基于光流的运动小目标检测算法

运动目标的检测是目标识别与跟踪的关键技术之一。光流技术是一种以物体的运动特征来检测目标的方法，它的提出为运动小目标的检测开辟了新的空间。在一个搜索跟踪系统中使用光流技术检测和跟踪空中小目标，目标大概为5～10个像素，而且背景复杂，相机抖动，普通分割算法无法得到小目标。在目标的运动明显异于背景的情况下，通过利用基于光流的目标检测算法来检测出小目标，同时运用高斯金字塔模型，提高算法的运算速度。试验结果表明提出的基于光流的检测算法在背景运动的红外图像中取得了较好的效果。

A Small Moving Object Extraction Algorithm Based on Optical Flow

Moving object extraction is a key technology of object recognition and object tracking.Optical flow technology is an object extraction method based on movement of object.It creats a new area for small moving object extraction.We use optical flow technologe to extract and track small object in one searching and tracking system.The object covers 5~10 pixels,along with complex background and shake of camera.It is hard for common method to get small object.After using movement characteristis of the viedo,it can distinguish small object(the movement of object is different form background).At the same time,using Guass model can improve the speed of algorithm.The testing result shows that the object extraction algorithm based on optical flow has good effect when the infrared image’s background is moving.