The Research On Button Defect Detection Algorithm Based On Machine Vision

With the continuous development of the textile industry,buttons occupy an important position as their accessories,and their quality directly affects the sales of clothing.At present,most companies still rely on manual detection of button defects.Due to the influence of the external environment and labor intensity,manual inspection has problems of low efficiency,low accuracy,and high costs.In the current process of continuous development of technology,the replacement of human vision with machine vision products will become a trend.This topic mainly adopts the way of machine vision and image processing,and proposes a button defect detection system.It provides important and reliable reference data for the quality rating of the button through the automatic detection system,which is of great significance to the enterprise.In this paper,the button defect detection system includes hardware platform,image processing algorithms and application software.Hardware platform from the image acquisition system,mechanical transmission and communication devices,and several other aspects were introduced.The flow of image processing algorithm is five parts: image acquisition,image preprocessing,template matching,edge detection and threshold segmentation.Among them,the image acquisition uses multi-threading technology to achieve asynchronous operation of image acquisition and image processing,improving the real-time detection.Image preprocessing part of the median filter selection of non-linear filter to achieve noise filtering.Template matching using edge-based template matching algorithm to improve the matching efficiency.Edge detection using Canny edge detection operator and sub-pixel algorithm to extract the edge,improve the edge detection accuracy.Threshold segmentation using Otsu algorithm to extract the surface defects.In addition,a deep convolution neural network method is adopted to classify and test the complicated buttons.The application software will be algorithm system integration,deployment to the windows operating system,the system closer to the product.Software features include user management,online testing,offline analysis,report statistics and more.Finally,the reliability of the algorithm and the system's indicators are verified through experiments.The test results show that the algorithm in this paper can effectively identify the edge of the button defect,clogged eyes and surface stains.The average detection speed of 102 ms,and able to better meet the online detection of the target.