卷积形态滤波与图像去噪

Convolution morphological filter and images noise removal

提出的卷积形态变换是一种新的形态变换形式,具有线性卷积的结构和形态变换的性质。这种新的形态变换以乘性结构元素为特征,它不同于具有加性结构元素的普通形态变换,对于它们的性质和结构的研究也是论文的主要工作之一。另一方面的工作是针对卷积形态核提出了一种结构化的自动生成算法。研究表明,卷积形态滤波与卷积积分变换一样,对于图像具有去噪和平滑作用,且在实验效果上具有较通常的形态滤波和线性卷积变换更优的去噪和平滑功能。

A novel morphological operator,called convolution morphological operator,based on the formation of convolution integral is presented.It employs muhiplicative structuring element which different from the additive structuring element in classical morphological operator.Some properties of the convolution morphological operators are discussed.An automatic generation structured algorithm for morphological kernel is proposed.Experimental results prove that an new morphological filters which constructed by the new operators performance dominates those of classical operators for images buried in salt＆pepper,speckle and Gaussian noise.