Research On The Methods Of Remotely Sensed Image Fusion And Its Application In The Inversion Of The Target Spectrum

In recent years, the technique based on remotely sensed image fusion (RSIF) has been concerned widely in both military and civilian filed. RSIF was commonly applied to combine the information of two or more images from a certain scene, and then create new integrated information about the same scene, which can not be gained only from a single remote sensing image. Based on the fusion of several images, the limitations and differences of single image at geometry, spectral and spatial resolution could be overcome, and the quality of the image was improved. Moreover, it was benefit to position, identify and explain the physical phenomena and events. Generally, there were three fusion levels, i.e. pixel level, feature level and decision level.The common used RSIF methods included intensity hue saturation (IHS) transform method, principal component analysis (PCA) method, brovery transform (BT) method, smoothing filter-based intensity modulation (SFIM) method and wavelets transform (WT) method, etc. Among them, the image fusion method using WT was researched popularly at present. In this paper, the above several methods were researched, while the experiments of fusing different remotely sensed images were processed.In order to objectively and quantitatively evaluate the quality of the fusion image which was disposed by different RSIF methods, some important parameters have been selected, including mean value, standard deviation, entropy, joint entropy, average gradient, correlation coefficient and deviation exponent. After analyzing the evaluation results of these parameters, one or two proper RSIF methods were obtained.Obviously, the physics base of the remote sensing technology is that landmark can reflect, absorb and emit electromagnetic wave. Actually, the essential attention of remote sensing and its application is focus on imaging and inversing spectrum characteristics. When the remotely sensed data were used to inverse the characteristic of object spectrum, their utilization would be improved, and then the ambiguity and uncertainty of object recognition also could be reduced. In this paper, various remotely sensed datasets were analyzed firstly. Then, taking airport as an example, the target spectrum was inversed on the basis of the different RSIF technologies.The results showed that: (1) As two methods used, WT together with PCA and SFIM, good characteristics including higher spectral fidelity and more information have been obtained in TM, ASTER and MODIS fusion; (2) The effect of combining ASTER and MODIS fusion was better than the effect of TM and MODIS; (3) Good relevance and coherence of the inversion results were gained. It was reflected that the object spectrum inversion based on the image fusion had much higher feasibility and reliability.