The Study Of Remote Sensing Image Data Fusion Methods And The Effect Of Evaluation

With the rapid development of remote sensing technology,the multi-platform,multi-temporal,hyperspectral and high-resolution satellite data come forth in large numbers, are widely used in various fields of geographic. Because different sensors have their own characteristics, different remote sensing images have clear differences in spectrum information and high-frequency information.In practical application, an image information is often not meet the purpose of the experiment, most of the need for a comprehensive variety of images analysis. Funding for some projects a relatively small, while the higher prices of high-resolution images, thus limiting the use of high-resolution images. In order to solve these problems, fusing the multi-source remote sensing images is one of the very effective means. Fusion images can be integrated in different image characteristics, to obtain the required information. At present the fusion of multi-source remote sensing image is a hotspot study in the field of remote sensing.There are many experts and scholars, have researched the image fusion methods. Generally.it can be fused at three levels:pixel level fusion, feature level fusion and decision level fusion. This paper will labor the fusion methods at the basis of previous studies, systemiced descripte the hiberarchy of image fusion, and detailed analysis the elements and characteristic of the fusion methods whose application has been mature. It will put the pixel level fusion and feature level fusion as the focus, such as the his transform, pea transform, brovey transform, hpf transform and wavelet transform, etc. Finally,it simple evaluate the fusion methods what can offer theoretical basis for the choice of fusion methods in the fourth and fifth part of the paper.Although the application of the fusion methods are quite mature, the evaluation of the fusion imaging's effect also did not form a unified standard. The current evaluation indexes are mostly based on statistical characteristics, they are only at a point of view or a particular aspect of image fusion to evaluate.it only consider single-factor index, there is no comprehensive evaluation of a variety of factors.Based on the analysis of traditional evaluation index, this paper found a systematic and comprehensive evaluation system, designed a new algorithm, with a view to a comprehensive, systematic and accurate for quantitative evaluation of image fusion effects to make it more objective and real, more convincing. Specifically:(1)evaluate at the spectrum fidelity and image definition two aspects to found spectrum information index and high-frequency information index. (2) combinating these two indexes to integrated evaluate the fusion image based on the amount of information. (3)combining the structure similarity principle, considering the person's visual information, established a combination of qualitative and quantitative evaluation index-fusion index.Because of the diversity of data sources, this paper analyzes the imge fusion of homogeneity sensor data and multi-source remote sensing data (different sensors). The fourth of the article, studying the image fusion as the Spot5 multi-spectral data and panchromatic data of Kaifeng City as example. The fith of the article, studying the image fusion as the ETM multi-spectral data and Quikbird panchromatic data of Tianjin City as example. They mostly use the pixel level fusion approach to analysis, and then progress integrate evaluating in the aspect of spectrum fidelity, high-frequency information into the degree, as well as the amount of information. According the fusion results, to validate the valuation of fusion methods in the second part and the reasonableness of the evaluation system in the third part; and further analysis the effect of the bands selection and sensor selection to the image merge.The first section of this paper,it described the multi-source remote sensing image data fusion need, application and significance,then it elaborated the study route of the paper. The second part, it made an system description and summary for the remote sensing data fusion methods, and evaluate the methods, then form a system of different levels.The third part, researching the evaluation of image fusion results,and establishing an evaluation system.The fourth and the fifth parts are case studies.At the last of this paper, it summarized the multi-source remote sensing image data fusion, and illustrated a number of objective problems and shortcomings, and then prospects for image fusion.