Research On The Process Of Video Mosaic

Image mosaic, that is developing with multi-subject such as technology of multimedia, virtual reality technology, theory of communication, signal processing and the pattern recognition, is one of the technology of image processing. Image mosaic has been applied broadly in the areas of space exploration, submarine detecting, medicine, weather reporting, video index and searching, the objects 3-D reconstruction and the high resolution processing of digital cameras. Video mosaic is an important application of the image mosaic, and the mosaicked video followed by the image mosaic has a larger visual angle and a higher resolution. It can be applied to the area of exhibition, vision entertainment and advertisement.This text aims at mosaicking the videos collected by two cameras which were se fast to each other. According to the flow of the video mosaic, firstly the restrict condition for the two cameras’ location relationship and the setup of the cameras’ parameters is proposed according to the experience of experiment. Then the content of the video pre-processing including video synchronization, video image enhancement and video image denoising is introduced. At last, according to the flow of image mosaic, summarizes the theory of image mosaic and the image fusing of image mosaic.This text researches a theory of denoising, smoothing and an improvement theory of image fusing using gradually in and out mainly. The first theory makes use of multiplying the eight domain pixels and these powers to compute the noised pixel value, and the powers are self-adaptive computed according to the eight pixels’ direction differential coefficient. This theory has a nice effect in both image denoising and image detail keeping, compare with the median filtering this theory has obvious dominance, and this theory is mainly used in the video pre-processing and eliminating the seam caused by mosaic. For the gradual change gene in the session of image fusing, this text proposes that uses the curve which has the characteristic of the Gaussian curve to take place the conventional linear curve. In addition, just make use of the middle one third of the overlapped area to form the fusing area so the most of the overlapped area will been kept up and can achieve one image transit rapidly to anther. This theory can reduce the blur and restrain generating virtual shadow, and the mosaicked image is real and natural. At last, to estimate the effect of image mosaic, referring to other image estimation, one method of making use of every column of the fusing image was introduced.Finally, the image in the mosaicked video looks natural and the luminance and chrominance transit smoothly without any man-made mosaic mark. The mosaicked video can apply to different projects environment.