Atmospheric Correction Algorithm Improvement For Tm Data Based On Multi-Threshold Region Growing

Affected by atmosphere, remote sensing data acquired by sensor is not accurate. We proposed a method to determine the dark pixel values for Shuangtaizi estuary. This improved algorithm has combined normalized difference vegetation index (NDVI), normalized difference water index (MNDWI) as well as multi-threshold region growing algorithm. The atmospheric corrected results of TM image show that the improved Dark-object method eliminates the atmospheric influence very well. It can make the typical ground object spectral curves close to the real spectral curves. In addition, compared with FLAASH and Dark Subtract atmospheric correction module, the improved Dark-object method has a better result in atmospheric correction for avoiding the correct excessive situation of FLAASH.