Research Of Three-dimensional Mesh Detection Based On Machine Vision And System Implementation

Visual detection technology is emerging as a relatively automatic detectiontechnology and has been widely accepted and used. This article is based on “defectdetection system applying on appearance of coating”, one project cooperated with Iwatanicorporation. This article carries out a detailed study on testing quality of a certain type ofthree-dimensional mesh and completed the entire platform design and systemimplementation.Firstly, paper introduces the source of subject, the research background andsignificance, and the research status. It also describes the main content and organizationalstructure of this article.After realize testing requirements and technical specifications of the detectionplatform. We describe the characteristics of the detected object, and then determine thedetection scheme, including make the detection principle, the choice of the lightingprogram, image acquisition card selection and CCD camera selection. We compared thepros and cons of all possible options.The third chapter studies image preprocessing of three-dimensional mesh, includingimage threshold segmentation, edge detection and skew correction. And then bycomparing the experiment to determine the most suitable method for this project.In chapter4, we have studied the method for judging surface quality and propose twomethod for detecting surface black spots and size deformation. At last, we did somecomparison test to proved that the proposed method is effective and feasible.The fifth chapter research shape recovery algorithm of the three-dimensional mesh.Attempts to recover3D reconstruction by using two pieces of defocus image.Finally, the article introduces the hardware and software of detection platform, andshows some photos that were taken when it is working.