Research On Large-size Workpiece Measurement Technology Based On Line-scan Camera

In recent years,with the rapid development of precision manufacturing industry in China,especially in the fields of aviation,shipbuilding and other fields,the demand for large parts and components is increasing,and the demand for precision measurement of various large workpieces is also increasing.Machine vision as a typical way of high precision measurement has been widely used in many industrial applications for measurement and inspection.Machine vision based large-size workpieces measurement technology has become a hot research topic.In this thesis,the research of machine vision measurement system is combined with large-size workpiece measurement demand,and the design and development of both image acquisition unit and software algorithm unit are completed.This research accomplishes the following tasks.First,this thesis builds an image acquisition system based on line scan camera.The working principle of the camera,lens,motion platform and other hardware systems is studied in detail,the appropriate hardware equipment is selected,and the characteristics and application scenarios of the line scan camera are analyzed.Based on this hardware system,an image acquisition system based on laboratory conditions is built.According to the imaging characteristics of the line scan camera,the external trigger mode is selected to make the speed of the motion match the line frequency,so as to avoid image stretching or compression during acquisition,thus laying the foundation for the following.Second,the image stitching algorithm is studied in this thesis.Aiming at the problem that the whole image cannot be acquired at one time during the measurement of large-sized workpieces,an image stitching method is introduced to obtain images with high resolution and complete information.This thesis introduces the image registration algorithm based on regions and feature points in detail,and the SURF algorithm is selected to extract features and form descriptors.Aiming at the characteristics of low iteration efficiency and low matching accuracy of RANSAC algorithm in the process of feature matching,a feature matching method based on MAGSAC algorithm is used,which effectively improves the stitching accuracy.Then,the evaluation criteria of image stitching are discussed to evaluate the quality of image stitching results.The experimental comparison shows that the registration method based on MAGSAC algorithm produces small distortion and high registration accuracy.Third,this thesis studies sub-pixel edge detection method.According to the actual application scenario of large-sized workpieces,guided filtering is used for image smoothing to maintain a better workpiece edge effect.In this thesis,the Zernike moment method is used to locate the sub-pixel edge of the image.Aiming at the possible dust,burr and other phenomena at the edge of the workpiece,the weighted least squares method is used to fit the line and circle to improve the fitting accuracy.The result is precise measurement of large-sized sheets.After camera calibration,the workpiece dimensions are measured and the data analysis proves that the image stitching-based algorithm can effectively meet the accuracy requirements of large-size workpieces.