A novel image fusion algorithm based on the support vector machine (SVM) is proposed. The original images are fused with different block sizes according to the positions of the original image blocks. Three features, i.e. the standard deviation, the DCT high frequency energy, and the spatial frequency extracted from each partitioned original image block are used to represent its clarity. Firstly the algorithm decomposes the original images into large image blocks. After the original large image blocks, which are clearer,are chosen using SVM, the original large image blocks that are on the boundary between the focused area and the blurred one are decomposed into small image blocks. Then the small image blocks are selected by the trained SVM. Finally the small image blocks that are on the boundary between the focused area and the blurred one are fused with the discrete cosine transform (DCT). Experimental results show that the proposed approach outperforms the conventional DWT-based and DCT-based image fusion methods and image fusion schemes using the fixed image block size.
展开▼