SEMI-SUPERVISED SPECTRAL CLUSTERING ENSEMBLE FOR TRAFFIC SIGN CATEGORIZATION

摘要

Traffic sign recognition is a significant part in advanced driving assistance system. In this paper, a semi-supervised spectral cluster ensemble model is designed for traffic sign recognition, and the spectral model is based on pairwise constraints. Then, the inference of the proposed model is illustrated in detail and the corresponding algorithm is stated step by step. At last, real datasets are selected for experiment and the experimental results show that the proposed algorithm can work well.