Earticle

초록

영어

The purpose of the real-time monitoring of target or incident is to have a timely alarm, to find the invasion of effective goal is the core technology. Distance measuring of intrusive targets is important for indoor monitoring, according to the detected intrusion target image, calculating the distance between the target and the camera, it can provide the basis for the computer to determine the type of invasion, and finally provide information for the security alarm. In order to meet the needs of the indoor environment monitoring, the paper hereby presents a set of binocular vision-based motion detection system. The system is divided into two parts, one part is offline training, and another is online computation. The offline part consists of stereo calibration module and revised distance parameter module. The online part consists of data acquisition and storage module, moving target detection module, the corresponding point matching module and ranging module. In Stereo calibration module, the internal and external parameters of two cameras are obtained. In motion detection module, for dual-channel video capture respectively, the background subtraction algorithm is used for object detection, and a mixed Gaussian model is used as the adaptive background updating method. Then based on static camera and fixed position, objects in binocular images obtained by binocular camera will be coarse matched firstly, then will be fine matched using scale invariant features transform algorithm to find the accurate match points under this system. In the mean time, revised distance parameter module provides a correction parameter for the traditional binocular parallax distance formula. Finally, the distance between the target and the camera is calculated. Experiments show that this system can be very effective in extracting moving targets, getting match points then measuring distance, especially in distance measuring of obstructed targets.

목차

Abstract
 1. Introduction
 2. Stereo Calibration Module
 3. Ranging Module
 4. Point Matching Module
  4.1. Scale Invariant Features Transform Algorithm [12]
  4.2. Rough Match Region Selection
 5. Experimental Results
 Acknowledgements
 References

키워드

저자

• BI Ping [ School of Electronics Engineering, Xidian University, School of Telecommunication and Information Engineering, Xi’an University of Posts and Telecommunications ]

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Hybrid Information Technology

• 간기

  격월간

• pISSN

  1738-9968

• 수록기간

  2008~2016

• 십진분류

  KDC 505 DDC 605