Detections Of Foreign Fibers In Cotton Based On Machine Vision And Image Analysis

Foreign fibers in cotton have troubled textile industry for a long time. Traditionally, the enterprises pick out the foreign fibers by manual sorting. The workers are working in a poor condition and the labor intensity is big. Besides, it is inefficient and difficult to pick out the foreign fibers. So, foreign fibers detection and elimination device comes to the market. However, those machine equipments overseas are too expensive and difficult to repair. There are some limitations in equipments in china. In this paper, the foreign fibers detection device is designed and built. The images of the raw cotton are acquired during the production. Foreign fibers are recognized based on machine vision and image analysis. Image processing and analysis algorithms that suit to production are researched.(1) The foreign fibers detection device is designed and the images are acquired. The images of the raw cotton are acquired during the production by the usage of a light-emitting diode (LED) that use direct current and low-voltage. The directional properties of the illumination are directed bright-field front light illumination.(2) On the basis of through analysis to the foreign fibers image and the requirements of real-time online, this paper put forward to several image processing methods of removal of several noises and image enhancement, such as image smoothing, gray value transformations, homomorphic filtering, adaptive adjustment algorithm that separates the foreground from the background with nonuniform illumination. The median filter can be used for noise suppression in a short time by analysis. Density histogram adjustment can be used to increase the contrast locally that caused by illumination. Adaptive adjustment algorithm can increase the contrast.(3) In order to get the foreign fibers in the raw cotton image, several image segmentation algorithms are discussed, such as thresholding and edge extraction. The dynamic thresholding can separates foreign fibers very well by analysis.(4) Smallest enclosing circle algorithm is developed for foreign fibers positioning. Spatial position of the foreign fibers is fixed according to the Centroid coordinates of it in the image and the dropping speed of cotton. The jet arrays of real-time separator are also discussed in this paper.(5) The foreign fibers detection system with image pre-processing, image segmentation and feature extraction is designed based on object-oriented programming design thought. The method applied in the system still has its limits which has a long distance to the utility because of the manifold foreign fibers, although this paper has acquired some achievements. However, this paper lays the foundation for next work.