Fusion Of MODIS Image And TM Image Based On Directionlet Transform

Image fusion is a kind of image enhancement, it is used to combine multiple images of different characteristics into a fused image, the characteristics of the fused image has more obvious, more abundant information. In recent years, image fusion is applied in remote sensing image, medical image and other fields. This paper mainly focuses on remote sensing image fusion, the fusion algorithm based on Directionlet transform.Compared with the classic wavelet transform, Directionlet transformation can get the edge of the image and outline details.MODIS is a new generation of spectral imaging instrument. The resolutions of its spectrum, radiance and time are high. MODIS data hold the advantages that the other data source can't offer because that MODIS image is open to public for free receiving all over the word. Compared with MODIS, TM image has high spatial resolution, but low spectral resolution and temporal resolution. Enhancing MODIS spatial resolution by Dirctionlet transform fusion with TM to take the advantage of TM high spatial resolution must spread MODIS data application, such as big scale mapping for LUCC dynamic monitoring and fire detection etc.Compared with traditional wavelet transform, Directionlet transformation of the basis function is anisotropy and direction. The direction of the basis functions directly related to the direction of the anisotropic target in the image, based on this, this article put forward an adaptive search for the main direction of image to construct Directionlet transformation matrix, which can improve the transform adaptive ability to capture images of anisotropic characteristics.This paper applies improved Directionlet transformation of MODIS data and TM data fusion, the fusion of images with high spatial and high temporal resolution. This paper adopts the traditional fusion algorithm and fusion algorithm based on multi-resolution, the results are compared with Directionlet transformation algorithm.In this paper, the evaluation of fused image using a combination of subjective and objective evaluation. From the objective evaluation standards, we can know that theDirectionlet transform algorithm could improve spatial resolution of MODIS data, and it is better than traditional algorithm.