摘要
彩色的視覺技術雖然擁有更豐富的資訊量，卻面臨了更高的複雜度與運算速度的需求。非常幸運的，近來高速低價中央處理器(CPU)與高解析度彩色CCD的普及，促成了彩色機械視覺的有利環境。本文決定利用這種優勢結合舊有的相關視覺技術、色彩學與自行開發的技巧為機械視覺貢獻一份心力。 本文運用平行結構的彩色雙攝影機達到了以色彩特徵為基礎之彩色物件三度空間移動路徑偵測與追蹤。為了達到這個目的我們在色彩學與立體視覺上做了兩項貢獻，在色彩學上首創RGB與HSI融合的色彩模型概念，應用此模型開發快速色彩量化分割的方法，使電腦能更快速的判定物件顏色；在立體視覺上重新定義與運用Andreas etc.所提出的區塊比對(Block Matching)演算法，摒棄以往規避態度應用彩色影像優勢直接面對克服左右影像對應問題(correspondence problem)，並將此技巧應用在追蹤問題上。 在其他相關整合技術上，本文提供了改良的Sequential Algorithm區域編碼技巧、前置位置估測器、立體空間幾何概念與一些實驗時設備的校正技巧和相關細節，最後並進行實驗證明理論的可行性與探討其優缺點。
並列摘要
Color vision technique has more adequate image information, but confronting more complexity problems and more calculate cost. Fortunately, height speed CPU and height quality color CCD are popular and inexpensive. This makes a well chance for machine vision. We decide to use the advantage with traditional vision technique, chromatics and develop some novel methods to devote myself for machine vision. In this thesis, we achieve the goal of 3D target tracking based on color stereo binocular vision. For this goal, we present two contributions in chromatics and stereo vision. In chromatics, we develop a novel color solid using RGB model and HSI model fused. This makes that computer can tell color region faster and stronger. In stereo vision, we redefine Block Match method that presented by A. Koschan 1996. This solved the correspondence problem directly and even tracking problem. We improve Sequential Algorithm for labeling color region, offering the target position prediction method, conception of stereo geometry and some technique to correct vision system. Finally, we have experiments to make sure that my vision system was working and discussing the problems in my system.
參考文獻
[1] W. E. L Grimson. From Images to Surfaces: A Computational Study of the Human Early Vision System. MIT Press, Cambridge, 1981.
[2] D. Marr. Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. W. H. Freeman and Company, San Francisco, 1982.
[4] T. Kanade Y. Otha, “Stereo by intra- and inter-scanline search using dynamic programming”, IEEE Trans. on PAMI, Vol.PAMI-7, No2, 139-154,1985.
[5] S.T. Barnard, “Stochastic Stereo Matching on the Connection Machine”, Proc. Image Understanding Workshop, 1021-1031, 1989.
[6] G. Xu, S. Tsuji, M. Asada, “A motion stereo method based on coarse-to-find control strategy”, IEEE Trans. on PAMI, Vol.PAMI-9, No2, 332-336,1987.