DocumentCode :
3050478
Title :
A multi-core architecture for face detection—Based on application specific instruction-set processor
Author :
Ren, Haiwang ; Che, Ming
Author_Institution :
Sch. of Comput. Sci. & Technol., Tianjin Univ., Tianjin, China
fYear :
2011
fDate :
26-28 July 2011
Firstpage :
3354
Lastpage :
3357
Abstract :
A detection system, which could extract the face information from image or video frame, is becoming more and more important in lots of field, such as Computer vision, image processing, artificial intelligence and so on. In this paper, a multi core architecture, based on ASIP, is proposed for Viola and Jones´s algorithm. A novel scan strategy is adopted to make the data accessing more regular. Meanwhile, through multiple levels optimization of detection processing and the application-oriented ASIPs, the design is able to meet the overall demand. The architecture is prototyped on Altera Cyclone II FPGA, and the result shows that with lower resource consumption it performs real-time processing and high detection rate.
Keywords :
application specific integrated circuits; face recognition; feature extraction; field programmable gate arrays; information retrieval; instruction sets; microprocessor chips; multiprocessing systems; real-time systems; Altera Cyclone II FPGA; Jones algorithm; Viola algorithm; application specific instruction-set processor; application-oriented ASIP; data accessing; detection processing; face detection; face information extraction; multicore architecture; multiple level optimization; real-time processing; resource consumption; scan strategy; video frame; Computer architecture; Face; Face detection; Field programmable gate arrays; Object detection; Random access memory; Real time systems; ASIP; AdaBoost; FPGA; Multi-Core Architecture;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Multimedia Technology (ICMT), 2011 International Conference on
Conference_Location :
Hangzhou
Print_ISBN :
978-1-61284-771-9
Type :
conf
DOI :
10.1109/ICMT.2011.6003093
Filename :
6003093
Link To Document :
بازگشت