基於人工蜂群演算法之物件追蹤研究

Objects Tracking Based on Artificial Bee Colony Algorithm

蔡尚麟(Shang-Lin Tsai)

指導教授 ： 鍾鴻源

國立中央大學/資訊電機學院/電機工程學系/碩士(2015年)

摘要

近年來，隨著攝影機與監視器的普及，影像追蹤成為了一個熱門的議題。為了提升追蹤目標物的精準度和解決目標物遮蔽的問題，本論文採用人工蜂群(Artificial Bee Colony; ABC)演算法來對目標物進行即時追蹤。
在偵測目標物這方面，本論文採用了背景相減法，因其可以切割出完整的目標物體且運算量低，容易運用於即時系統中。再來則是利用改良的種子區域生長法來標記各個目標物，區分出各個目標物後，再計算出各個目標物的中心位置。接著對各個目標物建構顏色直方圖模型以便做追蹤使用，追蹤過程中很容易受到光線變化影響，本文是採用HSV色彩空間中的色相，去掉了亮度的影響可以成功降低光線變化所造成的干擾。物件追蹤則是利用了人工蜂群演算法來尋求最佳解，擁有結構簡單、容易使用及收斂速度快等特性。遮蔽物問題一直以來都是物件追蹤的一個問題，為了有效的解決遮蔽物問題，本論文使用了一個可以調整搜尋框大小的機制，在追蹤不到目標物的時候放大搜尋框來增加搜尋範圍，追蹤到目標物的時候又能將追蹤框調整回原來的大小。

關鍵字

人工蜂群演算法 ； 物件追蹤 ； 物件偵測 ； 種子區域生長法

並列摘要

In recent years, as cameras and monitors become more and more popular, object tracking becomes a hot issue. In order to improve the accuracy of the tracking object and solve the occlusion problem, in this thesis, the Artificial Bee Colony (ABC) algorithm is used for object tracking in real time.
In terms of object detection, in this thesis, the background subtraction is used for it can cut out complete targets, has low computation and be easily applied to real-time systems. Besides, the improved seed region growing method is used to distinguish every target and calculate its center. Then, for model building, color histograms are used to build target models. In order to avoid the interference of light, in this thesis, the HSV (Hue, Saturation and Value) color space is used. Moreover, for object tracking, in this thesis, the ABC algorithm which has a simple structure is used to find the best solution for it is easily used and its convergence is fast. Occlusion is always a big problem for object tracking. Therefore, in this thesis, the adaptive searching window is applied to exclude occlusion; the searching window will zoom in or out, depending on its fitness value. If the tracking window loses the targets, the searching window will increase. If the tracking window finds the targets, the searching window will adjust to the original size.

並列關鍵字

ABC ； Object tracking ； Object detection ； Seeded region growing

參考文獻

[1]G. Paravati, A. Sanna, B. Pralio and F. Lamberti, “A Genetic Algorithm for Target Tracking in FLIR Video Sequences Using Intensity Variation Function,” IEEE Transactions on Instrumentation and Measurement, vol. 58, pp. 3457-3467, 2009.

[3]X. Dong, J. Cao, H. Yang, Z. Yu, H. Guo and C. Liu, “Object Tracking based on integrating the Genetic algorithm with complex method,” International Conference on Intelligent Control and Information Processing, pp. 205-209, 2013.

[5]T. Kobayashi, K. Nakagawa, J. Imae and G. Zhai, “Real Time Object tracking on Video Image Sequence using Particle Swarm Optimization,” International Conference on Control, Automation and Systems, pp. 1773-1778, 2007.

[7]Y. Zheng and Y. Meng, “Adaptive Object Tracking using Particle Swarm Optimization,” International Symposium on Computational Intelligence in Robotics and Automation, pp. 43-48, 2007.

[8]Z. Hao, X. Zhang, P. Yu and H. Li, “Video Object Tracing Based on Particle Filter with Ant Colony Optimization,” International Conference on Advanced Computer Control, vol. 3, pp. 232-236, 2010.