一种基于图像的运动物体检测及高度分析算法

现代监控系统要求做到无人值守，而智能视觉监控能很好满足这个要求。文中为实现这种智能视觉监控系统做了一些算法上的研究工作：实现了对运动物体进行实时检测和追踪，并利用透视几何中的消影点、消影线以及交比的约束方法对运动物体高度进行了分析。试验结果表明，此方法将自适应阈值应用到物体检测中使检测结果更加精确，并且自动估算出的物体的高度误差也很小，具有一定的实用性。

A Moving Object Detection and Height Analysis Based on Image

Modern surveillance system requires nobody to be on duty, and intelligence vision surveillance can do it. This paper has done some studies on arithmetic for this system: presents a method of detecting and tracing moving objects in real time, and analyses their height using the restriction of vanishing lines, vanishing points and cross-ratio in perspective geometry. The result of the experiment indicates that it is more accurate when have applied the adaptive threshold to detect moving object, and the error of the height estimated automatically is also small. There is some practicability.