Location:Home > Engineering science > Information and Communication Engineering > Research on Multisource Image Fusion Algorithm and Its Applications Based on Statistics and Reasonin

Research on Multisource Image Fusion Algorithm and Its Applications Based on Statistics and Reasonin

Downloads: []
Tutor: ZhangJunPing
School: Harbin Institute of Technology
Course: Information and Communication Engineering
Keywords: Multi- source image fusion,Feature Extraction,Statistical correction,Evidential
CLC: TP391.41
Type: Master's thesis
Year:  2007
Facebook Google+ Email Gmail Evernote LinkedIn Twitter Addthis

not access Image Error Other errors

In recent years, along with the rapid development of remote sensing technology, many ways reflect the goals of the characteristics of the surface features of the various remote sensors, multi-source data constitute the same area, and how these massive data of multi-source as much as possible as a whole comprehensive utilization in order to extract more complete and accurate information, and provide the basis for human resources, environment and disaster investigation, is a key issue in the field of remote sensing image processing. According to the imaging principle and characteristics of the different remote sensor, multi-source remote sensing image fusion technology can make use of different combinations of mode, image enhancement to improve image quality, improve the correct rate of image classification and target identification, such fusion results more accurate and reliable than the use of single-source remote sensing information. In this paper, several specific image source, on the basis of analyzing the principle of pixel level fusion algorithm to achieve spectral information retention improvement, and further discussion of feature level and decision-making fusion algorithm in order to improve the feature identification and classification accuracy. First, in the analysis of several conventional pixel level fusion algorithm such as the IHS, PCA, SCN, based on the fusion results in the introduction of the spatial information, spectral information will produce different degrees of damage, this paper proposes the use of statistical improved method of correction. That the statistical properties so that it has some degree of consistency between the adjustment of some components of the high spatial resolution image and multi / hyperspectral image. The experimental results show that the image of the fusion results in the increase while the spatial detail information, spectral information has improved retentive. Secondly, in order to different SAR imaging mechanism with multi-spectral image fusion and applied to the classification in fusion-based feature extraction algorithm, namely the use of texture statistical methods to extract texture feature of SAR images and multispectral the image spectral feature composed feature set as maximum likelihood classification in contingent classification input. The experimental results show that the method of classification accuracy than the result of the use of pixel level fusion classification and multispectral classification of SAR images directly. Finally, two specific applications to study the decision-making level of multi-source image fusion, multi-source image the target area to identify and multi-source image classification. This article discusses the reasoning fusion algorithm based on the evidence, the forthcoming single source decision-making into the basic probability assignment evidential reasoning, to a combination of the use of a combination of rules, according to the decision rule fusion results. On this basis, the basic probability assignment constructor. From the results can be seen to reduce the uncertainty of the single-channel source to the multi-source fusion, and improve the accuracy of identification and classification.
Related Dissertations
Last updated
Sponsored Links
Home |About Us| Contact Us| Feedback| Privacy | copyright | Back to top