Abstract | ||
---|---|---|
Extracting foreground objects is an important task in many video processing/analysis systems. In this paper, we propose a technique for foreground object extraction, under static camera condition. In our approach the spatial histogram of a single background image is modeled as Mixture of Gaussians and this model is updated after every few frames. To extract the foreground, input frames are compared with current background frame model and foreground pixels are classified according to intensity differences. To mitigate the errors caused due to movement of the background objects (e.g tree leaves in outdoor scenes), we also incorporate optical flow in an efficient manner. We demonstrate performance of our approach on various indoor and outdoor scenes. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1145/1924559.1924579 | ICVGIP |
Keywords | Field | DocType |
analysis system,foreground pixel,histogram-based foreground object extraction,g tree,outdoor scene,background object,single background image,extracting foreground object,efficient manner,foreground object extraction,current background frame model,hidden markov model,optical flow,traffic classification,mixture of gaussians,video processing | Background subtraction,Traffic classification,Computer vision,Histogram,Video processing,Pattern recognition,Computer science,Artificial intelligence,Pixel,Hidden Markov model,Optical flow,Mixture model | Conference |
Citations | PageRank | References |
0 | 0.34 | 10 |
Authors | ||
1 |
Name | Order | Citations | PageRank |
---|---|---|---|
Mandar Kulkarni | 1 | 40 | 5.21 |