Research On Visual Inspection Technology And Application Of Large Area Leather Contour

The collection and detection of leather images is the basic application in the leather processing industry,and the acquisition of high-precision images of large-area leather is one of the difficulties.The traditional linear CCD camera has high requirements for the servo control accuracy of the conveyor belt,but area scan CCD cameras do not have this requirement.Even if the transmission rate is uneven,the large-area leather image can still be effectively obtained by the stitching technology,which can effectively reduce the system cost.The work of this paper is to solve the related problems faced by large-area leather image processing based on area scan CCD cameras,and it has a good reference value for reducing the automation cost of leather processing.This thesis,which is based on semi-processed animal leather,focuses on the technology and application of visual inspection of large-area leather contours.This paper optimizes and improves the design of stitching schemes and leather image processing algorithms,plans a path for the cutting of leather fur edges,and finally implements a visual inspection system for leather contours.The main work of this paper is as follows:Firstly,using the camera to capture leather images in segments,and performing image preprocessing such as filtering.After analyzing and comparing the two algorithm principles of Gaussian filtering and bilateral filtering,a bilateral filtering algorithm is selected to reduce the noise of the leather image,according to the actual processing effect.Secondly,based on the collected leather images,this paper analyzes the theory of commonly used image stitching algorithm.After considering the characteristics of the leather image and the actual stitching test,a corner extraction algorithm was selected for stitching leather images.In addition,based on the improved Harris algorithm,subpixel technology is applied to corner detection to provide more accurate matching feature points for image stitching.Then,after completing the morphological boundary extraction,using the chain code algorithm to sort the boundary coordinates,and proposing a new method of leather contour vectorization.On the basis of the performed work,the trajectory planning of the leather edge cutting was completed,and the area calculation was performed.Finally,this paper designs and implements a visual inspection system for largearea leather contours.The system environment construction,program migration,software algorithm implementation,and interface design have been completed successively,and the detection effect is displayed through a friendly human-computer interaction interface.