基于灰度相似和双邻域的磨痕角自动检测方法

An automatic wear angle detection method based on double neighborhood and intensity similarity

为了确定磨斑图像中磨痕的方向,便于后续磨斑直径的测量、抗磨性分析和异常磨损检测等处理,文中提出了一种双邻域的磨痕角自动测定方法。方法同时考虑小、大两个尺寸邻域内的灰度变化度来推算磨痕的方向角,构建小尺寸的近邻域,计算像素及其近邻域像素的灰度差以得到像素不同方位的灰度变化,求出使差异度取最小值的方位值(很接近磨痕方向);考虑磨痕角的检测精度和耗时,利用大尺寸的远邻域及远邻域上的兴趣像素,得到兴趣像素的远邻域差异度;找到与灰度差异度的极小值(即灰度相似)所对应的磨痕方位值,并结合远邻域的尺寸推算出磨痕角。以人工判定的磨痕角为基准,对200组样本数据进行仿真。实验表明,文中方法检测方向角的平均绝对误差小于1.5°,最小和最大的绝对误差分别为0°和4.309°,平均每帧耗时约2.8 s。

In order to determine the direction of wear scar in the image and facilitate the subsequent measurement of the diameter of the wear spot,analysis of the wear resistance and detection of abnormal wear,a two⁃field automatic measurement method of the wear angles is proposed in this paper.Method at the same time consider small and big size of two adjacent neighborhood to calculate the direction of wear angles according to grayscale difference value.One near neighborhood with small size is constructed,and the grayscale difference between the pixel and its neighbor pixels in different directions is calculated,and then the azimuth of what makes difference degree take the minimum value(very close to the wear scar direction).In order to improve the detection accuracy,another far neighborhood and its neighbor pixels are constructed in a similar way as near neighborhood.Considering the time consumption of the algorithm,only the luminance difference between the pixel and its interest neighbor pixels in far neighborhood is calculated.It’s necessary to find the azimuth corresponding to the minimum value of gray difference(grayscale similarity),and calculate the wear angle by combing with the size of the far neighborhood.Based on the manually determined wear angle,the simulation experiment on 200 samples data shows that:the average absolute error of the direction angle detected by the method in this paper is less than 1.5°,the minimum and maximum absolute error are 0°and 4.309°respectively,and the average running time per frame is about 2.8 s.