Research On Detection Algorithm Of Foreign Fibers In Lint

Materials like chemical fibers, animal hair, plastic films and so on are not cotton fibers, materials like dyeing thread, dyeing rope, dyeing cloth and so on are cotton fibers, but the color of them are different from cotton fibers, and the above two kinds of materials all belong to foreign fibers in lint. In cotton ginning factory, content of foreign fibers in lint will seriously affect the quality and grade of cotton, and thus affect the trading price of cotton and the economic benefits of cotton ginning factory. In textile factory, foreign fibers are apt to cause spun yarn broken, working efficiency reduced, and defects on dyeing cloth, and seriously influence product appearance and sales, etc. Thus it can be seen that foreign fibers in lint have a great influence on the quality of cotton and the economic benefits of cotton ginning and textile factory, though the content is few. To solving the problem of foreign fibers, a lot of cotton ginning and textile factories use workers to pick foreign fibers, but this method has a series of shortcomings, such as high labor intensity, low working efficiency, unstable product quality and qualified rate, etc, and the labor costs are also growing, so picking foreign fibers by workers can’t solve the problem well.Aiming at the above problems, this paper did a deep research and analysis of foreign fibers detection system, image segmentation algorithm, and Gray Level Co-occurrence Matrix(GLCM), and on this basis put forward the improved two-dimensional Otsu algorithm and two-dimensional Otsu algorithm based on GLCM, thus realized the foreign fibers detection and increased the real-time and accuracy of foreign fibers identification technology respectively. Specifically, this paper primarily did the work at the following aspects:1. Foreign fibers detection system, mainly including image acquisition and image processing, was designed, the image acquisition was to collect lint images containing foreign fibers and the image processing was to analyse and segment the collected images to identify foreign fibers in lint.2. The image acquisition system was built. Firstly, the overall structure of detection system was design, then the models of required hardwares, such as CCD cameras, lens and light source, were selected and the layout of them was determined according to actual demand, finally the image acquisition system was completed successfully.3. The improved two-dimensional Otsu algorithm was put forward. Gray distribution characteristics of lint and foreign fibers were analysed with gray histogram, on the basis of it, threshold ergodic range of two-dimensional Otsu algorithm was reduced to about1/5, and thus segmentation time was shortened to about1/5of the original two-dimensional Otsu algorithm, effectively improving the real-time.4. Two-dimensional Otsu algorithm based on GLCM was put forward. The color of lint and white foreign fibers was too similar, so the identification accuracy of white foreign fibers in lint was quite low. However, the texture of different material was different, so texture images were segmented by two-dimensional Otsu algorithm, and an image denosing method which combined corrosion and median filtering was used to the segmented images, and finally a very good segmentation effect was obtained.5. The upper computer programming was completed with C#, interfaces and main functions were complised with C#, and image processing functions were realized with hybrid programming of C#and Matlab.Results of experiments show that the improved two-dimensional Otsu algorithm can effectively reduce segmentation time and improve real-time on the basis of guaranteeing segmentation accuracy, and the two-dimensional Otsu algorithm based on GLCM can accurately identify white foreign fibers in lint and significantly improve segmentation accuracy.