Improved JPEG compression algorithm based on saliency maps

摘要

JPEG compression algorithm plays an important role in modern image storage and transmission field. This paper proposes an improved JPEG compression algorithm combined with saliency maps in quantization. It can realize layered compression based on users' attention degree of image content and improve compression ratio on the premise of ensuring the quality of regions where users pay attention. This paper firstly analyzes and realizes of the algorithm based on saliency map. After that, it demonstrates the algorithm's reliability and effectiveness by experiments with different kinds of images. At last, threshold control, which may further improve compression efficiency, is discussed briefly.