Recognition and monitoring of gas leakage using infrared imaging technique with machine learning

Abstract

Gas leakage in the domestic sector leads to numerous dangerous hazards. The earlier prediction is one of the safety measures to prevent various consequences. The proposed system helps in the earlier detection of gas leakage using artificial intelligence techniques. This involves machine learning with infrared imaging techniques. Machine learning is the process of teaching machines to do tasks automatically by analysing and testing data. The obtained data are processed using image processing techniques. The image processing technique is used to extract information from the images involving various stages such as image enhancement and image analysis. The initial data are obtained in the form of images using infrared imaging techniques. It is the technique that utilizes the infrared portion of the electromagnetic spectrum to obtain the desired images. The obtained images are processed to obtain clear images in the dataset. The data is then tested and taught using machine learning evolving optimization techniques on the data. This helps in the accurate detection of gas leakage. To compare, the individual models' test accuracy ranged from 99.8% (based on Gas Sensor data using Random Forest) with the training accuracy of 99.8%. Experimental results demonstrate its ability to automatically detect and display gas leaks in high quality by establishing a background model, segmenting the gas-leak zone with motion characteristics, and rendering the gas-leak region in colour using grayscale mapping.