基于灰度阈值分割的锥体零件缺陷识别算法

根据人的视觉感官特性，针对锥体零件缺陷检测中的图像信号预处理，利用含有缺陷的灰度图像的灰度统计特征自动找出分割图像的灰度阈值．研究了目前使用边缘检测、阈值分割、区域增长等方法．利用曲线拟合方法计算出灰度的分布曲线，用数学分析的方法求出曲线的局部极值．对于出现多个极值情况时进行分析筛选，把分割缺陷比较理想的极值作为灰度分割的阈值．以此分割图像，其分割效果比较理想．

Defect Recognition Algorithm of Cone Object Based on Gray-level Threshold Segmentation

Considering the human's vision speciality,aims at the pretreatment of image signal of defect detection of cone object,age including some defect an algorithm is proposed to find the threshold,with defect statistic feature of gray-level image including defects.Some techniques,such as edge detection,threshold segmentation and region increase,are investigated.The distribution ccnve of image gray-level is calculated by fitting curve method.The local extremum of distribution-curve is solved by mathematic analysis.For multiple extremums,analysis and filter methods are used to search the correct extremum which is to be adopted as threshold of gray-level segmentation.The threshold is taken to segment and recognice image defect,the segmentation effect is ideal.