一种基于自适应轮廓波变换的图像编码方法

An Image Coding Method Based on Adaptive Contourlet Transform

为实现图像更加高效的稀疏表示,本文提出一种基于自适应轮廓波变换的图像编码方法。该方法基于熵准则,能够根据图像的频率分布,自适应地划分子带,从而实现图像的高效稀疏表示,提高编码质量。将本文所提算法用于多种标准测试图像的编码,分别与基于脊波、曲线波、轮廓波等变换的编码方法相比,实验结果表明,本文所提方法均优于上述方法,同等码率下峰值信噪比提升约0.5-2dB,且主观重构效果明显提高,证明了本文所提方法是有效可行的。

In order to achieve more effective sparse representation for images,in this paper we proposes a new image coding method based on adaptive contourlet transform. This method is based on entropy criterion,and can partition subbands adaptively according to the frequency distribution of images,so as to realize the efficient sparse representation for images and improve the coding quality. The proposed method is used on various standard test images,and compared to ridgelet,curvelet,contourlet based methods. The experimental results show that the proposed method is superior to these aforementioned methods and improves the PSNR of about 0.5-2 dB under the same bit rate. The visual quality of the reconstructed images is obviously improved as well.