基于最大模糊熵的红外图像边缘检测算法

Edge Detection of Infrared Image Based on Maximum Fuzzy Entropy

针对红外图像边缘检测这一难题,结合红外图像及其梯度图像的特点,在红外梯度图像模糊划分的基础上,提出了一种基于最大模糊熵的红外图像边缘检测方法。首先通过改进传统Sobel算子构造出红外图像的梯度图并研究其直方图特点,然后对其进行自然模糊划分,最后根据最大模糊熵准则确定最优模糊参数,进而确定梯度图像的最佳分割阈值,从而实现边缘提取。与传统的基于梯度的边缘检测算法进行对比实验,结果表明,该方法用于红外图像边缘检测能获得更好的效果。

Pointing at the difficulty in the edge detection of infrared image, the characteristics of the infrared image and its gradient image are analyzed. Through fuzzy partition to the gradient image, edge detection of infrared image based on maximum fuzzy entropy is introduced. First, the gradient image is constructed by improving the Sobel operator, and its characteristics is studied; then, implemented fuzzy partition to infrared gradient image; at last the optimal fuzzy parameter is obtained from the rules of maximum fuzzy entropy. At the same time, the optimal segmentation threshold value is achieved, and the edge detection is realized finally. The results show that the proposed approach has better performance than some classical edge detection methods based on gradient.