摘要

The goal of image compression is to achieve the best image quality with limited space. But it is difficult to evaluate image compression performance quantitatively for different applications. Quantization, i.e. truncation to original data, results in distortion directly in image compression process. This paper demonstrates the truncation problem into an optimization model to minimize distortion functions with constrained conditions. By constructing corresponding goal functions for special applications, the optimal R-D curves can be customized and the optimal truncation points on the curves are found iteratively. The model can embed easily into other coding methods and continue the high effectiveness. This paper presents a new distortion function, which emphasizes image details, and compares it with two other distortion functions. The experiments show image quality corresponding to different distortion functions and illustrate this method is valid.

  • 出版日期2009

全文