兩段式背景相減法偵測移動物體

Moving Object Detection Based on Two-Staged Background Subtraction Approach

蘇恕德(Shu-Te Su)

指導教授 ： 陳永耀

國立臺灣大學/電機資訊學院/電機工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.01779

摘要

本論文提出兩段式背景相減法偵測移動物體。在靜止的場景中使用背景相減法來偵測移動物體是常用的方法，然而此方法有許些問題需要解決，例如：場景中有光線的變化、新物體加入背景當中、場景或攝影機產生雜訊…等，甚至還有移動物體與背景有相近的顏色而導致偵測錯誤的情況。為了解決上述問題，本論文提出兩段式背景相減法偵測移動物體。第一階段是利用改善式移動平均背景建立背景與背景相減，改善式移動平均背景可以改善上述前三項問題。第二階段是修正第一階段的偵測錯誤，利用移動物體候選區的權重圖重新背景相減，即可改善上述的最後一項問題。最後，兩段式背景相減法有快速的計算效率、低記憶體需求、移動物體偵測的高準確率，同時亦能解決移動物體和背景顏色相近時的偵測錯誤。

關鍵字

移動物體偵測 ； 背景相減法 ； 背景更新 ； 自動取閥值演算法 ； 前景分析 ； 陰影偵測

並列摘要

Moving object detection based on two-staged background subtraction approach is proposed in this thesis. Background subtraction in image sequence is a popular approach for detecting moving objects, especially in a relatively static scene. However, there are some problems for background subtraction approach, such as the scene with varying illumination, the new static object in the background, and the captured frame with noise that caused by environment or camera. Moreover, sometimes the results in moving object detection have false detections due to near color in corresponding pixels of moving object and background. To solve above-mentioned problems, Two-Staged Background Subtraction (Two-StaBaS) approach is then proposed. The first stage is background modeling and background subtraction, the suggested background is Improved Running Average Background (IRAB). IRAB shows its improvements on the first three above-mentioned problems. The second stage is the modification of detection errors in the first stage, namely, it solves the near color problem. Background subtraction based on the weighting map in moving object candidate region, a number of false detections in background subtraction approach decreases. Finally, two-staged background subtraction approach has the advantages of fast computational speed, low memory requirement, and good accuracy for detecting moving objects.

並列關鍵字

moving object detection ； background subtraction ； background update ； auto-thresholding algorithm ； foreground analysis ； shadow detection

參考文獻

[1]. S.T. Su, Y.Y. Chen, "Moving Object Segmentation Using Improved Running Gaussian Average Background Model," Digital Image Computing: Techniques and Applications, pp. 24-31, 2008.

[2]. C.R. Wren, A. Azarbayejani, T. Darrell, A. P. Pentland, "Pfinder: Real-Time Tracking of the Human Body," IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 19, No. 7, pp. 780-785, 1997.

[3]. C. Stauffer, W.E.L. Grimson, "Learning Patterns of Activity Using Real-Time Tracking," IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 22, No. 8, pp. 747-757, 2000.

[4]. N.M. Oliver, B. Rosario, A.P. Pentland, "A Bayesian Computer Vision System for Modeling Human Interactions," IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 22, No. 8, pp. 831-843, 2000.

[5]. K. Toyama, J. Krumm, B. Brumitt, B. Meyers, "Wallflower: Principles and Practice of Background Maintenance," IEEE Inte'l Conf. on Computer Vision, Vol. 1, pp. 255-261, 1999.