Road sign text detection from natural scenes

Abstract

Texts on road signs contain important information which is quite useful for potential applications. We proposed a robust method for detecting road sign text from urban street scenes under different weather conditions. First, color Segmentation and morphological operations are employed to obtain candidate regions, and contours of candidate regions are mainly concern. Then, a linear support vector machine (SVM) classifier is followed for shape classification after shape features based on edge orientation histogram (EOH) of contours are extracted. Finally, binarization of road sign images is achieved by k-means clustering in the S channel, multi-scale rules and strokes merging are referenced to extract texts. Experiment results on a large amount of images demonstrate the effectiveness of the proposed method.

Chinese Traffic Panels Detection and Recognition From Street-Level Images

“…as it is proposed in [9]. ( , ) S x y is the saturation component of the image and its value range is [0,1].…”

“…In (9), aspect ratio is w divided by h. The lower limit of w is independent of Wid, because it is hard to find their relation, as is shown in Fig. 5.…”