【摘要】 提出了一种新的图像压缩算法,该算法在K奇异值分解(K-SVD)的基础上,将图像分解成4×4像素的图像块,进行字典学习和稀疏表示,完成图像压缩编码存储;在图像恢复的过程中,通过使用字典学习前保留的边缘,对恢复图像进行修复。实验表明:在压缩比为20:1时,该算法的峰值信噪比(PSNR)较JPEG算法高出4 d B;用边缘信息修复后,较JPEG算法高出近10 d B,比JPEG2000算法高出3 d B。更多还原

【Abstract】 A new efficient image compression algorithm was proposed based on K singular value decomposition( K- SVD). The algorithm decomposes an image into 4 × 4 pixel image blocks,completes image compression storage by dictionary learning and sparse representation. In the process of image restoration by using the dictionary before learning to retain the edge of the restored image to be restored to get better image restoration effect. Experimental results show that in the compression ratio of 20:1,the peak signal to noise ratio( PSNR) is 4 d B higher by using the proposed algorithm than the JPEG algorithm. With the edge information restoration,the PSNR is about 10 d B higher than the JPEG algorithm and also even 3 d B higher than the JPEG2000 algorithm.更多还原