The Research On Algorithms Of Multisource Remote Sencing Image Fusion And Its Application

As the rapid development of the correlative technologies such as sensor, remote sensing flat and data traffic, the multisource remote sencing image data about the same scene obtained by all kinds of sensors are more and more. However, the level of remote sensing application is dropped behand the development of remote sensing technology. Up to the present, along with the appearance and rapid development of the data fusion technology, multisource remote sencing image fusion, as an important branch of the data fusion theory applied in digital image processing and remote sensing, provides bran-new approaches to the process, analysis and application of multisource remote sencing image data. It is also propitious to enhance the function of multidata analysis and provides an excellent foundation to the research on the large scale remote sensing application.Image fusion means to process spacial registration with all kinds of image data about the same scene obtained by all kinds of sensors, integrate the notable or complementary information adopting some algorithm organically, and then to produce new image data and interpret them. This dissertation, based on analyzing and summing up existed literatures and research productions in the round, adoping high-resolution panchromatic images and low-resolution multispectral images as source data, starts from researching on the technology and application of pixel-level fusion, and develops several aspects of research jobs as follows:1. This dissertation expatiates on the conception of multisource remote sencing image fusion, analyzes and summarizes the intention, hiberarchies, flow and methods of multisource remote sencing image fusion systematically;2. To improve the fusion quality of remote sencing image, this dissertation analyzes and investigates the preprocessing technology of multisource remote sencing image fusion, it contains: characteristic analysis(TM image, ETM+ image, SPOT secondary planet and China-Brazil earth resorce secondary planet image), geometric rectification, enhancement and registration of remote sencing image;3. This dissertation expatiates on the conception of multisource remote sencing image pixel-level fusion, analyzes and summarizes the weighted fusion algorithm, product fusion algorithm, ratio fusion algorithm, high pass filter fusion algorithm in spatial domainin, and IHS transform fusion algorithm, principal component analysis fusion algorithm, pyramid fusion algorithm, wavelet transform fusion algorithm in transformation domain in the round, simulates all kinds of remote sencing images with Matlab software, and summarizes the function, advantage, disadvantage, and circumstance in point of the traditional fusion algorithms by contrasting the fusion images.4. Based on analyzing and summing up four regional characteristics, this dissertation introduces regional characteristic rule into traditional IHS transform and wavelet transform fusion algorithm, and puts forward a new improved IHS wavelet transform fusion algorithm based on regional variance rule.5. This dissertation investigates the subjective and objective evaluation criteria, and evaluates the result across-the-aboard in the experiment by adopting the evaluation criteria combining qualitative analysis with quntitative analysis. By analyzing the experiment results, it also sums up the characteristic and circumstance in point of some commen fusion algorithm, approves that the new algorithm is better than traditional fusion algorithms, and provides more scientific basis for the application of remote sensing image fusion.