| In recent years, with the continuous research on bag-of-visual-words(Bo VW) model, Bo VW model has been widely used in the field of digital image processing. Because the image plays an important role in people's life and work, it is significant to quickly and accurately extract the image information from the large scale of image database. This paper mainly focuses on the image classification method based on bag-of-visual-word model, and further makes some improvements on this kind of problem.The main focus in this paper can be summarized as follows:1. This paper proposes a ROI segmentation algorithm based on corner extraction and graph theory, which automatically determines the region of interest in the images. Firstly, the corner points are extracted from the image. Then, the diagonal structure is constructed and the diagonal points are selected according to graph theory. We proves the validity of the proposed algorithm on Caltech 100 set. 2. Based on the improved bag-of-visual-word model, the fuzzy membership function are exploited to optimize the Bo VW model. In the process of constructing the image histogram vector, different fuzzy membership function are exploited for fuzzy histogram representation of image. The image classification has been completed by combining with SVM algorithm. The experimental results show that the improved algorithm has improved the classification results. 3. Combining with the multi-core parallel computing technology, the Bo VW model has been optimized in parallel and thus the efficiency of the algorithm is improved. According to the principle of bag-of-visual-word model and multi core parallel computation, the algorithm is optimized, which includes three aspects: feature extraction, feature clustering and image histogram. The experimental results show that the parallel multi-core computing algorithm improves the efficiency of the algorithm and the practicability of the algorithm in image classification. |
PDF Full Text Request