Panoramic Image Mosaicing Technology For Regional Monitoring

The panoramic mosaics technology is widely used in military and civil fields, and has always been a research focus in the field of image processing. Based on the background of engineering application, this paper proposed an algorithm of panoramic stitching, which performance is robust and efficient. Then, in view of the particular simulation scenario from Armed Police for practical task, a system of panoramic stitching facing regional monitoring is designed and implemented.This paper focuses on the implementation of the panoramic stitching system by use of the ant colony algorithm(ACA) and mutual information measure, which work can be summarized as follows.(1) The related theories of panoramic stitching technology are introduced firstly, especially highlighted fears about the feature detection and description based on scale invariant feature transform(SIFT), and elaborates on the essence theory and technological process of SIFT. In view of the existing problems of large amount of calculation and more false matching in the SIFT algorithm, this paper proposes an image registration algorithm of Coherent Point Drift(CPD) based SIFT feature. The theory of the algorithm above-mentioned is analyzed and discussed in detail, and the specific steps are listed. Finally, Experimental result of the algorithm shows that, compared with the classical SIFT algorithms, the proposed method can improve the accurate matching result to some extent and is more robust to registration.(2) An image mosaic technology based on ACA and mutual information measure is proposed in this paper. On the basis of the mutual information between the images, the algorithm acquires the matching regions by computing the image information entropy. Secondly, the optimal matching region is gradually obtained by the use of ACA, which takes the mutual information measure as its fitness function. Then the scene images are joined together into a complete and large image containing all the images information. Finally, compared with the stitching method based on SIFT features, the mosaic experiments showed that our proposed algorithm has a better stitching effect and higher splicing efficiency.(3) A regional monitoring system of panoramic mosaics is designed and implemented. In accordance with the Armed Police for practical tasks, such as the group of large disturbances scene and detaining important target indoor, a software system with image registration and fusion is built based on the certain application environment, which consists of requirement analysis, functional design, class design and interface design. A series of tests and analysis show that the system has a good compatibility, and can well splice the sequence of images in a higher speed.