中文摘要：

针对红外视频图像的特点,提出了一种基于期望最大化算法的红外扩展目标鲁棒自动跟踪方法。首先利用局部Top-Hat形态学滤波进行背景抑制和去噪,并通过平台直方图技术突出跟踪区域的红外目标灰度信息;然后以考虑了像素空间位置信息的高斯加权直方图建立目标的灰度特征模板;最后通过期望最大化迭代计算来估计出各密度分布的最大似然函数的参数集,并由此确定跟踪目标的位置和形状尺寸。实验表明,该方法不仅实现了跟踪窗口随目标尺寸的自适应变化,而且有效克服了红外图像信噪比低的缺点,提高了红外目标实时跟踪的稳健性。

英文摘要：

A robust and automatic tracking scheme for moving extended object in infrared video images based on expectation maximization （EM） algorithm was proposed. Firstly,local morphological Top-Hat operator was adopted to restrain the image background and eliminate the noises,and the grey level of infrared objects became more salient using the platform histogram technique. Secondly,the gray level feature template of the infrared object was built based on the Gaussian weighted histogram which included gray pixel position. Finally,the localization and shape size of infrared object in every frame were obtained with the EM algorithm,which was an iterative numerical tool for computing the parameter set of maximal likelihood function of density distribution. The experiments demonstrate that the proposed method can track the infrared object with adaptive bandwidths and enhance SNR of the infrared image. It efficiently guarantees the accuracy,stability and continuity in real time tracking infrared object.