数学形态学滤波器的优化设计

Optimized Design for Mathematics Morphology Filter

数学形态学滤波器作为图像处理领域中的一类重要的非线性滤波器已被广泛地用于图像降噪、边缘检测和模式识别等图像处理技术之中,然而其处理效果的好坏却严重地依赖于结构元素的形状和大小。在传统的形态滤波器的设计中,结构元素的选择完全凭借设计者的经验,因而很难保证所选择的结构元素为最优结构元素。针对此问题,该文应用遗传算法对形态滤波器设计中的结构元素进行优化,通过对样本图像的学习训练,获得基于优化结构元素的形态滤波器,并在此基础上设计了一种简单、实用的自适应优化滤波算法。该算法的有效性用计算机仿真实验进行了验证。

Mathematics morphology filters have been used popularly in digital image processing for filtering noise,detecting edge and recognizing object as an important nolinear filter.However,the effect of processing image is greatly dependent on the size and shape of structure element.Because selecting of structure element relies on the experience of designer in classical design of morphology filter,it is difficult to ensure that the structure element used for processing image is best.To solve the problem,genetic algorithm is applied to optimize structure element by means of learning to sample image and obtain the best morphology filter based on optimized structure element.Based on this,a simple and practical filtering arithmetic is designed for reducing the salt and pepper noise in gray image.Validity of the arithmetic is affirmed by simulation test.