On Machine Vision Based Surface Flaw Detection System For Minitype Metal Parts

With the development of computer technology,the machine vision technology,which has advantages of non-contact and indefatigability comparing with traditional detection by people,can reduce the detection cost,strongly raise the efficiency of production, effectively ensure the quality of products,and raise value added of products.It has been used widely in the field of industrial online detection.On minitype metal parts detection, because of its complicated mechanism of production for surface flaws,surface flaws are versatile.Therefore,there are still many problems to be studied and resolved so as to apply the machine vision technology more widely and successfully.Based on the related literature and technology applications of the metal defect detection,the principle and approach of the minitype metal parts detection was studied by taking Ni-Fe-B workpiece as an example.Then,an online real-time detection system based machine vision algorithm was developed,which applied high-speed and high-resolution plane array sensors and had complete configuration and function of software and hardware, as well as user-friendly interface.Regarding surface flaw detection system of Ni-Fe-B workpiece,an implementation scheme for detection system was demonstrated firstly in this dissertation.And then regarding the features of workpiece flaws,low level dark field ring light resolution was designed based on the research of several common lighting systems.Also imaging mechanism and model of camera was analyzed to choose appropriate camera and camera lens,and then design and demonstration for image gathering resolution was achieved.Several image processing algorithms was introduced to meet the requirements of machine system.To solve the problem that mini flaws were easily confused with outside background of the workpiece when detected,an approach of selecting detection field with polygon Region of Interest(ROI) whose coordinate changed automatically with automatic locating result for workpiece was proposed as a result of high accuracy and low difficulty to the system.Regarding of existing abundant marginal points in workpiece image,a new approach of sequential similarity detection algorithm(SSDA) based on feature of marginal points was proposed on the research of several template matching algorithms,which solved the problem that other algorithms can't meet the requirements of real-time character in online detectiong system and was validated also by experiments.Then OTSU algorthm was combined with fixed threshold segmentation algorith to achieve image binaryzation based on research of threshold segmentation algorithm.And also surface targets were counted and their features were extracted based on the study and analysis for pixel tagged algorithm to judge the quality of workpieces.At last,the connection between camera and software platform,initialization of camera, image gathering and image processing were achieved based on VC++ 6.0 platform and API functions supplied by Alltime Image company to implement the development of software detection system.