A model for JPEG quantization

Presents a strategy for generating optimal JPEG quantization tables to approximate a target compression ratio. This uses a model to express the quantization coefficients as functions of compression ratio and their position in the quantization table. Simulated annealing of model parameters was used to find optimum models for an image which is a composite of several standard test images. Models of varying complexity with 1 to 6 parameters were optimized at three compression ratios, and a three parameter model was chosen to represent quantization tables. After further optimizations over a range of compressions, a general model was obtained by expressing each model parameter as a function of the compression. Application to three CCITT test pictures demonstrates the quality of recovered images.<<ETX>>

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

[2]  C. D. Gelatt,et al.  Optimization by Simulated Annealing , 1983, Science.

[3]  Michele Marchesi,et al.  Applications of simulated annealing for the design of special digital filters , 1992, IEEE Trans. Signal Process..

[4]  Barry G. Sherlock,et al.  Optimum DCT quantization , 1993, [Proceedings] DCC `93: Data Compression Conference.

[5]  Allen Gersho,et al.  Globally optimal vector quantizer design by stochastic relaxation , 1992, IEEE Trans. Signal Process..

[6]  N. Ahmed,et al.  Discrete Cosine Transform , 1996 .

[7]  Teresa H. Meng,et al.  Optimal quantizer step sizes for transform coders , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[8]  David E. Goldberg,et al.  Genetic Algorithms in Search Optimization and Machine Learning , 1988 .