ACCELERATION IMAGE COMPRESSION WITH REQUIRED VISUAL QUALITY

A.N. Zemlyachenko, E.O. Kolganova, V.V. Lukin

Automatic methods for lossy compression of images with required visual quality are considered. Metrics PSNR–HVS–M and MSSIM that take into account human vision system are used for controlling visual quality. For a large number of different images, iterative compression procedures are analyzed. It is shown that the number of iterations can be greatly reduced due to choice of the limits of possible change of parameters that control compression ratio. The results of the proposed solutions are presented for different encoders. Their comparative analysis is performed and it is shown that the compression ratio depends on characteristics of compressed images, and it can be significantly increased due to using modern compression methods tailored to visual perception.

Keywords: acceleration lossy compression, image visual quality metrics.