Design And Research Of Meat Sorting System Based On Machine Vision

With the rapid development of the meat industry in recent years,the manual sieving process in the traditional processing of dried pork has become a bottleneck restricting the rapid development of the dried pork industry,resulting in the fact that the production efficiency of the dried pork industry can not meet market demand.At present,the pork jerky industry plans to use the sorting of meat pieces according to the grain direction instead of manual sieving.And sorting can usually be achieved manually or by teaching robots.However,the temperature of meat pieces is low and its texture is difficult to identify.Staff sorting meat pieces for a long time will cause frostbite on the hands and deterioration of eyesight,and the production efficiency of this method is low.And the traditional teaching sorting robot cannot meet the sorting needs because it cannot recognize the texture of the meat.Therefore,after fully investigating domestic and foreign sorting systems and analyzing the specific needs of meat sorting,combining machine vision and Cartesian coordinate robot,a meat sorting system is designed,and carries out the following work around it.(1)The overall design for the sorting system.Firstly,according to the needs and problems of meat piece sorting,the overall scheme is completed,then the image acquisition unit and sorting unit are built according to the characteristics of the meat piece,finally the software system is designed.The processing flow of the image algorithm is presented.(2)Calibration for vision systems.Firstly,a visual recognition model is established;then Matlab and Halcon were used to calibrate the camera to solve the model parameters,and the results show that the model parameters solved by Halcon had higher accuracy;finally,the robot and camera configuration model is studied,the robot and the camera are calibrated by using the Eye-to-hand model.The camera is calibrated to obtain the calibration parameters of the robot and the camera.Combining the two calibration results lays the foundation for the subsequent positioning of meat pieces.(3)Enhancement and segmentation of meat images.Firstly,because a single image enhancement algorithm cannot meet the actual processing requirements,a hybrid image enhancement algorithm is proposed.Experiments show that the hybrid image enhancement algorithm can effectively remove the noise in the meat image and make the meat texture more prominent.Then,aiming at the problems of poor noise resistance and low efficiency in the traditional Otsu algorithm,GA-OTSU is proposed.Experiments show that GA-OTSU segmentation of meat images is better.Finally,the morphological processing is used to eliminate the small holes and burrs caused by image segmentation.(4)Design of feature extraction and control strategy for meat chunks.In the aspect of feature extraction,firstly,the four common edge detection operator is used to extract the contour of the meat piece in the image.The Experiments show that the contour of the meat piece extracted by the Canny operator is the most complete and clear.Then,in view of the problem that the template matching algorithm cannot extract the information of the meat pieces due to the different shapes of the meat pieces,a positioning principle based on the minimum circumscribed rectangle is proposed,and the meat pieces are approximated as the minimum circumscribed rectangle.The characteristic information of the meat piece is obtained by obtaining the characteristic information of the rectangle.In addition,in view of the particularity of the texture dispersion of meat chunks,a texture direction extraction algorithm based on Radon transform is proposed,which extracts the texture direction of the area with the largest texture proportion in the meat chunks as the texture direction of the meat chunks.In terms of sorting control strategy,firstly,a dynamic tracking control method based on PID is proposed to realize the synchronous tracking of the robot and the conveyor belt;then a real-time positioning method of meat pieces on the conveyor belt is designed,and for the problem of repeated meat pieces in the front and back images,a deduplication method is proposed.Finally,to eliminate vibration and shock problems caused by high speed,a trajectory planning method based on S-shaped velocity curve is proposed.(5)Experimental validation for sorting systems.Firstly,a complete set of sorting system is built,and then sorting experiments are carried out using different sizes of molds.It was verified that the sorting effect of the system was good,but the final product still has the problem of small holes due to the irregular shape of the meat pieces.Therefore,a minced meat spray device is added to further improve the sorting system,and finally the quality of the finished product is compared with that of the hand-made product.Experiments show that the finished product rate reaches 96% after improvement,and the scores of each index exceed 9.0to meet the industry standard,and the production efficiency is greatly improved.It has been verified that the quality of the finished product of the sorting system and the entire set of processes meet the industry requirements.