モフォロジー演算と相対色によるセグメンテイションを用いたロバストな交通標識の認識

概要

論文の詳細を見る
We propose a robust road sign recognition system under various illumination conditions. The proposed approach has two steps: segmentation and recognition. The segmentation, which is the focus of this paper, is performed using morphological operations and relative color. The segmented regions are recognized by a template matching method using modified standard deviation. The algorithm works for various types of circular and pentagonal road signs. In experiments under various illumination conditions, the segmentation rate was 100% in the daytime and evening and 80% even in the night-time, and the recognition rate was 100% for all of the segmented regions under all illumination conditions. The effectiveness of the proposed system was confirmed through experiments using 200 images of road signs taken under a great variety of illumination conditions including fog and light rain.
社団法人映像情報メディア学会の論文
2005-09-01