Intention estimation of a walker around pedestrian lights by using fuzzy rules

We propose a method to judge a walker´s intention around pedestrian lights by using fuzzy rules. We detect pedestrian object in a movie of crosswalk area by using the code book method and acquire contour information. To improve the processing speed in this stage, we use parallel processing technique based on CUDA (Compute Unified Device Architecture). Now, we remove shadow which causes shape distortion of objects. Shadow removed object is judged by using the Adaboost algorithm whether to human or noise. If the objects are recognized as a human, we analyze pedestrian objects´ movement, face area feature, waiting time to judge whether he has intention to across a crosswalk by using Fuzzy rules. Traffic signal can be controlled after the judgment.