Linear-Regression Model Based Wavelet Filter Evaluation for Image Compression
Gaofeng Wei, Hongxu Jiang, Rui Qin Yang · 2010
In current wavelet-based still image compression, the choice of wavelet filter is of great importance, and many issues relating to it remain unresolved. This paper presents a method on linear-regression model based wavelet filter evaluation for image compression. This method analyses several factors (including image's brightness, image's contrast, filter's ability to concentrate energy, and entropy distribution on bit-plane of wavelet transform coefficients), and evaluates wavelet filters by predicting the quality of restored images. In this way, it provides an accurate and objective way to evaluate wavelet filter for image compression without encoding or decoding, and has certain significance on the study of evaluation and choice of wavelet filters.