Development Of Visual System For Casting Surface Defect Detection

In the production process of castings the castings themselves will have defects due to various production reasons.The defect detection of castings is mainly detected by manual identification,but some manual identification of defects is not applicable,such as cracking.However,by adsorbing magnetic powder and fluorescent irradiation,the defect can be displayed.Since the light source is fluorescent,it will cause irreversible damage to the human eye that human directly looking at the fluorescence for a long time.Therefore developing a set of industrially applicable precision casting defects inspection system is very suitable for industrial needs.This subject is to develop a set of plate tooth defect system according to the actual needs of the plate tooth manufacturer,which can automatically detect the plate teeth produced by the manufacturer,and automatically save the pictures of the defective plate teeth for subsequent manual confirmation and System improvements.(1)Complete the overall design of the plate tooth defect according to the requirements,including the overall solution and technical realization of the defect identification,the selection of hardware devices,such as cameras,lenses,light sources,etc.,and the design of the software part.(2)Target extraction of the plate teeth.The image pre-processed by the camera is pre-processed to remove noise,and then the Blob analysis and template matching method are used to realize the extraction of the plate teeth.By comparing the extraction effects and subsequent requirements of the two extraction methods,the method of template matching is finally determined.Extract the plate teeth.(3)Plate tooth defect extraction.After the target extraction of the plate teeth,the outer contour of the plate teeth is removed by morphological processing,the internal contour is removed by drawing the ROI and the coordinates obtained by the template matching.And the smal lines are extracted by the feature selection,then combined by two lines.Some design of the parameters makes the interference lines basically impossible to combine,and the lines of the defective parts can be combined.Finally,the feature lines are extracted from the combined lines,and the lines of the defective parts are selected to complete.Extraction of defective parts.(4)Programming with VS development tools,and combined with the industry common machine vision software Halcon,finally completed the development of the defect detection system for the plate teeth,and tested the accuracy,detection accuracy and real-time performance of the system in the laboratory.