Research On Dimension Measurement Of Mechanical Parts Based On Image Recognition

With the rapid development of modern manufacturing industry,the industrial measurement technology is developing towards the direction with high speed and high precision.The traditional manual testing methods can not meet the requirements of high efficiency and high precision.Because the image measurement technology based on machine vision has the advantages of high precision,high efficiency and non-contact measurement,the method of image measurement has gradually become the focus of industrial production.Based on the analysis of image measurement technology,through the analysis of the traditional image recognition technology,the improved algorithm suitable for this project is proposed.It realizes the measurement of mechanical parts based on image recognition,and specific research work is as follows:(1)This paper analyzes the importance of the image measurement technology in the measurement of modern mechanical parts,as well as the development status and existing problems.A scheme of image measurement system based on machine vision is designed.Choose the appropriate camera,light source and other hardware equipment,etc.according to the requirements of the system to build a binocular vision measurement platform for the measurement of mechanical parts in the project.(2)The noise sources and types of the image and the corresponding filtering techniques are analyzed to choose the appropriate filtering method.Because the traditional median filtering algorithm has complexity in implementation,to solve this problem,the fast algorithm of median filtering is analyzed and the experimental results show that this method can greatly shorten the execution time of the filter.(3)With the test part image as research object,the basic principles of traditional edge detection operators,such as Sobel operator,Canny operator and traditional morphological operators,are studied and analyzed.Through several experiments,several classical operators are compared and analyzed.The traditional mathematical morphology operator is improved.Experiments show that the improved mathematical morphology gradient operator can effectively refine the image edge.(4)This paper analyzes the basic principle and algorithm steps of the Hough transform algorithm for line edge extraction,and uses the Hough transform algorithm to extract the geometric primitives(lines)in the image.In order to improve the accuracy of research,the improved sub-pixel positioning technology is used in the paper.According to the characteristics of the mechanical parts of the image edge,the method of sub-pixel edge location by linear interpolation is proposed to achieve a more accurate target edge sub-pixel positioning.(5)Programming and debugging the operating environment,running the software and conducting several experiments.The experimental results show that the improved algorithm and the experimental scheme proposed in this paper can be used to measure the size of mechanical parts,and the improved algorithm is obviously improved.Based on the analysis of the research results,this paper puts forward the ideas and directions for further research in the future.