Texture Classification Using High-Order Local Derivative Pattern and KNN Classifier

Texture classification is a fundamental issue in image processing and computer vision. It has been used in material classification, surface analysis, document analysis, and industrial automation. In this paper, a texture classification algorithm based on Local Derivative Pattern (LDP) is proposed. The algorithm extracts high-order directional texture features from grayscale images and represents them using normalized histograms. A K-Nearest Neighbor (KNN) classifier with cosine distance is employed to classify texture images into multiple categories. Simulation experiments on a practical texture image database demonstrate that the proposed algorithm can achieve accurate classification results with low computational complexity.