摘要
基于识别空间目标的需要,采用多视点特征法建立了绕双轴旋转的目标二维图像模型库,将常用滤波器和形态学相结合去除噪声并进行图像分割,得到了较好的分割效果。采用组合不变矩对分割后图像提取目标特征向量,并使用BP神经网络对其进行识别。实验结果表明：所提出的空间目标识别（space object identification,SOI）方法有效,可实现噪声干扰条件下运动卫星目标的识别。
In order to recognize the space object, the two-dimension image modeling storeroom is set up by multi-viewpoints feature method, the filter and morphologic are combined to wipe off the noise, and gets good effect of image segmentation. Combined moment invariants are extracted as the feature extraction, and BP neural network is used to recognize the space object. Simulation results show that the method of space object identification（SOl） is effective, and object recognizing under the noise of background can be achieved.
出处
《装备指挥技术学院学报》
2009年第6期55-59,共5页
Journal of the Academy of Equipment Command & Technology
基金
部委级资助项目
关键词
空间目标识别
多视点特征法
组合不变矩
图像分割
space object identification（SOI）
multi-viewpoints feature method
combined moment invariants
image segmentation
