基于时空分集理论的弱点状运动目标检测技术研究

A spatial-temporal diversity and linear combining based TDB technique for detection of dim moving point targets in image sequences

研究了一种基于时空分集与合并的图像序列中微弱点状运动目标高维积累实时检测技术,指出了同类文献中的其它算法可以归属为本文算法的几种特例。考虑目标尺寸、运动速率和亮度信息可变的实际情况,提出了相应的解决方案:针对以不同速率运动的多目标提出了金字塔式分辨率亚采样方案;针对不同尺寸的目标则提出了成批处理方案;针对可变亮度的目标则提出了可变集成度解决方案。研究了集成度、运动速度和目标亮度(信噪比)之间的具体数值关系,得到了有用的理论分析和仿真结果。

A spatial-temporal diversity theory based method for detection of dim moving point targets in image sequence was presented. It was pointed out that the methods presented in other papers could be seen as special cases of our algorithm. In many practical applications, targets were moving faster or slower than 1 pixel/frame which was the basis of most detection algorithms operates well. Further, the amplitudes and sizes of targets also were varying. To cope with those situations, the pyramid resolution sub-sampling solution for detection of targets moving with various velocities, the batch processing solution for proposed target sizes and the variable-integration-level solution for detection of targets with fluctuating luminance are also considered. And the inner relationships of those factors are also given in this paper.