针对成像平台运动情况下的运动目标检测问题，提出了一种从特征点稀疏运动场估计到运动分类的目标检测算法。

首先通过快速特征点检测与跟踪恢复出图像稀疏运动场；然后依据特征点之间运动一致性关系实现属于同一运动模式的特征

点分类，根据分类得到的各组特征点计算场景图像重建误差，剔除重建误差最小的特征点组，实现对前景目标的检测。仿真实

验对该算法在复杂场景中检测运动目标的有效性进行了验证。-】In order to detect target in the background motion video，an algorithm with 8parse motion field estimation，motion classification is prorMl．

Firstly，8pa_rl*motion field is recovered by fast local feature detection and吮king．The local features that belong to the same motion pattern are classified

according to their motion consistency．Then，the resulting local feature groups a_re used to reconstructed scene image，and the foreground targets aI_e identifled

by getting rid of the group with the least reconstruction en Dr．The proposed method is tested on the dataset of real complex scenarios and its effectiveness

is demonstrated in the results．