Enhancing group resolution of TM6 based on multi-variate regression model and semi-variogram function |
| |
Authors: | Ma Hongchao Ph D Li Deren |
| |
Institution: | (1) National Laboratory for Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, 129 Luyou Road, 430079 Wuhan, China |
| |
Abstract: | It is well known that Landsat TM images are the most widely used remote sensing data in various fields. Usually, it has 7
different electromagnetic spectrum bands, among which the sixth one has much lower ground resolution compared with the other
six bands. Nevertheless, it is useful in the study of rock spectrum reflection, geo-thermal resources exploration, etc. To
improve the ground resolution of TM6 to the level as that of the other six bands is a problem. This paper presents an algorithm
based on the combination of multi-variate regression model with semi-variogram function which can improve the ground resolution
of TM6 by “fusing” the data of other six bands. It includes the following main steps: (1) testing the correlation between
TM6 and one of TM1-5, 7. If the correlation coefficient between TM6 and another one is greater than a give threshold value,
then select the band to the regression analysis as an argument. (2) calculating the size of the template window within which
some parameters needed by the regression model will be calculated; (3) replacing the original pixel values of TM6 by those
obtained by regression analysis; (4) using image entropy as a measurement to evaluate the quality of the fused image of TM6.
The basic mechanism of the algorithm is discussed and the V C++ program for implemeting this algorithm is also presented. A simple application example is given in the last part of this
paper, showing the effectiveness of the algorithm.
Project supported by the National Natural Science Foundation of China (No. 40023004) |
| |
Keywords: | multi-variate regression model semi-variogram function image fusion template window V C programming |
本文献已被 维普 SpringerLink 等数据库收录! |
|