Intelligent land cover detection in multi-sensor satellite images

Abstract

Nowadays, the availability of satellite images is higher due to the launch of more number of earth observation and inter-planetary mission satellites. The satellite images are widely used in different early warning, risk assessment and disaster models. Hence, the need for efficient interpretation of these resources is also high. Generally, it is a primary issue in satellite image analysis to detect the specific region or individual objects in multiple sensor satellite images of varied spatial resolution. If traditional classifiers are used for such object detection, it needs much training time and tedious ground truth labeling. Hence, the proposed model is focused on reducing the above said complexities and offer efficient detection of the chosen land cover region using Speeded Up Robust Features (SURF), a widely used robust local feature descriptor. The performance of SURF has been evaluated using three different sensor images of moderate resolution for a common study region.