Automated Identification of Cotton and Viscose Fibers Using Image Analysis and Random Forest–Decision Tree Classification

To enhance cotton/viscose fiber recognition accuracy, we propose a data classification-based digital analysis method. A set of image preprocessing algorithm procedures has been developed to perform grayscale, denoising, and binarization on the image. In addition, the fiber cross-sectional image is marked, and the outer contour image of the fiber cross-section is obtained by using the edge detection algorithm, and finally the feature parameters are extracted. By training and analyzing the extracted fiber cross-section characteristic parameters, three recognition algorithms including K-nearest neighbor, backpropagation neural network, and random forest–decision tree are used to identify the fiber category. The experimental results show that the recognition rate of the random forest–decision tree algorithm is the highest, and the recognition accuracy is up to 97.5 %.