Image based roadside POI recognition and GPS correction scheme using smart phones

In recent years, advanced driver assistance system (ADAS) and intelligent traffic system (ITS) both are popular issues. This paper presents an efficient and accurate roadside sign recognition which can be used to collect traffic sign and point of interest (POI) information by any end-users with smart phones or driving video recorders. In order to achieve high recognition performance for the proposed scheme, the local binary pattern (LBP) operator is used for feature extraction with low computing complexity scheme, and the support vector machine (SVM) is also used to learn and classify roadside signs. Besides, this paper also utilizes an image based distance measurement method to obtain the distance between roadside signs and cameras by single camera view, thus the proposed scheme can be applied to general electronic devices with single camera such as smart phones, tablets, driving video recorders. The POI and distance information can be applied in ADAS, intelligent traffic system, navigation system, and GPS correction. Experimental results show the average accuracy of recognition for each roadside sign is 98.1873%, and the average recognition speed is less than 20 milliseconds per each sign. Also, the distance measurement method is workable to correct the GPS positions of POIs or users.