低视角红外图像行人监控的光流分析

光流法是运动目标检测的重要方法，在对运动目标参数的估计和跟踪方面有着重要的应用。红外图像的噪声相对比较大，对比度较低，光流法很少用于红外图像的目标的运动检测，特别是行人的检测。本文将光流法用于红外图像的行人监控，检测行人运动速度的大小和方向，并针对低视角情况下存在的由远近效应造成的误差，提出了一种低视角理想情况下行人光流分析的补偿方法，并对实验数据进行了测试，实验结果证明经该方法改进后的光流更符合实际情况。

Optical Flow Analysis for Infrared Pedestrian Monitoring in Low View

Optical Flow is a most important method for moving object detection,and most usually used in parameter evaluation and object tracking. Infrared images always have large noises and low contrast,so few people use optical flow for moving object detection in infrared images,especially for pedestrian detection.This paper uses optical flow to detect people's moving speed and direction by using infrared images.Because in low angle situation, the results of optical flow always have near-far effect and the results have large errors.So,this paper proposes a method to reduce this kind of errors caused by near-far effect,and we have also done some tests on the acquired infrared images.The results showed this method had a good performance and the optical flow results were more accurate after compensation.