Change Detection

Research on patch based change detection using probabilistic and deep learning approaches …

• I made significant contributions to the change detection research through: (i) efficient modeling of crop phenology, and (ii) efficient modeling of spatial context via image patches. Change detection using remote sensing time series poses two challenges: (i) distinguishing between real changes (e.g., due to damages, change in crop-type from season to season) from normal or expected changes (e.g., changes in green- ness over crop growing season, environmental effects), and (ii) scalability (continental scale). I developed a novel biomass monitoring framework that consists of two key components: change detection using Gaussian Process (GP) Learning, and change characterization using semi-supervised learning. Through a well designed covariance matrix (Toeplitz) that efficiently captures crop phenology, our GP-based change detection framework not only showed the improved accuracy (25% over Bayesian Online Change Detection), but also reduced the computational complexity (time complexity from O(n3) to O(n2) and memory usage from O(n2) to O(n)). In addition, efficient (mixed) parallel implementation has lead to the continental scale continuous monitoring system using daily MODIS satellite image data. This work was well received (won “Directors Best R&D” award at annual ORNL Lab Directed R&D project evaluations; selected as one of the six best papers at NASA CIDU Conference (2010) it got invited to a special issue of Statistical Analysis and Data Mining Journal (4(4), 2011)).
• In addition to the biomass monitoring (which is critical for food security), my research had also resulted in a key capability for monitoring man-made critical infrastructures which allows comparing multi-sensor and multi-resolution images which may not have been perfectly georegistered. Traditional remote sensing based change detection approaches compare individual (or first/second order neighborhood) pixels to identify changes. However, image to image and map to image registrations, through increasingly becoming accurate, contains errors (often by several pixels), as a result direct pixel-wise comparison is not accurate. In addition, often changes happen at a bigger spatial footprint than individual pixels. To overcome these limitations, we developed a novel and unique image patch-based probabilistic change detection that shown to be not only accurate than traditional methods, but also allows multi-resolution and multi-sensor data. The results were published in leading venues (Procedia Computer Science, Vol. 9, 2012; Demo paper at ICDM 2012), and has been extensively used in several NGA, DOD, and DHS sponsored projects at the ORNL.
  • Clayton Connors, Ranga Raju Vatsavai: Semi-supervised deep generative models for change detection in very high resolution imagery. IGARSS 2017: 1063-1066
  • Zexi Chen, Bharathkumar Ramachandra, Ranga Raju Vatsavai: Hierarchical change detection framework for biomass monitoring. IGARSS 2017: 620-623
  • Zexi Chen, Ranga Raju Vatsavai, Bharathkumar Ramachandra, Qiang Zhang, Nagendra Singh, Sreenivas Sukumar: Scalable nearest neighbor based hierarchical change detection framework for crop monitoring. BigData 2016: 1309-1314
  • Ranga Raju Vatsavai: Rapid Damage eXplorer (RDX): A Probabilistic Framework for Learning Changes from Bitemporal Images. IEEE International Conference on Data Mining (ICDM: Demo Paper) 2012: 906-909
  • Ranga Raju Vatsavai, Jordan Graesser: Probabilistic Change Detection Framework for Analyzing Settlement Dynamics Using Very High-resolution Satellite Imagery. ICCS 2012: 907-916
  • Varun Chandola, Ranga Raju Vatsavai: A Gaussian Process Based Online Change Detection Algorithm for Monitoring Periodic Time Series. SIAM Data Mining Conference (SDM 2011): 95-106
  • Varun Chandola, Ranga Raju Vatsavai: Scalable Time Series Change Detection for Biomass Monitoring Using Gaussian Process. NASA Conference on Intelligent Data Understanding (CIDU) 2010: 69-82. (Selected as one of the six best papers at the NASA/CIDU and published in the special issue of Statistical Analysis and Data Mining Journal)