基于侧抑制网络的改进Canny边缘检测算法

Improved Canny Edge Detection Algorithm Based on Lateral Inhibition Network

边缘检测是红外图像目标自动识别的关键步骤。针对红外图像对比度低、边缘模糊等问题,提出了一种改进的Canny边缘检测算法。该算法基于人类视觉系统的侧抑制竞争原理,利用侧抑制网络增强中心、抑制周围的特点,对原图像处理达到增强对比度、抑制噪声边缘的目的;采用OTSU实值法,自适应选取双阈值进行边缘检测。实验证明,该算法能够有效排除干扰边缘,对提高目标的边缘检测准确度有较大的实际意义。

Edge detection is the essential step in automatic target recognition of infrared images.An improved Canny edge de⁃tection algorithm is proposed for the purpose of detecting edges of infrared images with low contrast and blurred edges.Based on the lateral inhibition competition principle of human visual system,the algorithm utilizes the features of lateral inhibition network to en⁃hance the center and restrain the surrounding area,and achieves the purpose of enhancing contrast and restraining weak edges for the original image processing.This paper uses OTSU algorithm to select the two thresholds adaptively for edge extraction.Experi⁃ments show that the new algorithm can eliminate interference edge and has better edge detection effect for infrared images.