Research On Detection And Positioning Of Automobile Exterior Cover Based On Machine Vision

At present,automobile maintenance spraying mainly relies on manual operation,and the quality of spraying is limited by the technical level of the operator,and the efficiency is not high.In addition,due to the volatilization of toxic substances in the paint,the working environment gas is polluted.For a long time in this kind of harsh environment,the damage to the workers is extremely great.But the robot automatic spraying technology research is of great significance,and the automatic detection of parts and components And positioning is a technical difficulty to achieve automatic spraying.In this context,this paper proposes a machine vision-based detection and positioning scheme for automotive exterior covers,which solves the key automatic positioning problem for automotive maintenance coating.The main research contents of this thesis are as follows:firstly,the research source of the subject and the research status of related technologies are analyzed,and the research significance of using machine vision to locate the outer cover of automobile is clarified.Secondly,the light source type is determined for the operating environment and performance comparison,and the camera and lens model are selected according to the actual project application analysis,thereby establishing and calibrating the binocular vision system,obtaining internal and external parameters,and providing technical support for the subsequent three-dimensional reconstruction.Thirdly,due to the influence of the spraying environment and the detection accuracy requirements of the automobile outer cover,the image preprocessing is put forward higher requirements.a "neighborhood" is proposed for the image information fuzzy problem caused by the median filtering in smoothing noise.The most value determination method can preserve the original pixel value of the edge position,improve the image edge positioning processing and system real-time.Fourthly,in the aspect of detecting the characteristics of the outer cover of the automobile,two kinds of edge information processing algorithms are designed for the shortcomings of the previous single pixel-level edge detection method: for the smoother contour of the car,the processing at the original pixel level The Canny-Zernike sub-pixel edge detection method is proposed in the way;for the image information part of the workpiece with more corner information and more complicated edges,the feature detection is applied by corner detection,which is to improve the noise of the existing corner detection.A weighted filtering adaptive Harris corner detection method is designed.Finally,the three-dimensional coordinate recovery of the feature points is carried out on the basis of image stereo matching.The improved algorithm flow in the various aspects of visual measurement is applied to the positioning system.The experimental results show that the measurement accuracy of the positioning system can meet the requirements of automatic spraying technology.