Abstract: A new method for image compression which colligated the strongpoint of DCT and Cycle Spinning is presented.Because an image is cut apart during the DCT compressing process and DCT lacks of translation invariance，a direct application of DCT to compress an image will lead to the blockness and paeudo-Gibbs phenomenon（lead to artifacts）.To avoid the artifacts，cycle spinning is employed.The experiment results indicate that this method can get better visual effect and high PSNR value compared with the simple DCT method.

Key words: image compression, lossy compression, DCT, Cycle Spinning

摘要： 综合利用DCT变换和Cycle Spinning，提出一种新的图像压缩方法。由于图像的DCT压缩是分块完成的，且DCT变换缺乏平移不变性，直接应用DCT变换对图像进行压缩会使压缩重构图像产生块效应和伪吉布斯现象，导致图像失真。引入Cycle Spinning算法能有效地消除图像DCT压缩过程中产生的失真。实验结果表明，与一般图像DCT压缩方法相比，该方法明显改善图像视觉效果，显著提高图像的PSNR值。

关键词: 图像压缩, 有损压缩, DCT变换, Cycle Spinning