Recognition of Blue Traffic Signs Enhanced by Aggregating Fragmentarily Segmented Sign Images

Mohammed Hayyan Alsibai and Yuzo Hirai


Intelligent Transportation Systems, Blue Traffic Sign Recognition, Pattern Recognition


This paper proposes a new algorithm for aggregating fragmented images to enhance the recognition of blue traffic signs designating directions. The system is designed for the assistance of safe driving. Subspace method and Histogram of Oriented Gradients (HOG) features are used for classification. More than one subspace for each arrow class is used to cover complex distributions of arrow patterns. The system consists of several processing steps including labelling the blue objects in each frame, segmenting them from the background and classifying them by subspace method. One of the difficulties in this process is the fragmented detection of single signs. Direction signs consist of a white arrow surrounded by a blue background. Due to motion blur or some lighting conditions, the white area can spread into the blue area and blue background of the sign can be disconnected after the labelling step. To aggregate the broken backgrounds of the signs a new algorithm is designed by taking into account the shape information of the signs and arrows.

Important Links:

Go Back