Real-time Image Geo-registration Using Stochastic Kriging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current full image geo-registration from airborne platforms for Wide Area Moving Imagery (WAMI) is computationally intensive and typically conducted in post-mission processing, lacking real-time capabilities, especially for passive monocular Electro-Optical/Infrared (EO/IR) sensors where instantaneous geo-location of landmarks is challenging without combining information from different viewpoints.
Innovation Solution
A geo-registration system that uses a camera to identify features, compute vectors based on known camera location and orientation, and applies stochastic kriging techniques to create a geo-registration model by comparing these vectors with prior images, enabling real-time geo-location and geo-registration without external database access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If batch processing is used for full image geo-registration, then processing accuracy is improved, but processing time increases significantly
Solution Approach 1:
The system performs preliminary geo-registration using inertial navigation data and camera pose information before final refinement. By pre-computing initial geo-locations and using them as starting points for iterative optimization, the system achieves accurate results faster than traditional batch processing
Solution Approach 2:
The geo-registration process is divided into multiple independent stages: feature detection, initial geo-location using inertial data, vector computation, and stochastic refinement. Each stage can be processed separately and in parallel, reducing overall processing time while maintaining accuracy
2Measurement precision
If external database access is required for geo-registration, then geo-location accuracy is improved, but system complexity and bandwidth requirements increase
Solution Approach 1:
The system uses its own inertial navigation system and onboard sensors to provide geo-registration services without requiring external database access. By leveraging self-contained navigation data and stochastic kriging algorithms, the system achieves autonomous geo-location capability
Solution Approach 2:
Stochastic kriging serves as an intermediary method that combines inertial navigation data with limited sensor observations to infer geo-locations. This statistical mediator enables accurate geo-registration without direct access to external reference databases
3Productivity
If real-time processing is implemented, then processing speed is improved, but computational resource requirements increase
Solution Approach 1:
The system processes only the most critical features and uses approximate methods for less important regions. By applying full computational resources selectively to key features while using simplified processing for others, the system achieves real-time performance with reduced energy consumption
Solution Approach 2:
The system updates geo-registration at periodic intervals rather than continuously processing every frame. By updating at optimized frequencies based on platform motion and mission requirements, the system maintains real-time capability while reducing computational energy demands
Data Source
AI summary
A geo-registration system with a camera receives an image of a scene, identifies features in the image, and computes vectors for each feature based on the known location and orientation of the camera. The system then compares the computed vectors to vectors associated with the same features identified in prior images. The geo-registration system then uses stochastic kriging techniques to create a geo-registration model based on the vectors.