Image classification based on KD-Tree structure

Abstract

In this paper, a method of image classification based on a KD-Tree structure is proposed to perform multiple classifications for an input image according to the multi-layer classification model. The process of image classification based on the KD-Tree structure is conducted with the method of building a KD-Tree structure and training a set of classifier vectors. Therefore, some algorithms, including an image classification algorithm based on the KD-Tree structure, classifier vector training algorithm, and image classification models, are proposed. Based on this theory, the experiment is built on the image sets COREL, Wang, CIFAR-100, Caltech101, and Clatech256 with two groups of features: SCH and SCH36 and compared with other works with the same data set to demonstrate the feasibility of the proposed method. The results show that our method is effective and can be applied to image classification systems in different fields. At the same time, this is a new approach to the KD-Tree structure, applied to the image classification problem with higher performance than other methods with the same set of images.