期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2006
卷号：6
期号：12
页码：28-34
出版社：International Journal of Computer Science and Network Security
摘要：This paper presents a new framework for signboard recognition of signboards under adverse conditions. The framework is composed of two main stages newly proposed here: the main peak finding algorithm based on HSV color space to remove regions of occlusion and reflection, and the template matching to recognize objects. In the experiments, we used 300 images taken under a great variety of adverse conditions including occlusion, reflection, specular highlights, and so on. According to the experimental results, the proposed system could give 95% recognition rate for images taken under such bad conditions. The results are shown to demonstrate the robustness and effectiveness of the system. Moreover, the proposed method is applicable for vision-based car and pedestrian navigation assistance to provide up-to-date visual information in accordance with users’ demands, for example, for finding shops, billboards, landmarks and so on.
关键词：Signboard, Uniform color region, Occlusion and reflection, Main peak finding, Template matching
