車輛追蹤系統之自動化必須要克服的條件 非常眾多,本文提出新的辨認參數即是色彩,其 中包括有R(紅)、G(綠)、B(藍)以及(R-B)之影響,建 立了一不相似性指標做為計算辨認之門檻,另 外本文也提出多邊形視窗法以及Subostu三值法 解決實際影像的複雜度,由於多目標物體追蹤 尚是努力的領域,本文在二又二分之一D的側向 偵測的試驗中有較佳之效果。 An automatic vehicle tracking system has many important elements including preprocessing, detection, segmentation, features catching and targets recognition. This article adopts color features as the parameters of recognition and builds a dissimilarity index (D.I.) with the value of red (R), green (G), blue (B) and red minus blue (R-B). We try to use heuristic value(1200) as the threshold of recognition. Another we also try to establish two new segmentation processes, polygon window method and sub-ostu segmentation method, and we successfully segment the vehicle images. The study has 92. 41% recognition rate, on 2+(1/2)D condition.
Relation:
第六屆全國自動化科技研討會論文集(二)=Proceedings of the Sixth National Conference on Automation Technology (V olume=Two)，頁611-618
