DocumentCode :
2531353
Title :
Multicore DSP implementation of adaptive correlation tracker
Author :
Singh, Arun ; Gautam, Pradeep Kumar ; Jagtap, Maheshkumar P.
Author_Institution :
Adv. Numerical Researchand Anal. Group (ANURAG), Hyderabad
fYear :
2008
fDate :
19-21 Nov. 2008
Firstpage :
1
Lastpage :
6
Abstract :
This paper presents multicore implementation of an adaptive correlation tracking technique to track aerial target. The principle novelty of the proposed algorithm is its adaptiveness to changes in target size, illumination and target occlusion, etc during tracking. Tracking window size is adjusted according to the target size and tracking window is displaced depending on the tracking error generated during the course of tracking. Tracking correctness is achieved by estimating a confidence value for updating the template imagery periodically. During occlusion, target future positions are predicted by applying inverse exponential based prediction algorithm. The proposed tracking scheme is implemented on multi-DSP platform and fast tracking is accomplished at a rate of 100 frames / sec on image size of 640 times 512 pixel resolutions.
Keywords :
digital signal processing chips; image sequences; multiprocessing systems; object detection; target tracking; adaptive correlation tracking technique; aerial target tracking error; image sequence; inverse exponential based prediction algorithm; multicore DSP implementation; object tracking; template imagery occlusion; Digital signal processing; Electronic mail; Image sequences; Layout; Lighting; Multicore processing; Navigation; Prediction algorithms; Robustness; Target tracking; histogram equalization; image binarization; inverse exponential; multiprocessor; prediction; target occlusion; threshold value; track error; tracking window;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
TENCON 2008 - 2008 IEEE Region 10 Conference
Conference_Location :
Hyderabad
Print_ISBN :
978-1-4244-2408-5
Electronic_ISBN :
978-1-4244-2409-2
Type :
conf
DOI :
10.1109/TENCON.2008.4766773
Filename :
4766773
Link To Document :
بازگشت