Video Calibration Of The Driving Recorder

It is self-evident that driving recorder wills more and more popular on the development trend,because it has a broad perspective and long record time. Now it had had the utilization rate of9.1in foreign countries. And it is more perfect in the passenger car pickup mandatory to use, evenmany people began to spontaneous use it. However, with its rapid development, itself wide-anglelens to bring the optical distortion in the recorded video the same sharp distortion. Especiallyaround the image deformation serious so that the traffic recorder as evidence restore fact have agreat impact.Therefore, there is greatly necessity for its correction.Here, we tailor developed a correction for the distortion characteristics of the traffic recordersoftware. Were it experimental variety of correction algorithm effects, but not ideal. Especiallyspherical mapping projection correction, based on cambered correction method and the latitudeand longitude correction method losses plethora of the around data in a single frame image,result in a loss of the sense of the wide-angle lens, and the corrected image holes or virtual. Thenwe contrast to the complexity of the algorithm and computational because traffic recorder videocalibration requirements, the need to maintain the real-time characteristics, so the algorithm forthe calculation of the amount of time, to a large extent our priority problem. But due to ofconcentric circles correction method and polynomial correction method calculation is too big,taking up too much time so that each frame correction will lead to overall playback delaycorrection, real-time cannot be guaranteed. Finally, we decided to choose the grid method, butdue to its still computing problems, we improve its algorithm to reduce the number of dividedinto quadrants once and origin of symmetry analysis of the other quadrants. So that we can justuse a simple equation to complete other algorithms to complete the correction equation, calculatethe amount of time a lot of progress.However, due to the need to repeat each frame image correction coordinate transformationwork, time still real-time cannot be guaranteed, so we built table method of thinking to build thecode, we will be corrected image coordinates stored in a two-dimensional table, so only need torepeat each frame correction look-up table can be completed, thus ensuring its real-time.This paper for traffic recorder optical distortion correction procedures tailored our experiments comparing multiple algorithms select the most appropriate algorithm to ensure thesystem’s real-time and accuracy, and algorithm improvements and code optimization. Finally, ourexperiments show that the program can correct the distortion of large amount of data trafficrecorder recorded video, and procedures to ensure the real-time output.