论文部分内容阅读
大多数基于背景差的运动目标检测方法,主要运用背景图像与当前帧图像之差进行目标检测和提取,但对背景的实时更新和场景中的光线明显变化等情况不能很好的处理。本文结合Surendra背景更新算法和动态阈值背景差算法,给出了一种新的运动目标实时检测算法。首先采用Surendra方法动态更新背景,然后使用Ostu算法计算出的阈值与一个反映光线变化的增量之和为阈值实时检测运动目标。该算法既可以稳定地对背景进行实时更新,又可以适应场景光照变化的情况。
Most of the moving objects detection methods based on poor backgrounds mainly use the difference between the background image and the current frame image to detect and extract the target. However, the real-time updating of the background and the obvious change of the light in the scene can not be handled well. Based on Surendra background updating algorithm and dynamic threshold background difference algorithm, this paper presents a new real-time moving object detection algorithm. First, the background is dynamically updated using the Surendra method, and then the moving target is detected in real time using the sum of the threshold calculated by Ostu algorithm and an increment reflecting the change of light. The algorithm can not only stabilize the background in real time, but also adapt to the change of scene illumination.