多视点空间目标识别方法研究

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。

省情信息资源开发的理论和方法探析

信息是决策的基础。地方志是信息量最大、种类最全、最权威的信息源。利用现代信 息技术和地方志资源建设省情信息资源网络，将对“资政、存史、教化”起到积极的、不可 替代的作用。本文介绍了省情信息资源网络建网目的和基本功能，提出开发建设省情信息资 源网络既需要“硬”系统方法论，也需要“软”系统方法论，并简述了软系统方法论和多视 点法。基于软系统方法论的多视点法可以更好地解决系统建设的“软”问题。

Multi-view coordinate system transformation based on robot

The registration of point cloud is important for large object measurement.A measurement method for coordinate system transformation based on robot is proposed in this paper.Firstly,for obtaining extrinsic parameters,the robot moves to three different positions to capture the images of three targets.Then the transformation matrix X between camera and tool center point(TCP) coordinate systems can be calculated by using the known parameters of robot and the extrinsic parameters,and finally the multi-view coordinate system can be transformed into robot coordinate system by the transformation matrix X.With the help of robot,the multi-view point cloud can be easily transformed into a unified coordinate system.By using robot,the measurement doesn't need any mark.Experimental results show that the method is effective.