动态场景下运动目标检测方法研究

Research on Detection Methods of Moving Objects in Dynamic Scenes

动态场景下运动目标的检测方法是计算机视觉领域的热门研究课题,本文将对动态场景下的运动目标检测方法展开研究。首先,对比分析相邻帧差分法、背景差分法、统计模型法这3种检测方法的优劣,提出时空融合补偿差分与目标平滑模型相结合的新型运动目标检测方法。其次,研究新型运动目标检测方法中的运动估计与运动补偿技术。最后,对新型运动目标检测方法进行仿真验证,对比分析实验结果,以判断方法的可靠性、准确性。实验表明,本文提出的新型运动目标检测方法,能够满足检测要求,目标轮廓清晰、完整,检测准确性较高。

The research is carried out on the detection methods of moving targets in dynamic scenes,a popular research topic in the field of computer vision.In the wake of comparison and analysis of pros and cons of such three detection methods as adjacent frame difference,background difference method and statistical model method,there's a new moving target detection method by combining time-space fusion compensation difference with target smoothing model.Then the motion estimation and motion compensation of such method will be further studied.Finally,the experimental results from its application are analyzed to prove the reliability and accuracy of the research plan.The experimental results show that the new moving target detection method proposed in this paper meets the detection requirements by presenting a clear and complete object contour with satisfactory detection accuracy.