摘要
介绍了一种利用SOFM(自组织特征映射)网络的聚类功能进行全天星图识别的算法。应用C++软件编程,对全天导航星星表的星信息进行了星三角采样、分类,指出SOFM网络可以很好地提取、反映星表中的复杂信息。从实际观测星图的识别结果可以看出该算法能在中型以上视场的星图识别中发挥很好的作用,虽然识别时间比三角形算法长了些,但其抗噪能力远超过任何其他方法。
An autonomous star pattern recognition method using the tri-star clustering function of SOFM (Self-Organizing Feature Maps) network is described. C+ + software is used to classify tri-stars of the guide stars in celestial sphere and implement the method. It is concluded that the SOFM network can reflect the complicated information among star patterns better. The effect of classification is good, and it can be well applied to star pattern recognition. Its robustness with respect to noise exceeds any other method, although identification time becomes longer.
出处
《光学与光电技术》
2003年第5期47-50,共4页
Optics & Optoelectronic Technology