摘要
提出一种基于轮廓曲率去噪和仿射不变矩的目标识别方法,适用于激光主动成像这样的高噪声复杂应用场合。通过计算每个像素及其邻域的轮廓曲率,判断像素携带的信息量大小,据此对像素点进行分类。对分属不同类别的像素点,使用不同滤波参数的Lee滤波器进行滤波。对滤波后的图像再次提取出轮廓,计算轮廓的仿射不变矩,训练分类器进行目标识别。实验结果表明,本文算法在噪声环境下对目标的仿射变换具有较高的识别率,并且满足激光主动成像识别系统对于实时性的要求。
A target recognition method based on contour curvature and affine invariant moments is proposed, which is applicable for complex high-noise scenes such as laser active imaging system. By calculating the curvature of the pixels and their neighborhood, the information the pixels contain is determined, and then the pixels are classified. For the pixels belonging to different classes, Lee filter with different filter parameters are adopted. After the filtering, the contours of the object are extracted and the affine invariant moments are calculated, so as to train the classifier and identify the object. Experimental results indicate that the proposed method has a high recognition rate for affine object under high-noise conditions and meets the real-time performance requirement of laser active imaging recognition systems.
作者
张鲁薇
谢京江
孙涛
王锐
ZHANG Luwei;XIE Jingjiang;SUN Tao;WANG Rui(State Key Laboratory of Laser Interaction with Matter,Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun Jilin 130033,China;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《太赫兹科学与电子信息学报》
北大核心
2019年第2期305-308,共4页
Journal of Terahertz Science and Electronic Information Technology
基金
中国科学院创新基金资助项目(CXJJ-17-M132)
关键词
目标识别
激光主动成像
轮廓曲率
仿射不变矩
target recognition
laser active imaging
contour curvature
affine invariant moment
