Research Of Detection Methods For Cotton Foreign Fibers Based On Image Processing

China's cotton textile production is huge,which occupies a very important position in national economy.However,foreign fibers always mixed in cotton while picking,acquisition,storage,drying and other aspects.Although the content is relatively low,but it will affects the quality of cotton,which causes cotton will not easily be dyed.This has seriously affected the quality of the production of cotton,and causes very serious damage,which makes foreign fiber detection and identification particular important.Most Western countries use the automatic sorting machines to sort and pick cotton foreign fibers,but in China most of the domestic textile enterprises are still using manual sorting methods,not only time-consuming and inefficient.However,the price of some foreign machines and equipments is relatively high,and not very convenient to use and maintain.So it is very important to research and development our own relatively low prices,superior performance of foreign fiber identification and elimination systems which suit for our county.This paper describes the overall design detection system for foreign cotton fiber,analyzes the detection principle of foreign fiber,and introduces the system to select a light source,light direction,CCD cameras and optical lenses,etc.Collect cotton image online by building a f'oreign fiber detection device.To enhancement the image of foreign fiber cotton by histogram equalization and Retinex enhancement.Analyze the collected images and explore different segmentation methods of cotton foreign fiber.(1)In the cotton foreign fiber imagesegmentation method based on color,analysis the characteristics of cotton foreign fibers vary in different color space,proposed the idea of color perception to segment cotton foreign fiber image and improve the detection rate.(2)Research the method based on edge detection to segment cotton foreign fiber image,use different image,use different edge detection operator to segment the image of cotton foreign fiber,and analyze the results of processing.(3)In the cotton foreign fiber image segmentation method based on threshold,test the inter-fuzzy threshold method,iterative threshold and Otsu method,and its results were compared to analysis the segmentation effect(4)In the method based on clustering segmentation,experiment K-means clustering method and Mean-shift clustering method and compare the two methods.Finally,the finished segmentation images are being post-processing,which include filtering,morphological processing and heterosexual fiber positioning.The article used Visual C++ to development software and program the corresponding system,by analyzing the collected 400 pieces of cotton fiber image statistics,using the method of color perception to segment cotton foreign fiber image,the detection rate can reach 89%.