Design And Realization Of Workpiece Size Measurement System Based On Machine Vision

Machine vision is a frontier of high-speed development,the core is the use of machines instead of the human eye,the specific image information for complex processing to get all kinds of information process.With the rapid transformation of contemporary manufacturing,the traditional industrial size measurement methods have been phased out.Based on the SIFT feature of machine vision and image,this paper realizes the non-contact measurement of the workpiece size,and improves and realizes the measurement system.Before the image is acquired,the measurement system is first corrected to reduce the accuracy error caused by the mechanical device and so on to the workpiece image and improve the measurement accuracy.As the system uses a CCD camera,in order to make the camera photographic plane and the workpiece to be measured as much as possible to maintain the plane parallel to the system using the checkerboard grid calibration plate as a calibration tool.After the image acquisition,due to the workpiece and the camera space three-dimensional position relationship,will make the image to a certain degree of distortion,the system uses the standard checkerboard to calculate the affine transformation model parameters,and then get the image to do the correction processing.The system can be a small size of the workpiece can be done to complete the image information collection,but because the CCD camera range is limited,large-size workpiece can not do a line of image information acquisition,so to get the block image to get Complete artifact image information.According to the image stitching of the scale movement,the pixel position of the image stitching is determined by the coordinate of the image corresponding to each image,the display area of the new picture is taken into the whole picture,and finally the average weighting algorithm is used in the vicinity of the image splicing Image fusion to eliminate visual errors,to achieve efficient,fast image stitching effect.Based on the SIFT feature extraction and the best suture image mosaic technology,in the realization of the image than the grating image is slightly complicated,but improve the image mosaic accuracy.Through the SIFT algorithm to describe the key points of the workpiece image,the nearest neighbor method is used to complete the matching of the key points in the image,and then the RANSAC calculation is used for the second matching and the projection transformation model parameters are calculated.Then,the Max-Min-Cut is calculated to obtain the best suture and the image is stitched along the best suture.Finally,in order to eliminate the visual sense of the suture,the gray scale of the images on both sides is uniformly transitioned by the progressive fade method Image stitching.The proposed method can effectively eliminate the ghosting problem caused by image fusion and improve the accuracy of workpiece size measurement to a certain extent,while the key points of SIFT algorithm are accurate and stable.The shape detection and calculation of the workpiece is also limited by the resolution of the CCD camera.Based on the improvement of the physical resolution,sub-pixel subdivision technique is used to improve the accuracy of the measurement results.Sub-pixel positioning,including geometric methods,moment estimation method,interpolation method and least squares estimation method,each method has a scope and a certain limit.The moment estimation method is the most widely used for noise-insensitive characteristics.The system uses the moment estimation sub-pixel subdivision to identify and measure the shape of the workpiece.