Research on Image Compression Algorithm Based on Predictive Residual Coding
Yujie Guo, Wuli Zhou, Xiwei Peng, Yixuan Meng · 2025
In the field of image compression, lossless compression algorithms achieve high image quality, while lossy compression algorithms achieve high compression ratios. However, traditional algorithms struggle to balance image quality and compression ratio, and their encoding strategies often lack adaptability to varying characteristics across different image regions. To solve the above problems, an image compression algorithm based on predictive residual coding is proposed, which divides pixels into blocks according to their characteristics, selects the best one after traversing all prediction modes, calculates the prediction residual and performs integer transformation, quantization and encoding to obtain a compressed bitstream. The experimental results indicate that, compared to the traditional Joint Photographic Experts Group (JPEG) algorithm, the proposed algorithm achieves a 37.7 % improvement in compression ratio and a 19.4 % improvement in Peak Signal-to-Noise Ratio (PSNR) on the Kodak dataset,while the compression ratio increases by 29.5 %, and PSNR increases by 21.7 % on the Divik subdataset.