Study On Foreign Fiber Detecting System Based On Multispectral Technique

During the development of the cotton spinning industry, there is always a problem that how to detect foreign fibers in cotton accurately and efficiently. At present, many domestic and international enterprises have developed a variety of products which are related to the foreign fiber detection. However, due to a wide range of foreign fibers and different optical properties, these machines can only detect some certain types of foreign fibers in cotton. Detecting species is not comprehensive so it’s prone to miss, which is resulting in the unsatisfactory of the recognition rate. To solve this problem, this paper proposes a foreign fiber detection system which is based on the multi-spectral technology, which can detect a variety of foreign fibers simultaneously and highly develop the detection level, with a strong practical significance. Details are as follows:Firstly, we propose to use the multi-spectral technology into the foreign fiber detection system. By increasing the light band, we can get more foreign fiber multiband response characteristics. Then combing with the spectral characteristics of foreign fibers, we can get the best illumination wavelength of the multi-spectral detection system; through the experimental analysis of the factor which affecting the foreign fiber reflection imaging, we can get the optimum imaging conditions. Secondly, by using the multi-spectral detection principle, we design the structure of the multi-spectral foreign fiber detection system and set up a working platform; we propose the contrast pyramid image fusion method for images that taken from different CCDs; we do research on the image processing algorithms and compare the processing results by using the image quality evaluation function; then we select the edge-dimensional detection algorithm and achieve good recognition results. Finally, we write a software interface of multi-spectral foreign fiber detection system.The results show that: the missing foreign fibers in the single light source detection can get good complement detections in the multi-spectral detection. The types of detected foreign fiber are more and the recognition effect is better than the one in the single light source detection.