形态学多结构元多尺度熵权边缘检测方法研究

数学形态学广泛应用于图像处理和模式识别领域.针对形态学单结构元在边缘检测中边缘信息丢失的问题,提出了基于形态学多结构元多尺度熵权边缘检测方法.首先利用形态学高低帽运算对原始图像进行增强处理,由形态学运算调整结构元素尺度,采用抗噪型算子进行边缘检测,依据边缘图像的信息熵确定权值进行融合,改进了数学形态学边缘检测算法.实验结果表明,与传统边缘检测算法相比,该算法在保持图像边缘清晰的同时,有很强的去除噪声能力.

Edge Detection Based on Entropy Fusion of Morphology Multi-Structuring Elements and Multi-Scale

Mathematical morphology is applied widely in image processing and pattern recognition.The traditional method loses some edge information.In order to improve edge detection effective,this paper proposes a new edge detection method based on entropy fusion of morphology multi-structural elements and multi-scale.First,the original image is filtered using top-and-bottom hat operation.Second,the scale of structuring elements can be determined by morphology operation.A new edge image with a better quality can be extracted using de-noising algorithm,and fused by their entropy.This method can improve the traditional algorithm.Experimental result indicates that the new edge detection achieves better image processing effect than traditional method,and has strong ability of eliminating noise as well as keeping clear image edge.