Anti-shielding target tracking method based on multi-example learning and kernel correlation filter

The invention relates to an anti-shielding target tracking method based on multi-example learning and a kernel correlation filter, and belongs to the technical field of target tracking. The method comprises the step of jointly judging whether a target is shielded or not by combining the shielding degree predicted by multi-example learning and a shielding detection mechanism constructed based on akernel correlation filter; when the target is shielded, activating a re-detection mechanism to re-search the target, and pausing updating of the tracker to prevent the tracker from being interfered bya shielding object; and finally constructing a scale filter to determine the scale of the target. By implementing the method, the tracking method can be excellent in performance in a shielding sceneand meet the real-time requirement. 本发明涉及基于多示例学习和核相关滤波器的抗遮挡目标跟踪方法，属于目标跟踪技术领域。该方法包括：联合多示例学习预测的遮挡程度和基于核相关滤波器构建的遮挡检测机制共同判断目标是否发生遮挡。当目标发生遮挡时，激活重检测机制重新搜索目标，同时暂停更新跟踪器，防止跟踪器被遮挡物所干扰。最后构建尺度滤波器确定目标的尺度。实施本发明，能够使跟踪方法在遮挡场景下表现优异并且满足实时性要求。