Research On Fish Tail Removal Device Based On Machine Vision

With the continuous improvement of computer computing power and image processing technology,in developed countries,intelligent agricultural product processing equipment based on machine vision has achieved significant results.However,in my country,the application of machine vision technology to fish processing equipment is still in a blind spot.Domestic fish product processing equipment is basically based on traditional mechanization.This processing method has the characteristics of extensive form,outdated equipment,single function,low automation integration,low production efficiency,etc,which greatly affects the processing quality and economic benefits of fish products.Therefore,there is an urgent need to transform fish processing equipment from traditional mechanization to intelligentization in industrial production.At present,the main direction of equipment intelligence is how to rely on machine vision technology to guide the machine to perform intelligent human-like operations.In order to improve the processing quality and economic benefits of fish products,its processing equipment needs to be embedded with more efficient intelligent technology.The main work of the full text is as follows:(1)Collect a certain amount of fish body images,test and analyze the collected fish body images,and design a preprocessing process that conforms to the fish body images in this article.Through experimental research,it is found that when the fish species and environment change slightly,the traditional machine vision algorithm has a very low recognition rate for fish tails.Therefore,the Mask R-CNN neural network was used to build the machine vision model in this paper,and at the same time uses the COCO data set to pretrain the model,and focuses on analyzing the basic principles and processes of fishtail recognition.According to the characteristics of model checking,this paper introduces Affine Transformation to further improve the algorithm.(2)Based on the processing object of the device,the design of the control system of the fish tail removal device was completed.The system includes hardware selection,main program module,communication module,stepper motor control module,coordinate transformation module and other program design and writing.At the same time,it completes the path planning of the fish pushing mechanism and the design of the host computer LabVIEW real-time monitoring interface.And display the change range of the shear force required to remove the fish tail in the physical experiment on the real-time monitoring interface,which provides a basis for prototype production.(3)After the design,processing and statics inspection of the key components,the prototype of the entire fishtail removal device was completed.On the platform of the selfmade prototype,this article carried out the intelligent test experiment of fish body image acquisition,preprocessing,fish tail recognition,positioning,fish pushing,and fish tail removal.The experimental results are available.The prototype processing effect is very good.Within a certain range,the processing of fish tails for fish bodies of different species and sizes has met the expected work requirements and verified the feasibility of the machine vision fish tail removal device designed in this project.