Design And Implementation Of Scallop Grading System Based On Computer Vision

China is a large marine country with abundant resources.The aquaculture industry accounts for a large proportion of the gross national product.As an important part of aquaculture,scallop culture has attracted more and more attention.During the growth of the scallops,it is necessary to distinguish the grades several times,and then perform the subsequent process.At present,in scallop culture,the main grading method is manual measurement and grading.This method is cumbersome and has high volatility,which greatly affects the efficiency of scallop grading and the development of scallop aquaculture.By replacing the traditional manual grading with computer vision,it can avoid direct contact with scallops,effectively reduce scallop collision and improve scallop culture efficiency.This article is based on the process of the project,the content is more detailed,the main contents are as follows:(1)Scallop feature selection and image preprocessing design.Aiming at the problem that the scallop image information is not obvious and the shell structure is complex,the influence of scallop color,pattern and edge contour information on image processing is analyzed,and the image feature selection is carried out to determine that the color and contour are the main features of scallop image.A single channel RGB color processing method provided by OpenCV visual library is proposed.The scallop image is preprocessed by means of Gauss and median filtering.(2)Design of scallop image segmentation and measurement algorithm.The adaptive threshold binarization and edge detection method of Otsu method are used for scallop image after pretreatment.The scallop image is extracted by color,the scallop pattern is filtered effectively,the background and the front are separated,and the scallop contour information is obtained.Several existing scallop measurement algorithms are analyzed.Because of the irregularity of scallop contour,the calculation amount of dimension measurement is large.The method of minimum circumferential circle and equal area circle diameter mean of contour is proposed to replace scallop size measurement.By drawing the minimum circumscribed circle of the image and obtaining the circle of equal area with the scallop contour,the average diameter information is approximately transformed into scallop size information for classification,which greatly reduces the amount of calculation,controls the error rate and improves the efficiency.(3)System implementation and testing.This paper describes the practical possibility of the system implementation by using code,expounds the essence of the system implementation method,and introduces in detail the process of RGB color single channel extraction,filtering,binarization,edge detection and obtaining circular diameter information.By comparing the real data,the error rate of the experimental results is obtained,and the scallop grading item’s integrability is verified.This project researches and designs a scallop grading system based on computer vision.Compared with traditional manual measurement,it has the characteristics of avoiding contact and high efficiency,and has strong versatility.