Weather Radar Image Mosaic Based On Wavelet Multi-Resolution Analysis

Doppler weather radar is a device for measuring the intensity and regional distribution of the meteorological echo intensity and the regional distribution of the electromagnetic wave, and the radial velocity and the velocity spectrum of the scattering body. However, the Doppler radar detection range is limited, the forecast personnel to carry on the global weather situation to judge has certain limitation, reduces the forecast accuracy. At present, with the increase of the number of Doppler radar installation throughout the country, weather radar network technology has been gradually paid attention to. Network technology will be a large area of all radar images combined into a complete range of radar images, to facilitate more comprehensive observation of the weather. As an important branch of computer vision and image processing, image mosaic technology can effectively solve this problem.In this paper, a comprehensive comparison of existing mature image stitching technology, According to the characteristics of the radar image, wavelet contour features of radar images are extracted by the method of B wavelet modulus maxima. The centroid and Harris feature points are extracted from the contour of the radar image, and the initial matching results are made with the initial matching results of contour centroid and Harris feature points, using RANSAC algorithm to remove the pseudo feature points. Through the method of wavelet multi-resolution analysis, the decomposition of the different frequency components using different fusion rules, and ultimately complete the whole process of stitching. The improved algorithm uses MATLAB language to program. At the same time, a lot of experiments are done in this paper. The effect of the improved algorithm is compared with that of the traditional image mosaic method. After comparison, the algorithm proposed in this paper has been greatly improved the accuracy of the algorithm, and achieved good results.