Title :
Target Recognition by Fusing Images of Dual-Band Infrared Cameras
Author :
Wu, Qiongyu ; Peng, Chunyan ; Wang, Chengyou ; Liu, Yunhui ; Cai, Xuanping
Author_Institution :
Sch. of Electron. Sci. & Eng., Nat. Univ. of Defense Technol., Changsha
Abstract :
The paper presents a new algorithm for target recognition by fusing images of dual-band infrared cameras. The algorithm extracts a collection of features, which are invariable in translation, rotation and expansion, using the 2-D Fourier wavelet descriptor. The target is first recognized in each band of the images using the extracted features and then the recognition is refined by fusing the results and reliability of the bands using the D-S evidence theory. Experimental results show that the 2-D Fourier wavelet descriptor is robust in extracting invariants and the fusion algorithm proposed can improve the recognition rate from 82% to 97%
Keywords :
Fourier transforms; feature extraction; inference mechanisms; infrared imaging; object recognition; wavelet transforms; 2D Fourier wavelet descriptor; D-S evidence theory; dual-band infrared cameras; feature extraction; fusing images; target recognition; Cameras; Dual band; Feature extraction; Image recognition; Infrared imaging; Infrared surveillance; Noise robustness; Noise shaping; Shape; Target recognition; D-S Evidence Theory; Dual-Band Infrared Camera; Target Recognition;
Conference_Titel :
Mechatronics and Automation, Proceedings of the 2006 IEEE International Conference on
Conference_Location :
Luoyang, Henan
Print_ISBN :
1-4244-0465-7
Electronic_ISBN :
1-4244-0466-5
DOI :
10.1109/ICMA.2006.257800
