Content area
Full Text
An algorithm dedicated to binary image lossless compression is proposed. It uses the move-to-front transform associated to the two-role and the arithmetic encoders. Obtained results, when applying the technique on the benchmark CCITT test images, demonstrate clearly its effectiveness. Therefore, an average bit rate of 0.06 per pixel is achieved.
Introduction: Developed algorithms for data compression can be categorised into two main groups. The first one encloses the lossless techniques which guarantee the perfect preservation of the initial quality of the compressed data. Conversely, the second category includes lossy methods which can attain high compression ratio, compared to lossless strategies. However, they should conserve an adequate level of reconstruction quality appropriate to the application used. As an example, constraints of lossy compression of medical images are more severe than those in audio or natural image compression [1, 2].
One of the open issues that can be considered, is lossless bi-level (binary or two-tone) image compression. Many researches have been conducted to develop algorithms allowing high compression ratio (CR) [3 - 8].In[3], the authors presented a non-symmetry and antipacking (NAM) method for compression permitting both lossless and lossy compression. They reported that the proposed scheme outperforms the linear quadtree and the RLE encoder. In the same optic, Dai and Zakhor [4] suggested a lossless compression technique taking advantage of the two very dissimilar compression techniques: context-based modelling and Lempel -Ziv (LZ) style copying. It has been used for integrated circuit layout compression. The authors of [5- 7] used different strategies for efficiently partitioning the whole image into homogeneous rectangular regions that can be easily stored. Another alternative was shown in [8]. Its idea is based on the ordered binary decision diagram (OBDD). Later, an improved version was considered in [9]. Moreover, in [10], the logic functions and spectra were combined successfully to achieve results near to those of [9].
In this context, in this Letter we propose a method for encoding bilevel images. It includes the move-to-front (MTF) transform that reduces the image entropy. Associated to the two-role encoder (TRE), efficient encoding of the resulting data is ensured [11]. The two key steps of the proposed algorithm are: the MTF transform and the TRE encoder. In the following, we present an overview of the...