Evaluation of Landsat-8 OLI satellite imagery by a cross-comparison method combining spectral and texture features

摘要

As NASA’s eighth satellite in the Landsat series, there are substantial differences between Landsat-8 and other instruments onboard previous Landsat satellites. Consequently, it is highly necessary to perform the data quality assessment of Landsat-8 prior to various applications. This study presents a cross-comparison method to evaluate the Landsat-8 OLI against Landsat-7 Enhanced Thematic Mapper Plus imagery by combining spectral and texture features. A total of three kinds of indicators are used including spectral response function, statistical features and texture filters. The analysis results show that (1) the intersection areas of Landsat-8 OLI sensor decrease greatly among neighboring spectral bands on the spectral response curves and the slopes are higher in different spectral regions, which indicate that the spectral response of Landsat-8 OLI is more accurate; (2) much improvement of Landsat-8 OLI has been made in spectral resolution (9 vs. 7 bands), radiometric quantization (8 vs. 12 bits), and information quantity in the OLI image; and (3) Landsat-8 is also superior to Landsat-7 in the four texture filters including ASM (Angular Second Moment), correlation, contrast and entropy which are derived from GLCM (Grey Level Co-occurrence Matrix).