基于自适应正交有限脊波变换的图像编码算法

在有限脊波变换的基础上,根据对图像有限Radon变换系数的分析,提出了一种自适应的正交有限脊波变换。此变换根据特别设计的代价函数,可自适应地为每列有限Radon变换系数选择最优的正交变换形式,从而可高效地分析图像中的直线奇异。讨论了此变换在图像编码领域中的应用,通过对零树编码算法SPIHT进行改进,实现了图像自适应正交有限脊波变换系数的渐进编码。实验结果表明,对富含直线边缘的图像来说,与小波图像编码算法相比,该算法获得了较高的边缘重建质量。

Image coding based on adaptive orthonormal finite ridgelet transform

An adaptive orthonormal finite ridgelet transform is proposed based on the finite ridgelet transform.Depending on the cost function designed,this transform adaptively chooses the optimal orthonormal transform for each column of the finite Radon transform coefficients,so that it can deal with the linear singularities more efficiently.Furthermore,this transform was applied to the field of image coding,and SPIHT was modified to generate embedded bitstream for the adaptive orthonormal finite ridgelet transform coefficients.Experimental results show that for the straight line-rich images this algorithm can get better edge reconstruction quality compared with the traditional wavelet-based image coding.