基于邻域一致性的无人机航拍图像配准过程控制技术

Process control technology of UAV aerial image registration based on neighborhood consistency

摘要:

由于无人机航拍图像的采样节点重叠度较高，所以在无人机航拍图像配准过程中无法避免像素点错误匹配的问题，提出基于邻域一致性的无人机航拍图像配准过程控制技术。根据无人机航拍图像配准区域划定标准，识别关键图像标注点，通过计算一致性测度值指标设计基于邻域一致性的无人机航拍图像融合方法。结合图像融合结果进行航拍图像地理定位，根据采样灰度因子的数值水平分析图像采样节点之间的重叠关系，对航拍图像进行坐标变换与几何校正。结合坐标变换与几何校正结果提取图像边界特征点，将特征点作为配准过程控制点，根据控制点分布质量与尺度空间极值实现无人机航拍图像配准过程控制。实验结果表明，在邻域一致性原则作用下，无人机航拍图像采样节点之间的重叠度明显下降，像素点错误匹配问题得到了解决，配准效果好。

Abstract:

Due to the high overlap of sampling nodes of UAV aerial image, the problem of pixel wrong matching cannot be avoided in the process of UAV aerial image registration. A control technology of UAV aerial image registration process based on neighborhood consistency is proposed. According to the criteria of UAV aerial image registration area, the key image annotation points are identified, and the UAV aerial image fusion method based on neighborhood consistency is designed by calculating the consistency measure index. According to the numerical level of sampling gray factor, the overlapping relationship between image sampling nodes is analyzed, and the aerial image is subjected to coordinate transformation and geometric correction. Combined with the coordinate transformation and geometric correction results, the image boundary feature points are extracted, and the feature points are used as the registration process control points. The UAV aerial image registration process is controlled according to the distribution quality of control points and the extreme value of scale space. The experimental results show that under the principle of neighborhood consistency, the overlap between UAV aerial image sampling nodes is significantly reduced, the problem of pixel error matching is solved, and the registration effect is good.