Foreign Fibers Detection Using Improved Otsu-based Maximum Entropy Algorithm in Spinning Process

Aiming at the problem that the mixed anisotropic fibers in cotton seriously damage the quality grade of cotton textiles, a foreign fibers detection method using improved Otsu-based maximum entropy algorithm in spinning process is proposed. Firstly, the original image is transformed into grayscale image and preprocessed with median filtering algorithm, Then the extraction effects in the grayscale co-generation matrix are compared, which contrast is selected as the feature parameter for texture extraction. Finally, the feature image is fed into the maximum entropy improved Otsu algorithm to detect the foreign fibers, which achieves the fine foreign fibers detection. The model is applied to detect a variety of foreign fibers and fine foreign fiber samples, result shows that the proposed method is effectively improve the recognition effect of foreign fibers.