運用邊緣流量及蟻群最佳化於視訊中移動物件之主 動式輪廓追蹤研究

Active Contour Tracking of Moving Objects Using Edge Flows and Ant Colony Optimization in Video Sequences

賴東賢(TUNG-HSIEN LAI)

指導教授 ： 張元翔

中原大學/工學院/資訊工程研究所/碩士(2008年)

https://doi.org/10.6840/cycu200900630

摘要

物件的分割和追蹤在視訊應用中是相當重要的技術。本論文提出一個視
訊中移動物件的主動式輪廓追蹤系統，其方法包含前處理及物件輪廓分割兩
部分。前處理的目的在於擷取物件初始輪廓，物件輪廓分割的目的則在於使
移動物件的輪廓更貼近物件真實的邊界。本系統結合邊緣流量及蟻群最佳化
演算法以增進系統輪廓能量的收斂效率。經過實驗結果驗證，系統自動產生
的分割結果與手動分割產生的結果，其平均誤差已達到小於一個像素距離之
精確度。總結而言，本系統特別適用於動態物件分割與追蹤，且不需在場景
中建立背景模型。預計本系統將可以應用在物件導向的視訊編碼，或是其他
研究如視訊監控系統的行為分析等。

關鍵字

物件追蹤 ； 邊緣流量 ； 蟻群最佳化 ； 主動式輪廓模型

並列摘要

Object segmentation and tracking are important techniques in video
applications. In this paper, we present a novel system for active contour tracking
of moving objects in video sequences. Our method includes preprocessing to
identify an initial object contour, and object contour segmentation to refine the
contour of the moving object. The edge flows and ant colony optimization are
incorporated to improve the efficiency during system convergence. Experimental
results demonstrated that our system has achieved the automatic segmentation
accuracy of < 1 pixel on average as compared with manual segmentation results.
In summary, our system is particularly useful in segmenting and tracking a
moving object without constructing a background model for a video scene.
Ultimately, our system could be used in object-based video coding or other
analysis such as behavior analysis in video surveillance systems.

並列關鍵字

edge flow ； object tracking ； ant colony optimization ； active contour model

參考文獻

1. Xiang, T., Gong, S.: Video behavior profiling for anomaly detection. IEEE

Trans. Pattern Anal. Mach. Intell. 30, 893--908 (2008)

Mach. Intell. 30, 493--506 (2008)

3. Sundaramoorthi, G., Yezzi, A., Mennucci, A.C.: Coarse-to-fine

segmentation and tracking using sobolev active contours. IEEE Trans.