Algorithms for Image Texture Classification

• Published in: Image Texture Analysis pp. 51 - 102
• Main Authors: Lan, Yihua, Song, Enmin, Hung, Chih-Cheng
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2019; Springer International Publishing
• ISBN: 9783030137724; 3030137724
• DOI: 10.1007/978-3-030-13773-1_3

Image texture classification utilizes either unsupervised or supervised algorithms as a classifier based on textural features extracted from images. Many of these algorithms are from early research in the statistical pattern recognition. It has been proved that if the priori probabilities information are available about datasets, Bayes decision theory gives the optimal error rates in classification. However, this priori information may not be available in many applications. Hence, many classification algorithms using other measures such as similarity on the dataset are developed for categorization. There are also many variations of these algorithms which have been developed and used in image texture classification. This chapter will explain the basic concept of the following algorithms: (1) K-means, (2) K-Nearest-Neighbor (K-NN), (3) fuzzy C-means (FCM), (4) fuzzy K-nearest-Neighbor (Fuzzy K-NN), (5) fuzzy weighted C-means (FWCM), (6) new weighted fuzzy C-means (NW-FCM), (7) possibility clustering algorithm (PCA), (8) generalized possibility clustering algorithm (GPCA), (9) credibility clustering algorithm (CCA), and (10) support vector machine (SVM). Some algorithms utilizing optimization techniques are also introduced. This includes the ant-based K-means algorithm, the K-means algorithm using genetic algorithms, the K-means algorithm using simulated annealing, and the quantum-modeled clustering algorithm (quantum K-means). In addition, a pollen-based bee algorithm for clustering is included. These methods can be used in image textures for classification.Algorithms from neural computation [2, 15, 34] for image texture classification will be discussed in Chap. 10.1007/978-3-030-13773-1_9.