摘要: 在各向异性扩散(AD)的图像处理应用中,扩散终止机制是当前研究的主要热点。该文把热物理扩散中的扩散平衡概念引入到图像处理的AD中,定义图像扩散平衡(DB)与最优扩散时间(ODTS)的概念,提出利用扩散平衡估计最佳扩散时间尺度的方法。该方法与SURE, MSE及其他时间尺度估计方法相比在噪声压制与特征的保护上达到了一个合理的平衡且不需估计噪声强度(方差),能自适应于图像,在低信噪比下也能获得很好的估计结果,试验测试结果验证了该方法的有效性。
关键词:
各向异性扩散,
最优扩散时间尺度,
扩散平衡
Abstract: A stopping mechanism of anisotropic diffusion is one of interests in image processing at present. This paper introduces thermal diffusion to anisotropic diffusion for image denoise and restoration, and defines the concepts of Diffusion Balance(DB) and Optimal Diffusion Time Scale(ODTS). It proposes a novel estimation method of the optimal time scale based diffusion balance. The method can be adapted to image and need not be estimated noise intensity(variance) compared with SURE, MSE and others, and denoise and feature preservation can be reasonable balanced. The method can work well in the case of low SNR, the test results show it is effective.
Key words:
anisotropic diffusion,
optimal diffusion time scale,
diffusion balance
中图分类号:
严家斌;刘贵忠. 一种新的各向异性扩散时间尺度估计方法[J]. 计算机工程, 2008, 34(2): 43-45.
YAN Jia-bin; LIU Gui-zhong. Novel Time Scale Estimation Method of Anisotropic Diffusion[J]. Computer Engineering, 2008, 34(2): 43-45.