Discriminant splitting of regions in traffic sign recognition

Abstract

Mining discriminative spatial patterns in image data is a subject of interest in traffic sign recognition. In this paper, we use an approach for detecting spatial regions that are highly discriminative. The main idea is to search the normalized size blobs for discriminative regions by adaptively partitioning the space into progressively smaller sub-regions. Thus, each cluster of signs is characterized by an unique region pattern which consists of homogeneous and discriminant 2-D regions. The mean intensities of these regions are used as features. To evaluate the discriminative power of the attributes corresponding to detected regions, we performed classification experiments using a classifier based on Support Vector Machines. The proposed method has been tested in a real traffic sign database. Results demonstrate that the method can achieve a considerable reduction of features with respect to extraction from raw images while maintaining accurate. © 2013 Springer-Verlag.