Study Of Image Mosaics Technology

The automatic construction of large, high-resolution image mosaics is an active area of research in the fields of photogrammetry, computer vision, image processing, and computer graphics. More recent popular application is the digital panoramic mosaics, for applications such as the construction of virtual environments and virtual travel.In this thesis, the author detailedly makes a summary of the main methods used for image mosaics, analyzes the core techniques it should resolve, and emphatically discusses the classical algorithm-Transform Optimization. On the basis of it, by analyzing the available disadvantages of image stitching algorithms and image acquisition, a block-matching based multi-resolution image stitching algorithm (BMMR) is proposed. This algorithm took into account the precision of image stitching as well as the speed of it, as to the pick-up of the base feature block, a simple method of edge information threshold is used, which realized the automatic pick-up of base feature block, therefore improved the precision of image stitching. In the problem of searching matching block, a hierarchical search strategy is employed, which improved the speed of image stitching. The experiments show that our proposed algorithm greatly extended the scope of traditional image stitching algorithm and has good capabilities. Expect for the planar manifold, other mosaics manifolds are also discussed, and a fast cylindrical panoramic image creating technology (CPI) is proposed. Of this algorithm, the 360 degrees of circle panned image sequences are first projected to cylindrical coordinate separately, then block-matching method is used to register with multi-resolution, finally the cylindrical panoramic image display with high truth can be fast realized. In the ends, the fft-based stitching algorithm is simply discussed and its experimental results are also given. Thus, almost all methods on image mosaics are studied in this paper.