Research On The Size And Defect Detection System Of Door Panel Based On Machine Vision

With the continuous research and development of computer vision detection technology and the increasing level of industrial automation production,the product detection technology based on machine vision and industrial production line has the advantages of high detection accuracy,fast detection speed,high recognition rate,high safety,easy to save and query data,and gradually replaces the traditional product quality inspection of manual use of auxiliary equipment detection and recognition In this paper,based on the study of the on-line quality detection technology and image processing algorithm based on machine vision,the design of the door panel size measurement and the surface micro scratch detection scheme under the background of rust corrosion,and through the programming of the visual software for the detection project,the results show that the visual detection has the advantages of high measurement accuracy and accurate defect extraction.The main contents and innovations of this paper are as follows.In this paper,firstly,by analyzing the size measurement parameters of the steel anti-theft door panel and the scratch detection requirements under the rust texture,the measurement scheme of double cameras collecting side by side is designed,and then through the calculation to meet the measurement accuracy requirements,a reasonable experimental test platform is designea to select the camera with stable shooting and small distortion.The advantages and disadvantages of commonly used light sources and the influence of lighting methods on image recognition are analyzed.Finally,according to the gray distribution of scratches under the rust texture of the door panel,the light source category is reasonably selected,and the lighting method that can most restrain the rust texture from protruding scratches is designed.By analyzing the advantages and disadvantages of different edge detection operators and the actual edge extraction effect under the interference of rusty texture,and carefully studying the principle of the best line and circle fitting algorithm based on the least square method applied in industrial measurement at present,finally adopting the method of Canny operator edge detection combined with the least square method curve fitting to get the subpixel detection of the edge line and circle of the door panel Accuracy;compared and analyzed the accuracy difference between the two camera calibration measurement schemes and the local area calibration measurement schemes for single camera,the proposed local area calibration method for single camera can effectively improve the measurement accuracy and reduce the cost;considering the actual pipeline production mode,analyzed the advantages and disadvantages of a variety of template matching methods,and selected the one with the highest matching degree based on shape The template matching algorithm based on shape is used to realize the size tracking detection when the door panel moves.Through the analysis of scratch information under the rust texture,we can see that the scratch under the rust texture has the characteristics of shallow depth,irregular distribution,serious background interference,etc.after careful study,it is found that the scratch and the rust texture are nested each other,even the most advanced adaptive local neighborhood detection algorithm can not extract the scratch under the rust texture.In this paper,an adaptive filter for scratch detection under rusty texture is designed innovatively in frequency domain through Fourier transform,which not only filters out the interference of rust,but also adapts to the situation that the door plate is placed in different directions and the position of scratch appears,and realizes the real-time tracking detection of scratch defect when the door plate moves.According to the threshold selection of scratch binarization under different illumination after Fourier transform The dynamic threshold segmentation algorithm is proposed innovatively for different problems that need to be set manually.It is suitable for scratch adaptive binary detection when the light intensity in different shooting areas of the door panel is uneven and the light intensity changes.