효과적인 전경객체 검출을 위한 배경분리기법과 후처리 방법
A Background Subtraction and Post-Processing for Effective Foreground Object Detection
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 논문은 배경분리기법을 이용한 객체검출과 검출된 객체에 대한 효과적인 후처리기술에 관한 연구이다. 대표적인 배경분리기법으로서 AMF(Approximated Median Filtering)는 재귀적 방법으로 처리속도와 메모리 사용에서 효율적이다. 하지만 긍정곡선영역(Positively curved region)에 대한 처리 때문에 객체내부의 손실을 발생시키고 처리속도가 느리다. 이를 극복하기 위해 본 논문에서는 수정된 AMF방법을 제안하고, 적응적 가우시안 혼합 모델(Adaptive Gaussian Mixture Model)과 고유배경모델(Eigen-background)의 성능평가를 통해 수정된 AMF의 우수성을 증명한다. 또한 정확하고 빠른 객체검출을 위해서 제안된 배경분리기법을 통해 얻은 이진 전경 분할 마스크(Binary Foreground Segmentation Mask)에 대해 효과적인 후처리 방법을 제안하고, 실험을 통해 증명한다.
Abstract
This paper is studies on object detection using background subtraction techniques and effective post-processing technique on detected objects. AMF(Approximated Median Filtering) as a typical background subtraction technique is effective to processing time and memory use by recursive technique. But it generates loss on the interior of object and processing time slow because of process of positively curved region. To overcome problems, this paper is proposed to modified AMF and we prove a superiority of modified AMF by performance evaluations of adaptive Gaussian mixture model and Eigen-background. Also, to precisely rapidly detect objects, we propose a effective post-processing method on a binary foreground segmentation mask acquired through a proposed background subtraction techniques and prove through experiments.
참고문헌 (9)
- C. Stauffer, "Adaptive background mixture models for real-time tracking", CVPR99, 통권 : 2, pp.252~ (1999)
- N. Oliver, "A bayesian computer vision system for modeling human interactions", IEEE PAMI, 22, pp.831~843 (2000)
- C. R. Wren, "Pfinder: real-time tracking of the human body", IEEE PAMI, 19(7), pp.780~785 (1997)
- Z. Zivkovic, "Efficient adaptive density estimation per image pixel for the task of background subtraction", Pattern Recognition Letters, 27(7), pp.773~780 (2006)
- N. McFarlane, "Segmentation and tracking of piglets in images", MVA, 8, pp.187~193 (1995)
- S. Calderara, "Reliable background suppression for complex scenes", VSSN06, pp.211~214 (2006)
- R. Cucchiara, "Detecting moving objects, ghosts, and shadows in video streams", IEEE PAMI, 25(10), pp.1337~1342 (2003)
- R. Achanta, "Frequency-tuned salient region detection", Proceedings of IEEE Conference on, Computer Vision and Pattern Recognition(CVPR 2009), pp.1597~1604 (2009)
- An Evaluation of Optical Flow using Lucas and Kanade's Algorithm, "Carman Neustaedter", ~ (2002)