密集杂波和未知数目条件下的多目标检测算法

针对低信杂比和目标数量未知条件下的多目标检测问题，提出了一种基于Hough变换及扩展网络的检测前跟踪算法。首先采用标准Hough 变换及最小距离法生成侯选航迹，然后以每条侯选航迹为单位构成扩展网络，对侯选航迹进行杂波滤除，并利用功率均值法对侯选航迹进行二次筛选。最后通过剔除野点得到准确的航迹数及航迹参数。实验表明该方法不但实时性强，而且较好地解决了以往检测前跟踪算法在未知目标数量的条件下要求目标数量上限先验已知的问题。 

Detection of an Unknown Number of Targets in Heavy Cluster

This paper deals with detection of unknown quantity of lower observable targets in heavy clutter. A track before detect (TBD) algorithm based on Hough Transform and expanded trellis is proposed in this paper. First, a standard Hough Transform algorithm and a method of minimum distance are used to get the candidate tracks. And then each candidate track forms an expanded trellis which is applied to filter clutters further. In order to reduce false alarms, the mean power of all the measurements on each candidate track is calculated respectively to be as a threshold value. The number of the targets is conformed by a strategy of outliers eliminating and their tracks are extracted successfully at last. Experimental results indicate that the proposed algorithm not only meet the real time challenge, but also be capable of detecting multiple targets without the assumptions of an upper bound to the number of the targets.