基于自适应分割和自适应量化的图像压缩算法

Image compression method based on image adaptive segmentation and adaptive quantification

为弥补JPEG算法不能自适应不同属性图像的不足,提出一种基于自适应分割和自适应量化的图像压缩算法。利用多阈值分割将图像分割成不同尺寸的平滑子块和细节子块,不同子块采用不同DCT变换和量化方式.本文定义的压缩-信噪比增益(CPR)同时考虑了压缩比(CR)和峰值信噪比(PSNR),可有效评价图像压缩效果.实验结果表明,与传统JPEG算法相比,该算法在PSNR相同情况下,能够获得更高的CR;或者在CR相同情况下,能够获得更高的PSNR.

To make up the deficiency of JPEG method not self adapting to image with different properties, an image compres sion method based on adaptive segmentation and adaptive quan tification was developed. The image was divided into smooth and detail subimages with different sizes by using the multi thresholdsegmentation method, and every subimage used dif ferent DCT transform and quantification manner. Compression toPSNR rate（CPR） was defined by considering the compression rate（CR） and the peak signal to noise ratio（PSNR） simulta neously, which could evaluate image compression result effec tively. Experimental results show that the proposed method can give superior performance comparing with traditional JPEG method, and it has higher CR at the same PSNR, and higher PSNR at the same CR.