Research On Image Fusion Of SPOT5 Based On Pixel Layer

The fusion of digital image has become a valuable tool in remote sensing image processing in order to achieve improved information for decision making. The research selects SPOT5 image as the original fusion image. It is to improve the spatial and spectral quality of image, and to provide the fusion image of increased interpretation capability for the studied zone. The research approaches that this paper adopts include Brovey transform, spatial high pass filter(HPF), intensity-hue-saturation (IHS) transform, Principal component analysis (PCA) transform, wavelet transform, and the combination of the above approaches. And this paper evaluates the fusion images according to subjective direct observation and objective statistical parameters.The research result shows that the Brovey transform is simple and easy to actualize image fusion, but serious color and spectrum distortion occur in the fusion image. HPF fusion's result shows that image definition is improved largely, but the value of spatial correlation is the minimum compared with the application of other approaches. The advantage of IHS transform is that image entropy is big. It enhances spatial detail information of the fusion image. IHS transform fusion, however, can only be applied to three bands, while it brings obvious spectrum distortion. The image of PCA fusion has the pixel average value, which is the most proximal one to the ideal value, and the highest spatial correlation. PCA fusion approach can be applied to two or more bands. Thus, it is widely used in image processing.After theoretical research, fusion experiment, and fusion image's statistical parameters analysis, this paper makes the conclusions as follows. First, all the fusion images by different research approaches take on good look by subjective direct observation. Second, these fusion approaches not only enhance spatial detail information of the fusion image, but also preserve spectral information of the original multi-spectral image to a certain degree. The fusion image is the embodiment of the advantage complement from different images. Third, in order to lighten the spectrum distortion of the fusion image and the original multispectral image, histogram match is necessary in the fusion process. Forth, the fusion approach based on wavelet transform is the best among these approaches. Fifth, 3-level decomposition in wavelet transform fusion achieves a fine balance between spatial quality and spectral quality. Last, this research adopts straight addition wavelet transform fusion approach to simplify the traditional wavelet transform fusion, and gains more satisfying spatial and spectral quality fusion images than the result of traditional one.