Image Classification Based On Bag Of Visual Words Model

The bag-of-words(BOW)model was initially applied to text classification,in recent years,The bag-of-words model has been widely applied in the field of image recognition and image classification as deeper understanding about the text classification and its simplicity and efficiency.However,all scale-invariant feature transform(SIFT)features are clustered to construct the visual words which result in a substantial loss of discriminative power for the visual words.The corresponding visual phrases will further render the generated BOW histogram sparse.In this study,the paper aim to improve the classification accuracy by extracting high discriminative SIFT features and feature pairs.First,high discriminative SIFT features are extracted with the within-and between-class correlation coefficients.Second,the high discriminative SIFT feature pairs are selected by using minimum spanning tree and its total cost.Next,high discriminative SIFT features and feature pairs are exploited to BSIFT method to construct the visual word dictionary and visual phrase dictionary,respectively,which are concatenated to a joint histogram with different weights.The paper introduces the theory of the bag-of–words model and the characteristics of the state of art algorithm SIFT for low-level feature extraction,and proposes a binary BSIFT method according to the time consuming process of image matching and clustering,which testify the validity of our method.Then constructing the visual words and the visual phrases by using the high discriminative SIFT features and the high discriminative SIFT feature pairs,which are extracted from the training images with the within-and between-class correlation coefficients and minimum spanning tree.Then concatenated to a joint histogram with different weights according to their different roles Compared with the state-of-the-art BOW-based method,the experimental results on Caltech 101 dataset show that proposed method has higher classification accuracy.