Research On Identification And Classification Of Silkworm Cocoons Based On Shape Features

The screening of raw cocoons is an important process in silkworm cocoon production.At present,this process still relies on manual operation,and there are problems such as high labor intensity,low cocoon selection efficiency,and difficulty in standard control,which limit the production efficiency and product quality of cocoon silk production.Aiming at four types of silkworm cocoons with shape defects: double cocoons,malformed cocoons,perforated cocoons and crushed cocoons.Based on shape characteristics,this paper uses image processing and point cloud processing technology to realize the identification and classification of four types of silkworm cocoons.The specific research content and work are as follows:(1)A method for identifying and classifying double cocoons and malformed cocoons based on two-dimensional shape features is proposed.First,obtain the outer contour convex hull of cocoons by multi-step processing of the image.And then,calculate the minimum circumscribed rectangle and fitting ellipse of cocoons,respectively.Finally,calculate the slender length,the ellipse fitting degree and the division threshold for comparison to realize the recognition and classification.The experimental results show that the average recognition accuracy of the method in this paper is 97.1% for double cocoons,93.5% for waste malformed cocoons,and 89.5% for defective malformed cocoons.(2)A method for normalized rotation registration of point clouds of perforated cocoons and crushed cocoons is proposed.First,calculate the center coordinates and coordinate axes of the disk according to the rotation relationship and coordinate mapping relationship between any two sets of calibration model point clouds.And then,building the model of the mathematical model of the rotating disk.Finally,achieve registration.The experimental results show that when the average registration time of the method in this paper is 6.548 s,the average root mean square error is 0.557502,and the average distance is 0.433.On the premise of ensuring high registration accuracy,it can effectively improve point clouds of perforated cocoons and crushed cocoons registration efficiency.(3)A method for identifying and classifying perforated cocoons based on three-dimensional shape features is proposed.First,perform a series of prepossessing on point clouds of perforated cocoons,such as noise reduction,streamline,normal vector calculation.And then,identify the hole boundary points of point clouds of perforated cocoons.Finally,achieve registration and classification by calculating the hole centroid.The experimental results show that the method in this paper has an average recognition accuracy rate of 99.02% for perforated cocoons.(4)A method for identifying and classifying crushed cocoons based on three-dimensional shape features is proposed.First,sample and slice point clouds of crushed cocoons.And then,calculate the coefficient of variation of the average curvature of the sliced point cloud and compare it with the division threshold.Finally,achieve the identification and classification of crushed cocoons.The experimental results show that the average recognition accuracy rate of the method in this paper is 94.55% for waste crushed cocoons,and 91.9% for defective crushed cocoons.