基于双谱的猫眼目标激光一维距离像特征提取

发布时间：2018-06-02 11:17

本文选题：特征提取 + 猫眼效应　；参考：《激光与红外》2017年03期

【摘要】：为了验证一维距离像用于光学目标分类识别的可行性,解决复杂环境和远距离条件下光学目标的探测和识别难题,首次尝试利用激光一维距离像识别具有不同光学参数的猫眼目标,并提出基于双谱的激光一维距离像的稳定特征提取方法。该方法通过信号积累和滤波处理来提升复杂背景下的作用距离和回波信噪比增益,利用双谱计算同时实现远距离目标探测和猫眼目标的稳定特征提取;并通过奇异值分解和主分量计算降低双谱特征提取的运算量;通过处理不同距离、类型及入射角的真实数据样本,验证双谱特征的稳定性和有效性。该方法可以识别具有猫眼效应的光学目标和角反射器,与现有方法比较,具有远距离、低信噪比探测和识别的优点。
[Abstract]:In order to verify the feasibility of using one-dimensional range profile to classify and identify optical targets, and to solve the problem of detection and recognition of optical targets in complex environments and remote conditions, It is the first time to use laser one-dimensional range profile to identify cat eye target with different optical parameters, and a stable feature extraction method of laser one-dimensional range profile based on bispectrum is proposed. The signal accumulation and filter processing are used to improve the range of action and the gain of echo signal to noise ratio (SNR) in complex background. The detection of long range target and the stable feature extraction of cat's eye target are realized simultaneously by bispectral calculation. The computation of bispectral feature extraction is reduced by singular value decomposition and principal component calculation, and the stability and validity of bispectral feature are verified by processing real data samples of different distances, types and incident angles. This method can identify optical targets and angle reflectors with cat eye effect. Compared with existing methods, this method has the advantages of long distance, low signal-to-noise ratio detection and recognition.
【作者单位】：北京理工大学;装备学院;
【分类号】：TN249

【相似文献】