基于区域协方差矩阵的末制导目标跟踪

Terminal guidance target tracking based on region covariance matrix

针对红外成像末制导阶段目标尺寸和姿态迅速变化的问题,提出了一种基于目标区域协方差矩阵的粒子滤波器目标跟踪方法。为了在红外成像末制导阶段精确跟踪目标以及抑制逐渐显现的图像细节,本方法以目标位置和大小作为目标状态,通过粒子滤波器跟踪在实时帧图像中的目标,并在图像金字塔上迭代定位目标以提高跟踪精度。为了适应目标大小的变化,通过实时更新目标模板(区域协方差矩阵)以调整跟踪窗的大小。选择多种目标特征构造区域协方差矩阵,作为目标描述符来匹配、定位目标。仿真结果表明,该方法不仅能适应末制导阶段目标尺寸的急剧变化,而且对红外成像末制导阶段目标能实现精确跟踪,表现了很好的鲁棒性。

For the problem of rapid change in the size and pose of a target during infrared imaging terminal guidance, the region covariance matrix based particle filter is proposed to precisely track a target in this paper. In order to suppress image details and precisely track a target in the infrared imaging terminal guidance, the position and size of a target are served as the target state which is recursively estimated by the particle filter in the real-time frames, and the tracking precision is improved by the iterative location of a target in the image pyramid. For adapting the variable size of a target, the proposed approach can adjust the size of tracking window by timely update the target template （region covariance matrix）. The region covariance matrix as the target feature discriptor consists of multiple image features of a target to locate the frame-to-frame target by feature matching. Experimental results demonstrate that the approach can adapt rapid change in the size of target and presicely track target during infrared imaging terminal guidance, and has the good robustnes.