本文提出一種改進的即時追蹤演算法藉由改良凝聚式階層演算法和色彩空間，同時也結合了背景相減法、改良種子生長法和無線監控端，來實現即時追蹤自走車。在追蹤前對每個目標物進行顏色直方圖的建立與儲存，以便提升在追蹤時的速度。 由於計算機技術和半導體技術的進步，近年來處理器的速度大大提高。 越來越複雜的算法可以用於偵測或追蹤，並且目前相關的研究帶來了高效率的偵測和追蹤。科技的進步使的演算法在速度上取得了重大突破，可以實現偵測和追蹤系統。 基本上，過去幾年提出的物體跟追蹤法一般可分為以下幾類：基於區域式的追蹤，主動式輪廓追蹤，基於特徵的追蹤和基於模型的追蹤。雖然這些追蹤方法已經開發了很長時間，但仍然有許多問題需要解決。例如，複雜的背景和陰影可能會影響追蹤。本文使用改良的凝聚式階層演算法和改良顏色空間進行追蹤，可以有效地解決陰影問題。 ;This paper proposes an improved tracking algorithm by using “agglomerative hierarchical algorithm ” and “ color space ” for real-time tracking of the targets, which combines three methods to realize an automatic tracking system in real-time on wireless monitoring self-propelled vehicles. There are three methods including the background subtraction method, the seeded region growing method, and the wireless monitoring system to construct the color histogram models for each targets. Finally, the obtained image tracking pictures can be saved into flash memory to speed up loading. Due to the advance in computer technologies and the semiconductor technologies, the speed of the processors have been dramatically improved in recent years. More and more complex algorithms then can be used to detect or track objects, and the relating studies flourishes presently bringing about high efficiencies of identifying and tracking objects. It truly makes a major breakthrough in algorithms, and the real-time object tracking system can be achieved. Basically, the object tracking methods proposed in the past years can be generally divided into these categories: region-based tracking, active contour-based tracking, feature-based tracking and model-based tracking. Although visual tracking has been developed for a long time, it still has many problems which needed to be solved. For example, the complex background and the shadow may have an influence on tracking object. This paper uses improved agglomerative hierarchical algorithm and color space for object tracking that can effectively solve the shadowing problems.
