Imaging devices capable of capturing multiple modalities, such as visible light, depth, and infrared, are increasingly prevalent, necessitating efficient compression methods. Previous research independently compressed depth, infrared, and visible information. In contrast, this study aims to improve the coding efficiency of grayscale images by utilizing infrared images as auxiliary data. This approach employs 20 encoding methods, including 19 prediction methods integrating infrared images. We assume that both the encoder and decoder already have a common infrared image before encoding the grayscale image.

The design of entropy-constrained quantization is formulated as the minimization of quantization error with constraint which gives a maximum amount of information of quantized values. It is known that the optimization of unconstrained quantizer (e.g. a quantizer that minimizes summation of quantization error) is achieved by using approaches based on dynamic programming, which is called DP quantization. However, conventional DP quantization is an approach to optimize a quantizer whose quantization level is fixed. In this paper, we propose a complexity reduction algorithm for an optimal design for entropy-constrained quantizer by extending DP quantization.

Transform is one of the most important tools for image/video coding technology. In this paper, novel nonlinear transform generation based on genetic programming is proposed and implemented into H.264/AVC and HEVC reference software to enhance coding performance. The transform procedure itself is coded and transmitted. Despite this overhead, 0.590% (vs. JM18.0) and 1.711% (vs. HM5.0) coding gain was observed in our preliminary experiment.

In conventional video coding schemes such as H.264/AVC, prediction value p is obtained for the original pixel value x via inter/intra prediction and the prediction residue d = (x - p) is calculated prior to orthogonal transform and quantization. Encoding the prediction residue is nothing but ignoring the variety of distribution Pr(d\p) with respect to p, that is, using Pr(d) as a whole instead. Because H(d\p) les H(d), coding d using Pr(d) is not a good idea. For example, when p = 255, the prediction residue is always less than or equal to zero (i.e., d les 0) because 0 < x < 255 (for 8-bit case). Conventional prediction residue coding methods do not take this boundary into account and therefore involve coding inefficiency.

This paper tackles this problem by mapping the normal distribution into the geometric distribution before applying Golomb codes, which is optimal for geometric distributions. In our mapping, a pair of normally-distributed i.i.d. integers (say (x,y)) is concatenated and then mapped to one natural number z(x,y). The conditions that z shall satisfy are: minx,yZ(x,y)=0, l(x,y)z(x,y)(x,y) = (a,b), where l(x,y) is an arbitrary distance measure between the origin and the grid point (x,y), such as the Euclidean norm. The mapping can be easily obtained using a computer program. In addition, if the upper- and lower- bounds of the source is known, pre-calculated mapping table can be stored in the memory because it is independent of source statistics. Of course, this table is not needed to be downloaded/transmitted. After this mapping, z is made geometrically-distributed and conventional Golomb codes can be efficiently applied.

This paper provides a basic investigation of multispectral video coding techniques, such as interband decorrelation, visible/invisible separation, and nonlinear signal transform prior to encoding. A six-band video sequence (48 bit/pel) is used as the test data. The video/image codecs of JPEG 2000, MPEG-4 studio profile, as well as the newly developed standard H.264/AVC, are compared in a coding experiment. Investigation of the optimal encoding parameters, optimal bit allocation among components, subjective evaluation of decoded images, and optimal spatial resolution of each image band are some of the future works.

Lossless gray scale image compression is necessary for many purposes, such as medical imaging, image databases and so on. Lossy images are important as well, because of the high compression ratio. The authors propose a lossless image compression scheme using a lossy image generated with the JPEG-DCT scheme. The concept is, send a JPEG-compressed lossy image primarily, then send residual information and reconstruct the original image using both the lossy image and residual information. 3D adaptive prediction and adaptive arithmetic coding are used, which fully use the statistical parameters of the distribution of the symbol source. The optimal number of neighbor pixels and lossy pixels used for prediction is discussed. The compression ratio is better than previous work and quite close to the original lossless algorithm.>