JPEG 2000 is an international standard for still image compression, which is based on wavelet transformation and the Daubechies 9/7 filter adopted for lossy compression. Considering that a wavelet filter might be suitable for one image but not for the other in regard to compression quality, in this paper, we propose a novel filter design framework based on the Daubechies 9/7 filter, which employs chaos evolution programming (CEP) to optimize the wavelet filter for both the universal images and each specific image, respectively. The customized filter design is ready to incorporate into the JPEG 2000 compression/decompression since the filter coefficients can be constructed by only one tuning parameter, which can be easily packaged into the modified JPEG 2000 header. Experimental results show that CEP-trained filters achieve higher image quality.
All Science Journal Classification (ASJC) codes
- Control and Systems Engineering
- Signal Processing
- Computer Vision and Pattern Recognition
- Electrical and Electronic Engineering