Research Of Remote Sensing Image Fusion Based On Multi-Wavelet Transform

As the development of sensor technology, the multi-resource Remote Sensing image fusion is becoming to be one of the most important researches in the Remote Sensing image processing areas. In recent years, the multi-wavelet as a mathematic analysis theory, which has been widely used and made great progress in many image processing fields. A new Remote Sensing image fusion method combing with the features of multi-wavelet is proposed in this paper, which avoiding boundary interferences when regional processing in the high-frequency part. The major work is summarized as follows:Firstly, the original images are decomposed to different scales of high-frequency and low frequency using GHM,CL multi-wavelet; Secondly the image fusion is then implemented in different bands, regional blocking is adopted in high-frequency in order to avoid boundary interferences in the traditional regional processing and the larger absolute value of the region is taken as fusion coefficients, the Weighted Arithmetic Average algorithm is adopted in low-frequency; The finally fusion image is obtained by multi-wavelet reconstruction. Compared with the other methods, the experiment shows that this fusion method can get a better definition and the fusion effect is much better.For SPOT, TM satellite Remote Sensing images, due to the interference factors of the earth's rotation and shooting angle environment, the image pretreatment process becomes a particularly important part. According to the characteristics of the noise in Remote Sensing image, an adaptive weighted average filtering de-noising method is proposed. The experiments show that this method can effectively filter out the noise and suitable for remote sensing image processing. In the image registration aspects, a triangle automatic calibration method is adapted to realize the image registration automation and achieve good results.For the visualization of remote sensing image fusion process, an image fusion system is designed. It provides the read-image function, fusion methods and multi-wavelet selection, evaluation of fusion results and so on. The system intuitively shows the process and results of Remote Sensing image fusion, it is more practical and easy to operate.