梯度倒数加权平滑算法的改进与实现

Improvement and Implementation of Gradient Inverse Weight Smoothing Algorithm

根据图像的相关性和连通性原理,利用图像各部分的梯度信息,基于原梯度倒数加权平滑算法,提出了一种改进的梯度倒数加权平滑算法。处理结果表明,改进算法较原算法能够更好地去除椒盐和随机噪声,同时较好地保持了图像的边缘和细节信息,处理过程的运算复杂度与原算法相当。改进的梯度倒数加权平滑算法为目标识别、图像分割等后继处理与分析提供了有力的支持。

Based on the theory of relativity and connectivity of image and characteristic of each part, the paper presents a new improved gradient inverse weighting smoothing algorithm. The results shows that the improved algorithm can preserve the image edge and details perfectly and it, comparing to the old algorithm, can restrain the salt & pepper noise and impulse noise more efficiently. Furthermore, the running time of the improved algorithm is almost equal to the old one’s. It provides powerful supports for later procession and analyses, such as object identification, image segmentation and so on.