Cloud Classification in JPEG-compressed Remote Sensing Data (LANDSAT 7/ETM+)

Abstract

Environmental parameters required for geo-information modelling are subject to spatial and temporal dynamics. Remote sensing data can contribute to measure those parameters. For that purpose high-accuracy classifications of remote sensing data are required which can be very time-consuming due to the large data volumes involved. In many applications, however, the rapid provision of classified mass data is of higher priority than classification accuracy. One important focus on research and development efforts in the past years has been to optimise the automated interpretation of remote sensing data. Different investigators have shown that this interpretation can both be effective and efficient in JPEG compressed data with acceptable accuracy. This paper presents an operational processing chain for cloud detection in JPEG-compressed quick-look products of LANDSAT 7/ETM+-scenes (compression ratio is 10:1). Two well-developed conventional algorithms are applied to these datasets for cloud detection. Results show that the processing chain developed is stable and produces quality results with substantially compressed mass data.