BROWSE
Detail
Conference Paper
ROI Selection using Optical Flow for Object Detection from a Moving Vehicle
Cited - time in 
Share 
Share
- Authors
- Byoung-Kyu Dan, Yeo-Min Yun, Chun-Gi Lyuh, Tae Moon Roh, Sung-Jea Ko
- Issue Date
- 2011-01
- Citation
- International Workshop on Advanced Image Technology (IWAIT) 2011, pp.1-4
- Language
- English
- Type
- Conference Paper
- Abstract
- In this paper, we propose a region of interest (ROI) selection method for object detection from a moving vehicle. In the proposed method, the optical flow generated by the relative motion between the camera and the scene is utilized to estimate the forward and rotational motion of the vehicle. When the vehicle moves forward straightly, the optical flow vectors are converged to a certain point called the vanishing point. Considering the vanishing point, we establish a model for the optical flow with respect to the forward motion. In addition, the proposed model is extended to the rotational motion. By using the proposed optical flow model, we can find the direction of a moving vehicle and determine the ROI for object detection. Experimental results show that both the detection speed and the false positive problem are improved.
- KSP Keywords
- Detection speed, Forward motion, Moving Vehicle, Object detection, Optical flow vectors, Proposed model, Region Of Interest(ROI), Relative motion, Rotational motions, Selection method, Vanishing point