基于组合矩的激光成像雷达目标识别算法

随着激光技术的发展,激光成像雷达在现代战争复杂战场环境中逐渐获得了广泛的应用,目前激光成像雷达自动目标识别技术已成为国内外研究的热点问题。提出了基于组合矩的激光成像雷达目标识别算法,从激光成像雷达目标的距离像中提取低阶的Zernike矩、Hu矩和中心矩构成组合矩特征,该特征对距离像噪声不敏感,应用径向基函数(RBF)神经网络对三种地面目标进行分类识别。实验结果表明,该算法与应用Zernike矩和Hu矩特征进行分类识别相比,对三种激光成像雷达地面目标的平均识别率在高载噪比(20dB)下分别提高了1.0%和3.7%;在低载噪比(10dB)下分别提高了11.8%和42.5%;当载噪比高于17dB时,该算法的平均识别率达到100%。因此该算法取得了比较好的识别效果。

Target Recognition Algorithm Based on Combination Moments for Laser Imaging Radar

With the development of laser technology, laser imaging radar gradually possesses vast application in complicated battlefield of modern warfare. At present automatic target recognition technology for laser imaging radar is a hot problem at home and abroad. Target recognition algorithm based on combination moments for laser imaging radar is put forward. Combination moments feature including lower-order Zernike moments, Hu moments and central moments is extracted from range image of laser imaging radar target, this feature is not sensitive to range image noise. Radial base function (RBF) neural network is used to recognize three kinds of ground targets. Experimental result shows that comparing this algorithm with using Zernike moments and Hu moments feature to recognize targets, the average recognition rate of three kinds of ground targets of laser imaging radar is raised by 1.0% and 3.7% separately under high carrier-to-noise ratio (CNR) (20 dB); the average recognition rate is raised by 11.8% and 42.5% separately under low CNR (10 dB); when CNR is higher than 17 dB, the average recognition rate of this algorithm is 100%. Therefore this algorithm gains good recognition effect.