Improvement of Word Bag Model based on Image Classification

Bag of words was first used in the field of text categorization. In recent years, with the development of image processing technology, Bag of words has also been widely used in image classification and image retrieval. The use of Bag of words has simple and effective features, but in the image. There are still many defects in the classification. After researching and analyzing the SVM image classification model based on traditional Bag of words, it is found that there are two shortcomings in the traditional Bag of words in image classification: (1) the accuracy of visual word representation is not high; (2) the flexibility is poor. According to these two defects, combined with the contrast words and Hamming embedding technology, the traditional word bag model is improved, and a high discriminative force of Bag of words is applied and combined into the SVM image classification model. A SVM image classification model based on high discriminative Bag of words is presented. The experimental results show that the accuracy and applicability of the classification are improved.