結合前景抽取及人形偵測之視訊監測分析

Video Surveillance Analysis Based on Combining Foreground Extraction and Human Detection

曾惠淇(Hui-Chi Zeng)

指導教授 ： 賴尚宏

國立清華大學/電機資訊學院/資訊工程學系/碩士(2007年)

摘要

在這篇論文中，我們提出了一個在即時視訊監測上的前景抽取，結合使用AdaBoost機器學習演算法和HOG當特徵的人形偵測技術，並且利用以RANSAC為基礎的在時間軸上的追蹤理論來提高準確率。一般傳統的移除背景方法在有光影變化的情況下通常無法運作的很好。我們提出使用一個二步驟的前背景分離理論來移除背景以及那些受到陰影、自動白平衡和突然的光影變化所影響的背景。在前景抽取及人形偵測之後，我們利用時間軸上的資訊來提高人形偵測的準確率，使用以RANSAC為基礎的在時間軸上的追蹤理論來移除掉誤判及修復遺失的偵測結果。在一些真實的監視器影片上作的實驗結果證明，我們提出的前景抽取理論在各種有光影變化的不同環境下，以及人形偵測理論，都能有很好的性能。

關鍵字

監視器系統 ； 前景抽取 ； 人形偵測 ； 視訊監測

並列摘要

In this thesis, we present an adaptive foreground object extraction algorithm for real-time video surveillance, in conjunction with a human detection technique applied in the extracted foreground regions by using AdaBoost learning algorithm and Histograms of Oriented Gradient (HOG) descriptors. Furthermore, a RANSAC-based temporal tracking algorithm is also applied to refine and trace the detected human windows in order to increase the detection accuracy and reduce the false alarm rate. The traditional background subtraction technique usually cannot work well for situations with lighting variations in the scene. The proposed algorithm employs a two-stage foreground/background classification procedure to perform background subtraction and remove the undesirable subtraction results due to shadow, automatic white balance, and sudden illumination change. After foreground extraction and human detection, the temporal information is utilized to increase the detection accuracy by performing the RANSAC-based temporal tracking to remove the false alarms and recover the missed detections. Experimental results on some real surveillance video are shown to demonstrate the good performance of the proposed adaptive foreground extraction algorithm under a variety of different environments with lighting variations and human detection system.

參考文獻

[2] A. Elgammal, D. Harwood, and L. Davis, “Non-parametric model for background subtraction,” European Conference on Computer Vision, II: 751-767, June/July 2000.

[3] O. Tuzel, F. Porikli, and P. Meer, “A bayesian approach to background modeling,” IEEE Workshop on Machine Vision for Intelligent Vehicles, III:58, 2005.

[4] F. Porikli and J. Thornton, “Shadow flow: A recursive method to learn moving cast shadows,” IEEE International Conference on Computer Vision, I: 891-898, 2005.

[5] S. S. Huang, L. C. Fu, and P. Y. Hsiao, “Region-level motion-based foreground detection with shadow removal using MRFs,” Asian Conference on Computer Vision, pages 878-887, 2006.

[7] C. Benedek and T.Sziranyi, “Markovian framework for foreground-background-shadow separation of real world video scenes,” Asian Conference on Computer Vision, I:898-907, 2006.