.; Rohani, S.; Wood, H.C.; Pons, M.N. Particle shape characterization using image analysis and neural networks. Powder Technol.; 1996; 91, pp. 217-227. [DOI: https://dx.doi.org/10.1016/S0032-5910(96)03258-5]
47. Thurley, M.J. Three Dimensional Data Analysis for the Separation and Sizing of Rock Piles in Mining. Ph.D. Thesis; Department of Electrical and Computer Systems, Monash University: Melbourne, Australia, 2002.
48. Vallebuona, G.; Arburo, K.; Casali, A. A procedure to estimate weight particle distributions from area measurements. Miner. Eng.; 2003; 16, pp. 323-329. [DOI: https://dx.doi.org/10.1016/S0892-6875(03)00014-1]
49. Salinas, R.A.; Raff, U.; Farfan, C. Automated estimation of rock fragment distributions using computer vision and its applica-tion in mining. IEEE Proc. Vis. Image Signal Process.; 2015; 152, pp. 1-8. [DOI: https://dx.doi.org/10.1049/ip-vis:20050810]
50. Al-Thyabat, S.; Miles, N.J. An improved estimation of size distribution from particle profile measurements. Powder Technol.; 2006; 166, pp. 152-160. [DOI: https://dx.doi.org/10.1016/j.powtec.2006.05.008]
51. Koh, T.K.; Miles, N.J.; Morgan, S.P.; Hayes-Gill, B.R. Improving particle size measurement using multi-flash imaging. Miner. Eng.; 2009; 22, pp. 537-543. [DOI: https://dx.doi.org/10.1016/j.mineng.2008.12.005]
52. Aldrich, C.; Jemwa, G.T.; Van Dyk, J.C.; Keyser, M.J.; Van Heerden, J.H.P. Online analysis of coal on a conveyor belt by use of machine vision and kernel methods. Coal Prep.; 2010; 30, pp. 331-348. [DOI: https://dx.doi.org/10.1080/19392699.2010.517486]
53. Pant, L.M.; Huang, H.; Secanell, M.; Larter, S.; Mitra, S.K. Multi scale characterization of coal structure for mass transport. Fuel; 2015; 159, pp. 315-323. [DOI: https://dx.doi.org/10.1016/j.fuel.2015.06.089]
54. Moolman, D.W.; Aldrich, C.; Van Deventer, J.S.; Stange, W.W. Digital image processing as a tool for on-line monitoring of froth in flotation plants. Miner. Eng.; 1994; 7, pp. 1149-1164. [DOI: https://dx.doi.org/10.1016/0892-6875(94)00058-1]
55. Yang, C.; Xu, C.; Gui, W.; Zhou, K. Application of highlight removal and multivariate image analysis to color measurement of flotation bubble images. Int. J. Imaging Syst. Technol.; 2010; 19, pp. 316-322. [DOI: https://dx.doi.org/10.1002/ima.20208]
56. Aldrich, C.; Marais, C.; Shean, B.; Cilliers, J. Online monitoring and control of froth flotation systems with machine vision: A review. Int. J. Miner. Process.; 2010; 96, pp. 1-13. [DOI: https://dx.doi.org/10.1016/j.minpro.2010.04.005]
57. Li, J.Q.; Yang, C.H.; Zhu, H.Q.; Wei, L.J. Improved image enhancement method for flotation froth image based on parameter extraction. J. Cent. South Univ.; 2013; 20, pp. 1602-1609. [DOI: https://dx.doi.org/10.1007/s11771-013-1652-2]
58. Jahedsaravani, A.; Massinaei, M.; Marhaban, M.H. An image segmentation algorithm for measurement of flotation froth bubble size distributions. Measurement; 2017; 111, pp. 29-37. [DOI: https://dx.doi.org/10.1016/j.measurement.2017.07.023]
59. Xu, D.; Chen, X.; Xie, Y.; Yang, C.; Gui, W. Complex networks-based texture extraction and classification method for mineral flotation froth images. Miner. Eng.; 2015; 83, pp. 105-116. [DOI: https://dx.doi.org/10.1016/j.mineng.2015.08.017]
60. Fu, Y.H.; Aldrich, C. Flotation froth image analysis by use of a dynamic feature extraction algorithm: IFAC PapersOnLine. Proceedings of the 17th IFAC Symposium on Control, Optimization and Automation in Mining, Mineral and Metal Processing (MMM); Vienna, Austria, 31 August–2 September 2016; Volume 49, pp. 84-89.
61. Roy, S.; Mandal, S.K.; Das, A. Segregation and process features in a teeter bed separator as revealed by high-speed videography and image processing. Miner. Process. Extr. Metall. Rev.; 2013; 35, pp. 15-22. [DOI: https://dx.doi.org/10.1080/08827508.2012.700659]
62. Wang, Y.; Neethling, S.J. The relationship between the surface and internal structure of dry foam. Colloids Surf. A Physicochem. Eng. Asp.; 2009; 339, pp. 73-81. [DOI: https://dx.doi.org/10.1016/j.colsurfa.2009.01.021]
63. Wang, W.; Bergholm, F.; Yang, B. Froth delineation based on image classification. Miner. Eng.; 2003; 16, pp. 1183-1192. [DOI: https://dx.doi.org/10.1016/j.mineng.2003.07.014]
64. Moolman, D.W.; Aldrich, C.; Van Deventer, J.S.J.; Stange, W.W. The classification of froth structures in a copper flotation plant by means of a neural net. Int. J. Miner. Process.; 1995; 43, pp. 193-208. [DOI: https://dx.doi.org/10.1016/0301-7516(95)00003-V]
65. Bartolacci, G.; Pelletier, P.; Tessier, J.; Duchesne, C.; Bossé, P.-A.; Fournier, J. Application of numerical image analysis to process diagnosis and physical parameter measurement in mineral processes—Part I: Flotation control based on froth textural char-acteristics. Miner. Eng.; 2006; 19, pp. 734-747. [DOI: https://dx.doi.org/10.1016/j.mineng.2005.09.041]
66. Liu, J.; Gui, W.; Mou, X.; Tang, Z.; Li, Z. Texture feature extraction of flotation foam image based on gabor wavelet. Chin. J. Sci. Instrum.; 2010; 31, pp. 1769-1775.
67. Kaartinen, J.; Hätönen, J.; Hyötyniemi, H.; Miettunen, J. Machine-vision-based control of zinc flotation—A case study. Control. Eng. Pract.; 2006; 14, pp. 1455-1466. [DOI: https://dx.doi.org/10.1016/j.conengprac.2005.12.004]
68. Holtham, P.N.; Nguyen, K.K. On-line analysis of froth surface in coal and mineral flotation using JKFrothCam. Int. J. Miner. Process.; 2002; 64, pp. 163-180. [DOI: https://dx.doi.org/10.1016/S0301-7516(01)00070-9]
69. Park, H.; Bai, C.; Wang, L. A convolutional neural network for classification of froth mobility in an industrial flotation cell. Miner. Process. Extr. Met. Rev.; 2022; pp. 1-9. [DOI: https://dx.doi.org/10.1080/08827508.2022.2042281]
70. Pons, M.N.; Vivier, H.; Belaroui, K.; Bernard-Michel, B.; Cordier, F.; Oulhana, D.; Dodds, J.A. Particle morphology: From visualisation to measurement. Powder Technol.; 1999; 103, pp. 44-57. [DOI: https://dx.doi.org/10.1016/S0032-5910(99)00023-6]
71. Xiong, Y.; Zuo, R.; Carranza, E. Mapping mineral prospectivity through big data analytics and a deep learning algorithm. Ore Geol. Rev.; 2018; 102, pp. 811-817. [DOI: https://dx.doi.org/10.1016/j.oregeorev.2018.10.006]
72. Chen, C. CiteSpace II: Detecting and visualizing emerging trends and transient patterns in scientific literature. J. Am. Soc. Inf. Sci. Technol.; 2006; 57, pp. 359-377. [DOI: https://dx.doi.org/10.1002/asi.20317]
73. Onan, A.; Korukoğlu, S.; Bulut, H. Ensemble of keyword extraction methods and classifiers in text classification. Expert Syst. Appl.; 2016; 57, pp. 232-247. [DOI: https://dx.doi.org/10.1016/j.eswa.2016.03.045]
74. Maxwell, A.E.; Strager, M.P.; Warner, T.A.; Zégre, N.P.; Yuill, C.B. Comparison of NAIP orthophotography and RapidEye satellite imagery for mapping of mining and mine reclamation. GISci. Remote Sens.; 2014; 51, pp. 301-320. [DOI: https://dx.doi.org/10.1080/15481603.2014.912874]
75. Ramos Oliveira, G.J.; Horta, D.G.; da Silva, F.L.; von Krüger, F.L.; Araújo FG, D.S.; Mazzinghy, D.B. Determination of mineral liberation of a bauxite ore based on 3d compositional and textural characteristics using X-ray microtomography. Miner. Process. Extr. Metall. Rev.; 2021; 43, pp. 1-11. [DOI: https://dx.doi.org/10.1080/08827508.2021.1986707]
76. MacGregor, J.F.; Bruwer, M.J.; Miletic, I.; Cardin, M.; Liu, Z. Latent variable models and big data in the process industries. IFAC PapersOnLine; 2015; 48, pp. 520-524. [DOI: https://dx.doi.org/10.1016/j.ifacol.2015.09.020]
77. Ge, Z.; Song, Z.; Ding, S.X.; Huang, B. Data mining and analytics in the process industry: The role of machine learning. IEEE Access; 2017; 5, pp. 20590-20616. [DOI: https://dx.doi.org/10.1109/ACCESS.2017.2756872]
78. Xu, W.; Song, Y.; Dai, K.; Sun, S.; Liu, G.; Yao, J. Novel ternary nanohybrids of tetraethylenepentamine and graphene oxide decorated with MnFe_2O_4 magnetic nanoparticles for the adsorption of Pb(II). J. Hazard. Mater.; 2018; 358, pp. 337-345. [DOI: https://dx.doi.org/10.1016/j.jhazmat.2018.06.071]
79. Hoang, N.T.; Koike, K. Comparison of hyperspectral transformation accuracies of multispectral Landsat TM, ETM plus, OLI and EO-1 ALI images for detecting minerals in a geothermal prospect area. ISPRS J. Photogramm. Remote Sens.; 2018; 137, pp. 15-28. [DOI: https://dx.doi.org/10.1016/j.isprsjprs.2018.01.007]
80. Bunin, I.Z.; Chanturia, V.A.; Anashkina, N.E.; Ryazantseva, M.V. Experimental validation of mechanism for pulsed energy effect on structure, chemical properties and microhardness of rock-forming minerals of kimberlites. J. Min. Sci.; 2015; 51, pp. 799-810. [DOI: https://dx.doi.org/10.1134/S1062739115040177]
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.