基于图像数据重排的产品表面缺陷检测算法研究

Research on Product Surface Defect Detection Algorithm Based on Image Data Rearrangement

产品检测时许多无关图像信息增加了参与计算的图像数据量,为解决该问题,提出了一种基于图像数据重排产品表面缺陷检测算法.根据先验知识将产品表面结构明显且不会出错的特征作为产品成像后的周向位置特征,建立起样本周向位置特征库图像序列;同时提取每个周向方位下样本表面各个待识别子目标区域图像数据或变换特征进行有效数据重排作为产品的检测识别特征,建立起产品的检测识别特征库图像序列.检测识别过程中,提取产品位置特征寻找其在标准样本库中最优解确定产品的周向方位,然后提取产品不同方位下的多个待检测区域并进行相同的数据重排,最终完成快速检测识别.经过仿真验证,本文的算法可以大大节省缺陷检测所用时间.

During product inspection,many unrelated image information increase the amount of image data involved in the calculation.In order to solve this problem,an algorithm for rearrangement of product surface defects based on image data is proposed.Based on the prior knowledge,the features with obvious characteristics and no error on the surface of the product is taken as the circumferential position features after imaging,and then the library is established with the sample circumferential position features.At the same time,the image data or the transformed features of each sub-target region,which need to be inspected on the sample surface under each circumferential orientation,are extracted for effective data rearrangement as a detection and recognition feature of the product,and then the library is established with the detection and identification features.In the process of detecting and identifying the product to be tested,extracting the location features and finding the optimal solution in the standard sample library can determine the current orientation of the product,and then to-be-detected areas in different orientations of the product are extracted and used in the same data rearrangement,and finally the rapid detection and identification are completed.After simulation verification,the algorithm in this paper can greatly save time for defect detection.