期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2013
卷号：50
期号：1
出版社：Journal of Theoretical and Applied
摘要：Scene classification is an important research direction in the computer vision. However, it is not an easy task. We face many serious difficulties and challenges when classifying the nature scenes. A novel approach is proposed to recognize the nature scenes. Based on the traditional Bag of Visual words (BOV) model, the feature field and space field are combined by introducing the Markov Random Field (MRF) when quantifying the image into a collection of unordered visual words. And then the Latent Dirichlet Allocation (LDA) model is applied to learn the topic distribution of scenes. At last, the Support Vector Machine (SVM) is used to build a classifier in order to categorize a new image. The experimental results on the dataset of 15 nature scenes demonstrate that the introduction of the contextual semantic information on the basis of the traditional method can improve the classification performance.
关键词：Scene Classification; Markov Random Field; LDA; Bag of Visual Words
