Abstract:

Aiming to improve the efficiency of detection the foreign fibers, a new approach of detecting foreign fibers in cleaned seed cotton before the ginning stage was proposed. In the experiment, cleaned seed cotton, in which organic impurities such as boll shells, stems and leaves were removed, and 21 kinds of common white or color foreign fibers were used as detection samples. Images were acquired under the double illmination of white LED and red line-laser. Then, an improved Sobel edge edtection algorithm was used in the Red channel of RGB color space and the Saturation channel in HSI color space separately. And a one-dimension maximum entropy thresholding method was also implemented in the Saturation channel for increasing the successful detecting rate. Expeiment results indicate that the double illumination imaging and the image processing algorithm reduce interference such as shadows. The successful detecting rates of white and color foreign fibers are up to 74.7% and 70.8%, respectively. This paper provides a reference for detecting forgign fibers in seed cotton.

Key words: seed cotton, foreign fiber, double illumination imaging, color space, image processing