基于DSP的各向异性扩散滤波算法优化实现

摘要

Algorithm for Image Filtering based on Anisotropic Diffusion can control the noise of image efficiently while keep detail edges. But complicated algorithm is hard to content with the requirement of real time responsibility. This paper adopts a histogram-based statistical method to determine the median value of image gradient magnitude array. This method has improved the algorithmic efficiency of determining gradient image threshold. According to the structure and instruction characteristics of ADSPTS201. This paper optimizes and implements the algorithm. Algorithmic efficiency can meet the requirements of the real-time responsibility.%基于各向异性扩散的图像滤波算法可以在有效抑制图像噪声的同时较好地保持边缘细节,但复杂的算法很难满足系统实时性要求.采用基于直方图的统计方法来确定图像各向梯度模值数组的中值,有效提高了确定图像梯度阈值算法的效率.针对ADSP-TS201的结构及指令特点对算法进行了优化实现,较好满足了系统实时性的要求.