机载复杂遥感场景下特定建筑区检测跟踪算法

发布时间：2018-03-01 17:36

本文关键词： 机载视频跟踪建筑区检测局部描述子复杂场景　出处：《电子学报》2016年06期 　论文类型：期刊论文

【摘要】：特定建筑区域的自动检测及跟踪在机载遥感视频数据处理中具有重要意义,是非悬停机载平台面向任务对地搜寻及火控系统引导的关键技术之一.针对实际复杂应用场景,提出一种特定建筑区检测跟踪算法.首先,将离线获得的目标区参考图进行局部描述子特征提取.目标区在线检测阶段,为保证时效性,提出基于边缘显著性的快速SIFT(Scale Invariant Feature Transform)特征提取方法,对待检测大视场图像中疑似建筑区域提取局部描述子特征;并通过设计层次化的特征点匹配方法实现目标区的高可靠检测.目标区在线跟踪阶段,通过自适应开窗及轨迹预测技术有效圈定疑似目标区,并在圈定的有限范围内进行最终目标确认及稳定跟踪.实测数据仿真结果表明,本文提出的算法能在复杂场景条件下,实现对特定建筑区域的快速检测及稳定跟踪.可为将来实际系统应用提供关键技术支撑.
[Abstract]:Automatic detection and tracking of specific building areas is of great significance in airborne remote sensing video data processing. It is one of the key technologies of mission-to-ground search and fire control system guidance for non-hovering airborne platforms. A detection and tracking algorithm for specific building area is proposed. Firstly, the local description sub-feature is extracted from the off-line reference map of the target area. A fast SIFT(Scale Invariant Feature transform (SIFT(Scale Invariant Feature transform) feature extraction method based on edge salience is proposed to extract local descriptor features for suspected building areas in large field of view images. The high reliability detection of the target area is realized by designing hierarchical feature point matching method. In the online tracking phase of the target area, the suspected target area is effectively delineated by adaptive window opening and trajectory prediction techniques. And the final target confirmation and stable tracking are carried out in the limited range of the delineation. The simulation results of the measured data show that the algorithm proposed in this paper can be used in complex scene conditions. The rapid detection and stable tracking of the specific building area can provide key technical support for the practical application of the system in the future.
【作者单位】：北方工业大学电子信息工程学院;北京理工大学雷达技术研究所;
【基金】：国家自然科学基金(No.61171194) 北京市教委科技计划面上项目(No.KM201610009004)
【分类号】：TP751

【相似文献】