The Research Of Image Fusion Techniques Based On Multiwavelet Transform

Recently, multi-source data fusion is the hotspot of international remote sensing research, and it becomes the key technique of the useful information segmentation from mass data. In the paper, some typical methods on multi-source data fusion of remote sensing are introduced and experimented, including IHS(Intensity-Hue-Saturation), HPF(High pass filter), PCA(Principal components analysis) and Wavelet Analysis. The transforms of IHS, HPF and PCA, applied in fusing remote sensing images, have been very popular. And many methods have been produced at the base of these three methods. So it is very important that we learn the inner characters of these three methods. By the experiments in this paper, the three methods are analyzed. Wavelets analysis fusion method is the one that has been developed in recent years. It has reached the mature phase that the method is applied in fusing images by using wavelets multi-resolution analysis. Based on the study and experiments in scale wavelet analysis, we attach the importance to the multi-wavelet analysis. We adopt IHS, scale wavelet, and multi-wavelet methods to fuse the SPOT and TM images. The results show that the multi-wavelet method can acquire better results. Moreover, we find the character of multi-wavelet in multi-spectral image fusion by the deep research in multi-wavelet. It can provide the theoretical support to the reconstruction of multi-wavelet for the image fusion. Every method has different character. Considering from different applied fields, the performance evaluation results of fusion images can be acquired differently. In addition, because the remote sensing data are complex, and people have high request to the performance of the data fusion system, it is better that several fusion methods can be combined and be used together in the image fusion. So in fact, we should adapt different fusion methods according to practical conditions and applied areas and applied aims.