Abstract
Abstract. There is an increasing demand for robust methods on urban sprawl monitoring. The steadily increasing number of high resolution and multi-view sensors allows producing datasets with high temporal and spatial resolution; however, less effort has been dedicated to employ very high resolution (VHR) satellite image time series (SITS) to monitor the changes in buildings with higher accuracy. In addition, these VHR data are often acquired from different sensors. The objective of this research is to propose a robust time-series data analysis method for VHR stereo imagery. Firstly, the spatial-temporal information of the stereo imagery and the Digital Surface Models (DSMs) generated from them are combined, and building probability maps (BPM) are calculated for all acquisition dates. In the second step, an object-based change analysis is performed based on the derivative features of the BPM sets. The change consistence between object-level and pixel-level are checked to remove any outlier pixels. Results are assessed on six pairs of VHR satellite images acquired within a time span of 7 years. The evaluation results have proved the efficiency of the proposed method.
Highlights
Remote sensing data are playing a key role for tracking land cover changes
Detecting single building changes from satellite images is becoming possible with the growing availability of very high resolution (VHR) satellite imagery, such as IKONOS and WorldView-2
This study demonstrated that satellite image time series data (SITS) stereo data can be used to generate a detailed and robust building change detection task
Summary
Along with the improved image data quality and the increased amount of available satellite data in recent years, the details and accuracy of the obtainable information are assumed to be enhanced. Recent studies have proved the advantages of change analysis based on satellite image time series data (SITS) (Yang and Lo, 2002; Bontemps et al, 2008). The increase of spatial resolution and multi-view data capability of satellite sensors are exploited to use SITS for building monitoring at small scales. Time-series data could be helpful to improve the quality of each single dataset, as they provide redundant information. A two-step procedure has been developed: firstly, the quality of single images is improved by using spatial and temporal information. The improved time-series building probability maps are analysed to highlight building changes and identify the type of change which took place at a given building site
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
