Scanning orders of bitplane image coding engines are commonly envisaged from theoretical or experimental insights and assessed in practice in terms of coding performance. This paper evaluates classic scanning strategies of modern bitplane image codecs using several theoretical-practical mechanisms conceived from rate-distortion theory. The use of these mechanisms allows distinguishing those features of the bitplane coder that are essential from those that are not. This discernment can aid the design of new bitplane coding engines with some special purposes and/or requirements. To emphasize this point, a low-complexity scanning strategy is proposed. Experimental evidence illustrates, assesses, and validates the proposed mechanisms and scanning orders.

The release of the CUDA Kepler architecture in March 2012 has provided Nvidia GPUs with a larger register memory space and instructions for the communication of registers among threads. This facilitates a new programming strategy that utilizes registers for data sharing and reusing in detriment of the shared memory. Such a programming strategy can significantly improve the performance of applications that reuse data heavily. This paper presents a register-based implementation of the Discrete Wavelet Transform (DWT), the prevailing data decorrelation technique in the field of image coding. Experimental results indicate that the proposed method is, at least, four times faster than the best GPU implementation of the DWT found in the literature. Furthermore, theoretical analysis coincide with experimental tests in proving that the execution times achieved by the proposed implementation are close to the GPU's performance limits.

Quality scalability is a fundamental feature of JPEG2000, achieved through the use of quality layers that are optimally formed in the encoder by rate-distortion optimization techniques. Two points, related with the practical use of quality layers, may need to be addressed when dealing with JPEG2000 code-streams: 1) the lack of quality scalability of code-streams containing a single or few quality layers and 2) the rate-distortion optimality of windows of interest transmission. Addressing these two points, this paper proposes a mechanism that, without using quality layers, provides competitive quality scalability to code-streams. Its main key-feature is a novel characterization of the code-blocks rate-distortion contribution that does not use distortion measures based on the original image, or related with the encoding process. Evaluations against the common use of quality layers, and against a theoretical optimal coding performance when decoding windows of interest or when decoding the complete image area, suggest that the proposed method achieves close to optimal results.

This study measures the effect of lossy image compression on the digital classification of crops and forest areas. A hybrid classification method using satellite images and other variables has been used. The results contribute interesting new data on the influence of compression on the quality of the produced cartography, both from a "by pixel" perspective and regarding the homogeneity of the obtained polygons. The classified area in classifications only carried out with radiometric variables or with NDVI and humidity (for crops) increases as image compression increases, although the increase is smaller for JPEG2000 formats and for crops. On the other hand, the classified area decreases in classifications which also take into account topoclimatic variables (for forests). Overall image accuracy diminishes at high compression ratios (CR), although the point of inflection occurs at different CR depending on the compression format. As a rule, the JPEG2000 format gives better results quantitatively for forests (accuracy and classified area) and visually (images with less "salt and pepper" effect) for both land covers.

This paper introduces a probability model for symbols emitted by bitplane image coding engines, which is conceived from a precise characterization of the signal produced by a wavelet transform. Main insights behind the proposed model are the estimation of the magnitude of wavelet coefficients as the arithmetic mean of its neighbors' magnitude (the so-called local average), and the assumption that emitted bits are under-complete representations of the underlying signal. The local average-based probability model is introduced in the framework of JPEG2000. While the resulting system is not JPEG2000 compatible, it preserves all features of the standard. Practical benefits of our model are enhanced coding efficiency, more opportunities for parallelism, and improved spatial scalability.

This paper proposes an HEVC-based method for lossless compression of Whole Slide pathology Images (WSIs). Based on the observation that WSIs usually feature a high number of edges and multidirectional patterns due to the great variety of cellular structures and tissues depicted, we combine the advantages of sample-by-sample differential pulse code modulation (SbS-DPCM) and edge prediction into the intra coding process. The objective is to enhance the prediction performance where strong edge information is encountered. This paper also proposes an implementation of the decoding process that maintains the block-wise coding structure of HEVC when SbS-DPCM and edge prediction are employed. Experimental results on various WSIs show that the proposed method attains average bit-rate savings of 7.67%.