Study On Video Mosaic With Synchronic Adjustment Technology

Multi-source video splicing technology can synthesize a large field of view, theultra-high-resolution video images, avoid expensive overhead using a special device,and to facilitate universal application, built in satellite remote sensing imageprocessing, medical image processing, virtual scenes, computer vision and otherneighbor GF application domain has become a hot topic in the video imageprocessing. At home and abroad on this subject were a lot of research, but the researchwas not yet time synchronization for different video sources calibrated scholars toresearch. For reconfigurable processor applications, the hardware implementation ofthe algorithm detailed and in-depth research. Video image stitching algorithmreasonable division of the hardware and software. Of this work are as follows:(1)For different video sources stitching time synchronization calibrationalgorithm. Alignment is not only the position in space of the different video sources,and calibration in the time domain for the different video sources. First, the videosource based on the static image registration. The advantage of multi-scale analysisfeature point detection, to increase the robustness of the scale transformation. Willfeature points coarse matching by normalized cross-correlation algorithm. After usingRANSAC ideological the mismatching points on culling and get the correct affinetransformation matrix. Secondly, through the motion estimation, as well as the affinematrix time error of the different video sources by the displacement difference on thetrajectory an enlarged manifested. Inter-frame interpolation of the video source, thealignment of the video image to time error correction of the different video sources,and thus can be a time-space time parameters as a weighting coefficient. By themulti-scale fusion method, the image of the different video sources at different scaleslayer pixel level fusion, the use of the advantages of the multi-scale transform cansmooth the transition of the overlapping region of the two pattern, avoiding theobvious visual sense of the gap.(2)The video frames of the progressive scan recording time of the imagecalibration method. Calibration in time for progressive scan cameras with frameimage while videotaping the shortcomings modeling studies and proposedcompensation method for this problem, to reach the video image of the same frame ofimage pixels at the same time images. Matlab modeling experiments and analysis.(3)FPGA-based reconfigurable processor video splicing system. Use ofreconfigurable processor advantages of video algorithms, this algorithm utilizesreconfigurable processor and FPGA hardware circuit combination method to constructthe embedded video image stitching system. Rational division of hardware andsoftware, the complexity of the image pre-processing operation assigned to thereconfigurable processor calculations, in order to improve calculation accuracy andcomplexity; achieve the fusion operation and the video stream to the FPGA hardwarecircuit, in order to enhance the fusion calculation speed to meet the needs of the videostream. Finally, the video image is displayed on the terminal device.