Spectral normalization between Landsat-8/OLI, Landsat- 7/ETM+ and CBERS-4/MUX bands through linear regression and spectral unmixing

Monitoring changes on Earth's surface is a difficult task commonly performed using multi-spectral remote sensing. The increasing availability of remote sensing platforms providing data makes multi-source approaches promising, since it can increase temporal revisit rate. However, Digital image processing techniques are needed to integrate the data, since sensors can be quite different in terms of acquisition characteristics. This work addresses the spectral normalizing of three medium spatial resolution sensors: Landsat8/OLI, Landsat-7/ETM+ and CBERS-4/MUX, through linear regression and linear mixture model approaches. The results showed slight better results when using the linear regression approach.