A Fine Wool and Cashmere Identification System Incorporating Decision Tree Algorithms

Abstract

In practice, the identification of cashmere and wool fibers is still manual. With the expertise of experts, the shape of the fiber surface is observed and distinguished. This manual identification method is difficult because it requires at least years of training for qualified inspectors to observe a large number of fiber images, which is time-consuming and cannot avoid subjective human intervention. Although researchers have proposed many alternative identification methods, none of them can combine efficiency and cost. Therefore, it is particularly important to find a convenient and efficient identification method. The main purpose of this paper is to fuse the decision tree algorithm to study the identification system of fine wool and cashmere. This paper systematically introduces the method of cashmere and wool fiber identification based on image processing technology and feature extraction technology through fiber apparent morphological features. The process of fiber recognition is improved, and the features used for fiber recognition are obtained. The relevant theories and implementation methods are introduced in detail. Two key steps, image processing and feature extraction, are realized based on the theory. Finally, the effectiveness and feasibility of the method are verified through experimental tests, and good results are achieved.