A Novel Image Classification Algorithm Based on Word Bag Model and Feature Extraction

Image classification is based on the different characteristics of the image information to distinguish the images into different categories. It makes quantitative analysis of the image, interpreting the image or the image of each pixel region into a plurality of categories or in one, to replace the human visual interpretation. Traditional classification methods are usually based on single feature analysis. In this paper, a novel algorithm based on word bag model is described. This algorithm extracts a variety of image features at the same time, then establishes comprehensive access to multiple feature vectors, and finally gets more accurate image classification results. Bag of words model is a generalization of the traditional bag of words model. The core idea of the bag of words model method is that: all the words in the corpus statistics composed of words, words for each document statistics on the frequency of use, composed of the words frequency histogram to express this document. Corresponding to this model, we present the texture feature extraction and analysis steps.