Study On The Technology Of Video Mosaic Based On DSP

Image mosaic is a technology that carries on the spatial matching to a series of image which overlapped with each other, finally builds a seamless and high quality image. Image mosaic has become an important research technique of image processing and computer vision, it also has a wide range of applications in the medical, virtual reality, monitoring as well as smart home that is very popular at present and so on. This paper mainly studied image mosaic based on the Harris feature, and makes a deep study and analyze on the Harris feature. Then we build the platform of image processing on the DSP and FPGA, and transplant the algorithm on the platform. This paper’s research contents are as follows:Firstly, this paper present an improved method specific to the weakness of setting the threshold value of Harris in the image mosaic.We first compute the edge points of each pixel point by integral image in the binary edge image, and then compute the pixel’s threshold value of Harris corner according to the numbers of in the neighborhood.According to the numbers of edge corners in the neighborhood changes the threshold value of Harris feature of every pixel in the image, this can solve the problem that there too much Harris features in area with intensive edges result from a same threshold value in the Harris detection.Secondly, we build a platform of image processing based on DSP and FPGA, then we transplanted the improved algorithm on DSP. We optimize the program on the level of program and algorithm combined with the structure of DSP, convert floating point to fixed point, call the linear assembly and put a simple method of the Gaussian smoothing.