Research Of Wavelet Image Fusion Approach Based On Grayscale Variation Significance

With the development of multi-sensor technology, image data has increased dramatically. How to utilize image data of different sensors to get more reliable images is becoming a hot research field of image processing. This is precisely the contents of image fusion.There are many image fusion approaches. Wavelet-based image fusion approach is currently an important research direction. Wavelet transformation decomposes an image into low-frequency section containing the image contours and high-frequency section containing the image details. Most researches of wavelet-based image fusion focus on the fusion of high-frequency section, for retaining more image details in the fusion process. Research in the low-frequency fusion algorithm is not much. But when processing the image with large-scale cloud or haze, low-frequency information will become a crucial factor of fusion results.Wavelet image fusion is significantly affected by wavelet parameters. Wavelet parameters should be different when fusing different types of images. In this paper, a series of experiments are performed to determine the selection of wavelet parameters.In order to solve the fusion problem of images with large-scale cloud or haze, a series of researches have been done in this paper. Firstly, characteristics of cloud and haze are analysed. Secondly, model of grayscale variation is established. Finally, a wavelet image fusion approach based on grayscale variation significance is proposed.As is shown in the experiments, this approach can get more satisfied results when fusing the images with large-scale cloud or haze.