Analysis of image fusion and classification for high resolution SAR data on-line

SAR and optical remote sensing image, with highly complementary characteristics, can enhance the integration of information utilization of remote sensing data. Adopting the new Cosmo-Skymed SAR high-resolution image data, we inhibit speckle impact using enhanced Lee filtering. Then we fused this image with a CBERS image using local use standard deviation based on wavelet packet method. Because of fully integrating the characteristics of each image, it can retain the spectral characteristics and details of properties to the maximum extent, improve signal-to-noise ratio, and be conducive to information extraction. The experiments show that the automatic classification accuracy significantly increased and classification Kappa coefficient increased from 0.47 to 0.93 after fusion of Cosmo-Skymed and CBERS02 data. Meanwhile, This paper employ a geospatial information processing concept model complied interoperable system framework and an implementation approach for accessing geospatial information openly by chaining individual service module to assemble complex geospatial processing and executing the processing model to deliver information.