摘要： 针对计算机桌面图像压缩问题，提出一种基于颜色聚类的图像压缩算法。将桌面图像划分成16×16的非重叠块，归为文本/图形块、自然图像块及混合块3类。对色彩丰富文本/图形块进行颜色聚类，以降低块的颜色种类数，并做无损压缩。对自然图像块采用H.264帧内预测编码方法，对混合块采用混合编码方法。实验结果表明，该算法所得图像的峰值信噪比和结构相似度均高于传统算法。

关键词: 计算机桌面图像, 混合图像, 块分类, 颜色聚类, 图像压缩, 图像编码

Abstract: This paper proposes an image compression algorithm based on color clustering to compress computer desktop image. It divides computer screen image into 16×16 non-overlapping blocks, then each block is classified into text/graphic block, hybrid block or picture block. For text/graphic blocks with rich color, a color clustering method is used to reduce the number of colors, then text/graphic block is coded by lossless compression. Natural image block is coded by H.264 intra prediction coding method. Hybrid block is coded by hybrid coding method. Experimental results show that the image by proposed algorithm has higher PSNR and SSIM than traditional algorithms.

Key words: computer desktop image, compound image, block classification, color clustering, image compression, image coding

中图分类号: 

• TN911.73