Study Of Online Identification And Elimination System For Incomplete Shrimp Based On Machine Vision Technology

The shrimp industry has been playing an important role in the field of aquatic products in China.Shrimp products are popular with consumers because of their delicious taste and rich nutrition.With the continuous improvement of people's living standards,the demand for shrimp and prawn at home and abroad has been daily on the increase and the quality requirement of shrimp products has been also boosting.In the process of fishing,transportation,mechanical grading and high-temperature and high-pressure steaming,some shrimps will be damaged to be the defective shrimps,which affects the appearance and reduce the corresponding quality of products.Thus,this kind of impurities should be removed during the processing state.The traditional shrimp integrity detection is done mainly through the artificial sensory evaluation.However,the artificial detection has some shortcomings,such as strong subjectivity,high labor costs,fatigue etc.Therefore,multi-fields technical means,including mechanical design,machine vision and electronic control,are used in this thesis.Taking fresh shrimps and cooked shrimps as the research object,the thesis explores feasibility of applying machine vision technology to real-time on-line detection rejection of defects shrimps.The main contents and conclusions of this paper are as follows:1.In order to solve the problem of single-grain output of shrimps,a kind of shrimps tiled single-particle device is designed in the thesis to achieve the state of single-grain output,so that the situation of shrimps sticking can be avoided as much as possible,which achieves 95.1%effect of single-grain.2.In order to solve the problem of on-line integrity detection of shrimps,this study designed and implemented the SF of shrimps,based on the included angle contour analysis,and the FS of shrimps,based on the shrimp template matching method,so that to realize the integrity discrimination of different specifications.1128 pictures of fresh shrimps were used in the experiment,which includes 749 whole fresh shrimps images and 379 incomplete fresh shrimps images.There were 1123 pictures of cooked shrimps using in this study,including 769 whole cooked shrimps images and 354 incomplete cooked shrimps images.By respectively testing the proposed algorithms,the results showed that the template matching method can achieve the overall recognition accuracy of 99.6%compared to the method of included angle profile analysis of 86%.What's more,the average processing time of each image by the latter method is 15.1 ms faster than the former.For the overall recognition accuracy of 86.4%of cooked shrimps,compared with the method of included angle contour analysis,the template matching method can achieve the overall recognition accuracy of 99.4%.At the same time,the average processing time of the image by the latter method is 14.8ms faster than the former.It can be concluded that the template matching method is better both in recognition accuracy and image processing speed.3.In order to solve the problem of how to remove the detective shrimps,an implementing agency for removing shrimps with multiple defects is designed in this study.By adopting the method of gas blowing,the removal process will be achieved during the drop of shrimps.In this study,100 shrimps were used to remove only without identification to optimize the executive agencies.It showed that the critical parameter C is 158ms under the stable running state of the equipment.At that time,the baffle distance was 70mm from the surface of conveyor and the best culling rate is 90.3%at the end section of 30mm.4?On the basis of the above research,a set of automatic sorting equipment for on-line identification and rejection of incomplete shrimps,which is based on machine vision technology,was designed and built.The whole machine detection was conducted on fresh shrimps and cooked shrimps respectively.The results show that when the speed of loader is six,the vibrating frequency of tiled single granulation unit is 33Hz,the angle of vibrating screen is 10 °,and the speed of the visual detection unit is 0.667m/s,the rate of the camera frame is 60fps,the N value of the soft triggering is 8 and the air compressor pressure is 0.4-0.7MPa,the whole rate of machine recognition can reach more than 80%,and the speed of system processing surpasses the sorting speed of skilled workers.