【摘要】 运动目标状态分析已经成为计算机视觉研究的焦点问题,该文在介绍了运动目标成像系统的基础上,提出了基于帧间差分和差分背景图像信息融合的自适应背景模型以适应外界环境的变化,将α-β滤波的预测信息来提高聚类分析的实时性,同时将聚类结果来预测运动目标在下一帧图像的位置,根据运动的平滑性进行帧间匹配,从而有效地获得多目标轨迹。更多还原

【Abstract】 The paper introduces the system to capture image of moving object,advances adaptive background image model which based on information fusion difference frame-to-frame image with difference image to adapt variational environment.It improves real-time of clustering with the forecast value of α-β filterand,and uses the result of clustering to forecast the object position in the next image.The trajectories matching in frame-to-frame image based on motion smoothness,so the trajectories of the multi-object can be obtained effectively.更多还原