Title :
GPU Accelerated Multimodal Background Subtraction
Author_Institution :
RSISE, Australian Nat. Univ., Canberra, ACT
Abstract :
Although trivial background subtraction algorithms (such as temporal averaging) can execute quite quickly, they do not give useful results in most situations. More complex algorithms usually provide better results, but are typically too slow for widespread use. Here, we examine the architecture of the GPU and describe how a multimodal background subtraction algorithm can be implemented on graphics hardware to provide useful results in real-time.
Keywords :
coprocessors; image segmentation; GPU; graphics hardware; multimodal background subtraction; Acceleration; Central Processing Unit; Graphics; Hardware; History; Layout; Pixel; Robustness; Traffic control; Vehicle dynamics; GPU; background subtraction;
Conference_Titel :
Digital Image Computing: Techniques and Applications (DICTA), 2008
Conference_Location :
Canberra, ACT
Print_ISBN :
978-0-7695-3456-5
DOI :
10.1109/DICTA.2008.77