Research On The Method Of Obtaining Complete Image Information Of Cocoon Shell

In order to ensure the production quality of silk reeling process,before the cocoon is used for silk reeling,it is necessary to select the cocoon and filter out the cocoon that cannot be used for silk reeling or cannot be reeled into normal silk.As the reeling process requires high accuracy in cocoon selection,the improvement of the accuracy of existing cocoon selection research is limited by the integrity of cocoon shell image acquisition.Therefore,based on the need for obtaining the complete image of cocoon shell,this paper researches the method of acquisition the complete image of cocoon shell.Firstly,according to the requirements of cocoon shell image acquisition and the shape characteristics of cocoon,this paper designs a cocoon shell image acquisition device,whose core mechanism is cocoon turning mechanism.The cocoon turning mechanism mainly includes: cocoon turning roller,plane mirror and cocoon light source.Two synchronously rotating cocoon turning rollers enable the cocoon to turn between the rollers to obtain the surface image of the cocoon shell side.The plane mirror which installed on both ends of the roller is used for reflecting the surface image of two ends of the cocoon.The cocoon light source provides the light source for the inside of the cocoon,and assists the camera to collect the internal light image of the cocoon.According to the work flow of the image acquisition device,the control system of the device is designed.It mainly includes hardware circuit design and control program development.In addition,the human-computer interface is developed by using MFC to collect and save the cocoon image.Secondly,pre-processing the image of cocoon side surface which is collected by the device.The background area on the image is removed and the edge of cocoon is extracted.According to the ellipse fitting of the contour of the cocoon,the mathematical model of the cocoon is established.The mathematical model of cocoon is transformed into the equivalent step cylinder model,and the mapping relationship of pixel coordinates between the equivalent step cylinder model surface image and the unfolded image is solved,so as to realize the equivalently unfolding of the cocoon shell surface image,which provides the environmental basis for the research of cocoon shell surface image stitching.Finally,the color space of the unfolded image is transformed from RGB to HSI,and each channel of the HSI image is separated to assist in extracting the defect image in the unfolded image.The SIFT feature detection algorithm is used to detect and match the feature points of two defect extraction images with overlapping areas,and the coordinate distance filtering method and RANSAC algorithm are used to remove the mismatches of the two images.According to the coordinate correspondence of the matching points,the mosaic and fusion of the defect image are realized,and the complete image of the defect on the surface of cocoon shell is obtained.The research of this paper realizes the acquisition of complete image information of cocoon shell,as well as the mosaic of cocoon shell surface defect image.The research makes up for the shortcomings of the cocoon selection research,and provides a good technical support for improving the accuracy and reliability of cocoon defect feature recognition.