Research On Carding Tooth Detection For CNC Tooth Machine Based On Machine Vision

In Textile machinery, the quality of carding tooth on tooth cylinder for combing machinecan directly influence the effect of comber, thereby affects the quality of final products. Thetraditional detection technique requires a large number of workers, not only time consuming,but also it is difficult to guarantee the consistency of product quality．With the character ofnon-contact, real-time, fast speed, strong scene of anti-jamming ability, the Detectiontechnology of machine vision can replace the manual detection to eliminate human error.Applying the technology of machine vision, image processing and image analysis technology,under the existing experimental conditions, this paper adopts carding tooth as the researchobject, research and analysis of its machine vision detection technology, and finally establisha carding tooth detection system based on machine vision.1.First this paper expounds the research background and significance and topic source;Then the machine vision detection technology is summarized, and introduces its applicationin industrial detection research status at home and abroad and its trend of development;Finally, illustrates the main research tasks and content of this article.2.The overall summary of the comb tooth plate detection system; Based on the actualsituation and needs, the carding tooth detection system has been designed by the appropriatecamera, lens, the light source and other hardware instruments.3.The pretreatment technology (such as histogram equalization, image smoothing, imagesharpening etc.) and classical edge detection technology are introduced, through contrastanalysis, choose the appropriate method to meet the need of carding tooth image detection; Byusing the method of Gauss curve fitting and interpolation in the gradient direction so that thepositioning precision of image edge to sub-pixel level, and also lay the foundation of thefollowing image processing.4.Finally, on the basis of analyzing the traditional Hough transform, the linear leastsquares fitting method was adopted to realize on line detection; Through experiment contrast，an improved method was proposed to realize the circular test and the detection results byusing least squares method based on the moment processing, the experimental results showthat the improved method is fast, high detection accuracy, makes the detection performance ofthe system is greatly improved; Using Moravec corner detection algorithm for cornerdetection, laid the foundation for further image recognition and advanced processing.5.To verify the validity of the research results, based on VC++6.0design the cardingtooth detection system software development platform, realize the whole process of combteeth to detect.