Dynamic Monitoring Using Image Fusion by Wavelet Transform

Remote sensing images are primary data source for dynamic monitoring in land using,at the same time, DEM, relief map, land using detail survey map and other literal information are used as assistant tools in monitoring. There are many kinds of automatically detecting methods applied to detect changing information in land using, including post-classification comparison method and pixel to pixel comparison method, and image fusion method is belong to the latter. Remote sensing image fused by wavelet transform, which is an important image information enhancing method, can be used to detect land using changing patches automatically. After principle and advantages of two dimension wavelet transform theory have been introduced, an idea applying the theory in remote sensing dynamic monitoring in land using to automatically detect land using changing patches is presented. After utilizing principal component transform on a temporal multi-spectrum image, the first principal component is fused with another temporal multi-spectrum image by computing the tow temporal images' wavelet coefficient using weighted mean value, and cubic spline wavelet function is used to transform image to wavelet coefficient. When the three bands which are formed by multi-spectrum image fusion are used in false color displaying, there will be spectrum characteristic changes in areas where land using is changing and the areas are losing spectrum harmony with around objects, so, the changing patches can be detected automatically. Proved by checking indoor information and field validating, the precision can meet practical application demand for the detected changing information with low noise.