Forward looking infrared target matching algorithm based on depth learning and matrix double transformation
- 75 Downloads
To effectively reduce traffic accidents caused by night driving and provide initiative whole system for car driving in environment of lower visibility, sub-model of pedestrians at night based on far infrared sensor technology was designed according to basic requirements of driver assistant system of cars in the industry. Original data source was extracted for this model via far infrared sensor and its ROIs were obtained by using grey statistical technology. Matching detection was conducted on data source on basis of constructing multi-scale probability template, and detection rate as well as rate of leak detection of designed models could be effectively improved via comprehensive treatment technology of multi-frame verification. Experiments show that probability template of this model is improved on matching precision compared with common methods in the industry. It is applicable to two kinds of traffic road conditions of suburb and downtown at the same time, so it has good practicability.
KeywordsNight driving Automobile driver-assistance Grey statistics Multi-scale Multi-frame verification
Funding was provided by Science and Technology Project of Zhejiang Province, No. 2015C31171 and National Education Information Technology Research Plan, Project Nos. 146241806 and 136241559.
- 1.Su, J., Xu, Q.S., Liu, G.: A forward looking infrared target recognition algorithm based on edge matching. Acta Armamentarii 33(3), 271–277 (2012)Google Scholar
- 2.Li, X., Cao, Z., Yan, R., et al.: Forward-looking infrared 3D target tracking via combination of particle filter and SIFT. Proc. SPIE Int. Soc. Opt. Eng. 8918(48), 08 (2013)Google Scholar
- 5.Rizvi, S.A., Nasrabadi, N.M.: Neural network algorithms for automatic target recognition using forward-looking infrared imagery: a survey. Proc. Indian Acad. Sci. A 67(2), 243–276 (2001)Google Scholar
- 6.Borghys, D.C.J., Acheroy, M.P.J.: Long-range automatic detection of small targets in sequences of noisy thermal infrared images. Proc. SPIE Int. Soc. Opt. Eng. 2235, 4–8 (1994)Google Scholar
- 9.Jiang, H., Liao, G., Qu, Y.: Compensation of clutter spectrum for airborne forward-looking radar based on the clutter subspace transformation. In: Proceedings of the International Radar Conference Surveillance for a Safer World. IEEE, pp. 1–4 (2009)Google Scholar
- 10.Litvak, S., Brailovsky, L., Yanir, T.: Learning based estimation of hand and finger pose: US, US 20130236089 A1 [P] (2016)Google Scholar
- 11.Young, S.S., Kwon, H., Nasrabadi, N.M.: Adaptive target detection in FLIR imagery using the eigen space separation transform and principal component analysis. Proc. SPIE Int. Soc. Opt. Eng. 85(11), 26 (2003)Google Scholar
- 14.Feldman, V., Gopalan, P., Khot, S., et al.: New results for learning noisy parities and halfspaces. In: Proceedings of the IEEE Symposium on Foundations of Computer Science, 2006. FOCS ‘06. IEEE, pp. 563–574 (2006)Google Scholar