DCT quantization matrices visually optimized for individual images

Several image compression standards (JPEG, MPEG, H.261) are based on the Discrete Cosine Transform (DCT). These standards do not specify the actual DCT quantization matrix. Ahumada & Peterson and Peterson, Ahumada & Watson provide mathematical formulae to compute a perceptually lossless quantization matrix. Here I show how to compute a matrix that is optimized for a particular image. The method treats each DCT coefficient as an approximation to the local response of a visual `channel.' For a given quantization matrix, the DCT quantization errors are adjusted by contrast sensitivity, light adaptation, and contrast masking, and are pooled non-linearly over the blocks of the image. This yields an 8 X 8 `perceptual error matrix.' A second non-linear pooling over the perceptual error matrix yields total perceptual error. With this model we may estimate the quantization matrix for a particular image that yields minimum bit rate for a given total perceptual error, or minimum perceptual error for a given bit rate. Custom matrices for a number of images show clear improvement over image-independent matrices. Custom matrices are compatible with the JPEG standard, which requires transmission of the quantization matrix.

[1]  Heidi A. Peterson,et al.  Luminance-model-based DCT quantization for color image compression , 1992, Electronic Imaging.

[2]  Gregory K. Wallace,et al.  The JPEG still picture compression standard , 1992 .

[3]  N. Graham.  Visual detection of aperiodic spatial stimuli by probability summation among narrowband channels , 1977, Vision Research.

[4]  Joan L. Mitchell,et al.  JPEG: Still Image Data Compression Standard , 1992 .

[5]  M. A. Bouman,et al.  Spatial Modulation Transfer in the Human Eye , 1967 .

[6]  Albert J. Ahumada,et al.  Improved detection model for DCT coefficient quantization , 1993, Electronic Imaging.

[7]  Journal of the Optical Society of America , 1950, Nature.

[8]  William B. Pennebaker,et al.  Quantization of color image components in the DCT domain , 1991, Electronic Imaging.

[9]  N. Graham,et al.  Detection of grating patterns containing two spatial frequencies: a comparison of single-channel and multiple-channels models. , 1971, Vision research.

[10]  G. Legge.  A power law for contrast discrimination , 1981, Vision Research.

[11]  Gregory K. Wallace,et al.  The JPEG still picture compression standard , 1991, CACM.

[12]  Albert J. Ahumada,et al.  The visibility of DCT quantization noise , 1993 .

[13]  J. M. Foley,et al.  Contrast masking in human vision. , 1980, Journal of the Optical Society of America.

[14]  Andrew B. Watson.  Receptive Fields And Visual Representations , 1989, Photonics West - Lasers and Applications in Science and Engineering.

[15]  Jacob Nachmias,et al.  Summation of asynchronous gratings , 1980, Vision Research.

[16]  J. Robson,et al.  Probability summation and regional variation in contrast sensitivity across the visual field , 1981, Vision Research.

[17]  J. Robson,et al.  Grating summation in fovea and periphery , 1978, Vision Research.

[18]  A. Watson.  Probability summation over time , 1979, Vision Research.

[19]  H. Barlow.  Dark and Light Adaptation: Psychophysics , 1972 .