Research And Application Of Image Mosaic Based On Feature Point And Fusion Algorithm

Image mosaic is geometric registration for more than two or two images of the same scene with each other spatially overlap region. It contains all the image information to form a clear wide viewing the scene through the seamless integration of the interpolation method for a new image technology. As an important research area of digital image processing, image stitching technology is widely used in aviation major image processing, remote sensing technologies, medical diagnostics, measurement science.This thesis summarizes the existing splicing technology generalizations, for images with translation and rotation transformation, a detailed analysis of image stitching algorithm based on feature points. Based on the image feature points through the stitching algorithm of the space, and the luminance information of the feature points, extracts the image feature points, and the points extracted by combining the main direction and relevance of the initial matching feature points, and then use the similarity pseudo-matching algorithm to measure point to be removed to obtain a valid image sequences between feature points; then select the appropriate geometric transformation model, based on the obtained feature points to calculate the parameters of transformation model, the image in the same coordinate transformation system, the results obtained mosaic images. This thesis focuses on the extraction of feature points, and introduces the analysis of several commonly used feature extraction algorithm, the advantages and disadvantages between the algorithm through simulation experiments to compare. In order to accelerate the accuracy and speed of feature points and feature points in the reasonable choice, Based on a significant part of the unstable region to exclude the feature points, the experiment proved that the effect is significant.For the case generated mosaic image stitching problems, this thesis introduces the bilinear interpolation method eliminates image stitching holes later, and then forced cancellation algorithm to eliminate splicing gap. Since the sampling time and angle different color for a larger image, you need to multi-source image fusion algorithm to handle. This thesis describes a multi-source image fusion algorithm in image fusion algorithm based on wavelet analysis. the algorithm used to solve stitching ghosting blur experimental results prove significant.