基于改进区域分割遥感图像的航天器目标自动识别方法

Automatic recognition method of spacecraft target based on improved region segmentation remote sensing image

摘要:

：传统的航天器目标自动识别方法识别精准度差，为了解决这一问题，基于改进区域分割遥感图像研究了一种新的航天器目标自动识别方法，通过人工排查的方式来追踪航天器所提供的位置信息，并建立三角形立体体系，提取出航天器所追踪的目标和航天器之间的位置关系，实现航天器目标检测，分别针对复杂场景和运动场景对目标进行识别，引用击穿识别方法，基于遗传算法以及变换算法，实现了在复杂的自然遥感图像中能够识别多种目标，但是对于残缺和不完整的目标识别性差，因此又在方法中引入了自动学习智能识别算法，解决了在遥感图像中残缺不完整的目标识别效果差的问题。设定对比实验，结果表明，相较于传统方法，基于改进区域分割遥感图像的航天器目标自动识别方法识别准确率提高了15.23%。

Abstract:

The traditional automatic recognition method of spacecraft target has poor recognition accuracy. In order to solve this problem, a new automatic recognition method of spacecraft target is studied based on the improved regional segmentation remote sensing image. The position information provided by the spacecraft is tracked by manual screening, and the triangle stereo system is established to extract the target tracked by the spacecraft and the target of the spacecraft In order to realize the detection of spacecraft targets, the position relationship between them is used to recognize the targets in complex scenes and moving scenes respectively. The method of breakdown recognition is used. Based on genetic algorithm and transformation algorithm, the recognition of multiple targets in complex natural remote sensing images is realized. However, the recognition of incomplete and incomplete targets is poor, so the automatic learning intelligence is introduced into the method The recognition algorithm can solve the problem that the recognition effect of incomplete target in remote sensing image is poor. The experimental results show that the recognition accuracy of the spacecraft target automatic recognition method based on the improved region segmentation remote sensing image is improved by 15.23% compared with the traditional method.