3D Earth Surface Reconstruction for GPS-Free Position and Orientation

The position determining system is segmented into multiple independent components: visual sensors capturing images, 3D reconstruction module creating geometric models, DSM generation extracting height information, orthophoto generation creating corrected images, and matching module comparing with source data. Each component operates independently, eliminating single-point failure and providing robust positioning without GPS dependency.

The system performs self-positioning by using its own visual sensors to capture images, reconstruct 3D geometry, generate DSM and orthophoto, and match these self-generated products with source data. The system serves itself without external GPS signals, achieving autonomous position determination that is immune to electromagnetic interference affecting remote communication systems.

The complex computational task is segmented into distinct modules: image acquisition, 3D reconstruction, DSM generation, orthophoto generation, and matching. Each module processes specific aspects of the data independently, allowing for optimized computation in each stage and reducing overall computational burden compared to monolithic approaches.

The system performs preliminary actions by pre-generating and storing source DSM and orthophoto data for the region of interest before the actual positioning task. During operation, only matching against these pre-prepared reference data is required, significantly reducing real-time computational complexity compared to performing full 3D reconstruction and matching from scratch.

Multiple sensor data streams from calibrated cameras are merged into a unified 3D reconstruction. The individual images from multiple cameras are combined through stereo vision and multi-view geometry to create a single coherent 3D model, DSM, and orthophoto. This merging process consolidates redundant information while preserving complementary data, improving measurement precision without proportionally increasing processing requirements.

The system transitions from 2D images to 3D reconstruction by adding the height dimension through depth estimation from multiple camera views. This dimensional transformation allows the system to extract rich geometric and height information (DSM) from the 3D model, enabling accurate position and orientation determination while efficiently utilizing the data from multiple cameras.