基于改进粗糙集和NSCT的红外遥感图像增强

摘要

提出了一种基于改进粗糙集和NSCT的红外遥感图像增强方法.该方法首先利用NSCT对图像进行分解,得到多层多方向子带系数；然后根据相邻尺度和不同方向的子带中图像噪声、脆弱边缘等不同成分的系数分布,使用粗糙集对其进行分块,并制定合理的决策规则；再通过集合运算对系数中不同子块采用不同的处理方法,一方面抑制噪声,另一方面保护图像中的脆弱边缘结构,并采用增强函数对其进行不同程度的增强；最后对处理过的NSCT系数进行重构,获得增强后的红外图像.实验表明,该算法相对于其他传统红外遥感图像增强算法具有较好效果.%A novel algorithm for infrared remote sensing images enhancement based on rough sets and NSCT was proposed in this paper. Firstly, after NSCT decomposition for an image, coefficients in different scales and several directions was obtained. Secondly, considering the distribution law of coefficients of different structures in adjacent scales and different directions, we made partitions and decision rules for the coefficients based on rough sets. After that, we enhanced the different partitions of coefficients in different ways by set operation, and coefficients of cracked edges were enlarged more than strong edges. Finally, we made a NSCT composition for the modified coefficients, and the enhanced image was obtained. Compared with traditional enhancing algorithms, our method has a better noise reducing performance in preserving edges and improving the quality of images.