Study Of Phase Matching Algorithm Based On Binocular Stereo Vision

Abstract:Binocular vision imitates the human’s eyes which observes objects from the two different viewpoints, and then gets the spatial information and the contour information under the help of some imaging geometry principles which can get a result by calculation, and then get the result of the subject’s3D information. Stereo matching, which is the core technology of the Three-dimensional reconstruction and also the focus study and a "pathological" problem of stereo vision, is the process of searching the same point in the imaging process of the pair of images and obtaining the deviation of images.Based on the analysis of the matching method on the present, it focuses on the study of the image matching method based on phase information in this paper. The anti-noise performance of phase correlation method is very good because the phase itself reflects the position information of image structure, and the matching speed of phase correlation method is very fast because it’s processed with the correlation coefficient of the whole image. But in the process of image mathcing with shift variation, it will be failure when the algorithm of image matching based on phase correlation is used in the conditions which edge is unstable. Aiming at handling the problems of the unstable edge in images, it’s used with phase correlation method based on template matching to match images in this paper. This method uses searching style with small window in the matching process, the same as tradition template matching, needing the whole image traversal, so it results in the increase of calculation and influences the speed of this method. In view of the deficiency of the method, it’s used with searching method of large window and extending the template to the size of original image, and then processed with the method based on phase correlation. This method of phase correlation based on the extended template can greatly reduce the amount of calculation because of using the global searching method with big window, so as to improve the matching speed.Since the existence of affine transformation is often appeared in actual images, only the matching method based on phase correlationcan can not meet the demand. so, the image matching of phase correlation using Fourier-Mellin transform with its invariance of rotation and scaling obtains the the parameters of displacement, rotation and zoom through converting an image from the spatial Cartesian coordinates to the log-polar coordinate. However, the ambiguity will be produced inevitably when the image has scale change, thus the adaptation range of the Fourier-Mellin matching method based on phase correlation is limited in scale change. Therefore, it’s introduced with the theory of Gaussian scale space in this paper, and multiscale structure images are constructed by using simulation of the mechanism of human visual imaging, and then match each scale image with Fourier-Mellin transform. This approach reduces the ambiguity caused by changing the size in images and expands the adptation range in scaling change to a certain degree in the matching method using Fourier-Mellin transform base on phase correlation. The bilinear interpolation algorithm is introduced and the scale space theory based on bilinear interpolation is proposed by the understanding of Gaussian scale space theory in this paper, and multiscale images based on bilinear interpolation are obtained by replacing Gaussian kernel function with bilinear interpolation kernel function and matched each one with Fourier-Mellin transform. This approach carries on processing to the pixel structure itself, so it can greatly expand the adaptation range in scaling change in the matching method using Fourier-Mellin transform base on phase correlation.At the same time, the experiments will prove the effectiveness of the image matching based on phase correlation proposed in this paper through the standard image and the images collected.