基于改进Canny算法的齿轮表面边缘检测  

Gears surface edge detection based on improved Canny algorithm

作　　者：李小龙 闫娟 杨慧斌 刘向前 LI Xiaolong;YAN Juan;YANG Huibin;LIU Xiangqian(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学机械与汽车工程学院,上海201620

出　　处：《智能计算机与应用》2022年第10期180-183,共4页Intelligent Computer and Applications

摘　　要：齿轮的边缘检测是保证齿轮使用安全的重要手段,传统的Canny边缘检测算法对噪声敏感,但是在滤除噪声时容易丢失弱边缘信息,其固定参数适应性较差。针对上述问题,本文提出了改进的Canny边缘算法用于齿轮表面边缘检测,将离散余弦变换(DCT)和非局部均值(NLM)相结合,取代了Canny算法中传统的高斯滤波器。实验结果表明,该方法在边缘检测和噪声抑制方面具有良好的效果。Gears edge detection is an important means to ensure the safety of gears use. The traditional Canny edge detection algorithm is sensitive to noise, but it is easy to lose weak edge information when filtering noise, and its fixed parameter adaptability is poor. To solve the above problems, an improved Canny edge algorithm is proposed for gears surface edge detection. The combination of discrete cosine transforms(DCT) and nonlocal mean(NLM) replaces the traditional Gaussian filter in Canny algorithm. The experimental results show that this method has good results in edge detection and noise suppression.

关 键 词：边缘检测 CANNY算子 DCT系数 非局部均值 齿轮 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]