Research And System Development Of Part Geometry Measurement Technology Based On Machine Vision

The geometric measurement of parts occupies an important position in industrial production.The test results not only affect the qualification of the parts themselves,but also play a decisive role in the subsequent reprocessing of the parts and the overall assembly of the product.With the improvement of China's industrial manufacturing level,modern industry has put forward higher requirements for measurement technology.At this stage,traditional measurement methods can no longer meet the heavy quality inspection tasks and increasingly strict accuracy requirements.Machine vision measurement technology has been widely studied and applied to the field of part geometry measurement because of its advantages such as high measurement efficiency,good stability,and non-contact.Aiming at the problems of single measurement items and insufficient accuracy research in the current research,this thesis designs and develops a set of machine vision measurement systems that can measure multiple geometric features,and conducts research and experimental analysis on key technologies that affect measurement accuracy.Firstly,the core image processing algorithms in machine vision were studied.Through the calculation principles of the algorithm,simulation and comparative analysis,the comprehensive consideration was made from the aspects of algorithm performance,operation speed,and applicability.Finally,a set of image filtering,threshold segmentation and edge detection algorithms that meet the needs of this subject are selected.Then researched the two key technologies of the system:image edge detection algorithm and camera calibration technology.The gray moment sub-pixel edge localization algorithm was introduced and compared with Canny operator.It is verified that the gray moment method has higher edge positioning accuracy and better anti-noise performance.The principle and calibration steps of calculating pixel equivalent calibration and Zhang Zhengyou calibration method are also described.In terms of geometric quantity measurement,the algorithm of region of interest extraction was introduced,and the measurement algorithm was implemented in combination with a mathematical model.Then the function of the measurement software was analyzed and modularized.The system measurement software and supporting database system were developed.Through the selection of system hardware and measurement software,a part vision measurement system based on machine vision was constructed.Finally,the standard ring gauge,gauge block and angle gauge block were used to measure the diameter,roundness,distance,straightness,parallelism,and angle.Experiments were performed on two different edge detection operators and system calibration techniques.The effect of accuracy summarizes the experimental conclusions.The research results have realized the measurement of geometrical quantities of mechanical parts based on machine vision,and have important application value in automated production.