为实现从单站光测图像中估计出已知3D模型的空间目标姿态,利用Vega Prime提出了一种采用仿真图像进行相关度局部最优搜索的姿态估计方法, 该方法无需建立2D-3D特征投影关系和大量的模型匹配库。首先, 对输入图像进行图像预处理, 获得目标原始图像。然后, 利用Vega Prime加载目标3D模型生成仿真匹配图像, 并进行图像预处理获得目标匹配图像,计算两幅图像相关度。最后, 更新3D目标模型姿态, 直至仿真匹配图像与目标原始图像的相关度值取得局部最优, 输出目标模型姿态。仿真实验结果表明, 采用本文所提方法的观测仿真图像姿态平均估计误差为385°,仿真原图可实现姿态准确估计, 表明该方法是一种空间目标姿态估计的有效方法。

To estimate the attitude of the known 3D model space object from the mono-view image, a method is proposed by searching the local optimum of the correlation between the simulation image generated from Vega Prime and the observation image, which need not establish the relationships of target feature from 2D to 3D and abundant model matching library. Firstly, the input image is preprocessed to obtain the original target image. Secondly, Vega Prime program loads 3D model of the target, and generates a simulation matching image. The simulation image is preprocessed to obtain the matching target image. And the correlation of these two images is calculated. Finally, the attitude of model is updated until the correlation of the matching image and the original image is local optimum. Experimental results show that this method can estimate the attitude from the original simulation image accurately, and the average estimation error of the simulation observation image is 385°, which indicates that the proposed method is an effective method for space object attitude estimation.