Development Of High Speed Screening System For Precision Parts Based On Machine Vision

With the development of industrialization,there are more and more kinds of mechanical manufacturing equipment,but the basic dimension inspection as the inspection standard of manufacturing industry is lagging behind,especially in the field of finish machining,traditional detection methods,represented by manual detection,still occupy a large proportion,not only time-consuming and labor-consuming,but also have too much variability of detection accuracy.Therefore,it is urgent to strengthen the detection of small parts in the field of finish machining.Based on machine vision image processing technology,this paper designs a high performance-price ratio,high precision and high efficiency measuring system for the precision machining of special-shaped studs in 3C industry.In order to improve the detection precision,this system analyzes the construction of the detection system,the software programming flow,the image processing technology and so on,and finally realizes the non-contact measurement which meets the 0.02 mm detection requirement.The main contents of this thesis are as follows:First,design the hardware and software system for the project.According to the requirements of the project and the characteristics of the work piece to be tested,the characteristics of the system hardware and the minimum performance index are analyzed.The software is programmed by the combination of c # and OPENCV image library.The software system includes the design of image processing flow and so on.In order to improve the processing speed,the system cuts the image before processing the image algorithm.Secondly,Image preprocessing is carried out according to the algorithm flow.Firstly,the gray-level conversion is carried out to simplify the number of image processing channels and improve the processing efficiency.After analyzing the noise sources of the system and the common noise reduction methods,the median filter is selected to reduce the noise after the experiment,the OTSU automatic threshold is used to segment the region of interest and solve the adaptive threshold,which is convenient for the next step.Then,image analysis and processing are carried out after preprocessing.Compared with the traditional edge detection technology,Canny operator and adaptive threshold are selected as the system’s rough edge location algorithm.Based on this,the sub-pixel processing technology of the special-shaped bolt image is analyzed,and the system selects the BILINEAR interpolation for sub-pixel processing,and the detected floating-point edge coordinates are fitted to obtain the high precision work piece feature information,the line is Hough combined with the least squares,and the circle is least squares.Finally,an experimental platform is built to calibrate the system from Pixel to physical scale,and the measurement items are analyzed according to the requirements of the project,and the method of measurement is programmed.Experimental results show that the maximum accuracy of the system is 0.0076 mm,which meets the requirements.