非均匀噪声环境下的滑窗式恒虚警检测器

发布时间：2018-02-27 18:36

本文关键词： 信号处理滑窗式恒虚警目标检测　出处：《红外技术》2017年03期 　论文类型：期刊论文

【摘要】：为了解决传统目标检测算法在非均匀噪声环境下检测性能严重下降的问题,提出了一种自由滑窗式恒虚警检测算法,在被检测单元两侧各取N-1个参考单元作为初始自由参考窗,然后从左向右依次滑动N次,得到N个滑窗,根据滑窗噪声功率的平均值与被检测参考单元功率值的比较结果,选择相应的滑窗噪声功率平均值集合取均值,再乘以参数因子T得到比较门限阈值S,根据被检测单元的功率与S进行比较,确定是否为有效目标。经过与其他算法进行仿真对比,该算法具有最优的检测性能,检测率98.93%,误检率2.28%,并成功应用于车辆开门预警系统,经测试,预警率大于98.10%,虚警率小于2.80%。结果表明,该算法提高了非均匀噪声环境下目标的检测概率,具有良好的检测性能。
[Abstract]:In order to solve the problem that the detection performance of traditional target detection algorithm in non-uniform noise environment is seriously degraded, a free sliding window CFAR detection algorithm is proposed, in which N-1 reference units are taken as the initial free reference window on both sides of the detected unit. Then N sliding windows are obtained from left to right. According to the comparison of the average noise power of the sliding window and the power of the detected reference unit, the average value set of the noise power of the corresponding sliding window is chosen to take the mean value. Then multiply the parameter factor T to get the comparison threshold threshold S, and compare the power of the detected unit with S to determine whether it is an effective target. Through simulation and comparison with other algorithms, the algorithm has the best detection performance. The detection rate is 98.93%, and the false alarm rate is less than 2.80. The detection rate is 98.93%, and the false alarm rate is less than 2.80. The result shows that the algorithm can improve the detection probability of the target in the non-uniform noise environment and has good detection performance, after testing, the early warning rate is more than 98.10 and the false alarm rate is less than 2.80.
【作者单位】：安徽工程大学计算机与信息学院;奇瑞汽车股份有限公司前瞻技术研究院;
【基金】：国家自然科学基金资助项目(91120307) 安徽省自然科学基金项目(KZ00215072)
【分类号】：TN957.51

【相似文献】