License Plate Localization With Efficient Markov Chain Monte Carlo

	License Plate Localization With Efficient Markov Chain Monte Carlo
	Lijun, Cao; Xu, Zhang; Weihua, Chen; Kaiqi, Huang
	2014-06
会议日期	2014-6
会议地点	厦门
关键词	License Plate Localization Feature Likelihood Mcmc Proposal Probability
英文摘要	This paper presents a novel efficient Markov Chain Monte Carlo (MCMC) method for License Plate (LP) localization. The proposed method formulates the LP image feature and prior knowledge into a unified Bayesian framework. Then the localization problem is derived as a maximizing-a-posterior (MAP) problem, which integrates color, edge and character feature of LP. We propose an efficient MCMC method, taking integrated local geometrical likelihood as proposal probability to make the inference feasible. The experimental results on real dataset are very promising in terms of detection rate and localization accuracy.
会议录	Proceeding of International Conference on Internet Multimedia Computing and Service
内容类型	会议论文
源URL	[http://ir.ia.ac.cn/handle/173211/11838]
专题	自动化研究所_智能感知与计算研究中心
通讯作者	Lijun, Cao
作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	Lijun, Cao,Xu, Zhang,Weihua, Chen,et al. License Plate Localization With Efficient Markov Chain Monte Carlo[C]. 见:. 厦门. 2014-6.