Application Of Image Processing Techniques On The Detection Of Foreign Fibers In Cotton

The amounts of cotton textile possess a large appropriation in our national economy, which plays an important role in our country. However, foreign fiber has troubled textile industry for a long time, which has a serious influence on cotton production, as well as the quality of the cotton product, resulting in great economic losses. Traditionally, the textile enterprises pick out the foreign fiber by manual sorting. But workers are working in a poor condition with big labor intensity. Besides, this way is inefficient and difficult to ensure to pick out all impurity. Therefore, the research and exploitation on detection and elimination system of foreign fiber has profound significance and influence, which is suitable for the textile company requirement and has a relative lower price and the superior performance.In our seminar past research, the visible light is used as the detecting illuminant; just one foreign fiber is captured in the raw cotton picture. When there is a big color difference between foreign fiber and the raw cotton, the foreign fiber can be discerned easily. However, it is very hard to be detected with small color difference, such as polypropylene. So, the polypropylene fiber is needed to be detected for the next step. In this thesis, image processing techniques are used to the detection and elimination system of foreign fibers.(1) The detection system is introduced, including LED illuminant,linear CCD camera,the direct positive lighting way,the industrial board and electromagnetic valve. The polypropylene image is scanned continually using this system. So, this step prepares for the image processing and foreign position and elimination.(2)The image pretreatment is discussed. The image equalization and transformation of gray scale are operated to enhance the contrast foreign fiber and cotton, grayscale-log-conversion method is chosen; the median-filter-method is used to remove the image noise. So, this step is proposed the image segmentation and position elimination.(3) Under the ultraviolet illuminant, the segmentation method based on normal distribution can detect the five foreign fibers including those which are in the raw cotton involved in this thesis. When the ultraviolet illuminant is used, equivalent color fibers can be detected better. Even the polypropylene yarn and film in the cotton completely can also be detected better. The pine must be opened fully before working, which helps to lighten the pressure of testing.(4) The longer the length of fiber, the more surface, the better the segmentation effect. When we select and configuration light source,CCD camera and other hardware and determine the image processing algorithm, the litter size foreign fiber can be chosen to be detected. As long as the fiber is in the CCD camera range, it can be detected, even if it is not in the center of image. When there are two or three foreign fibers in the cotton image, the same hardware configuration and algorithm can be used. But the ultraviolet illuminant must be added, so that the equivalent color fiber, polypropylene yarn and film can be detected easily.