Airborne Camera Multi-Exposure Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current moving object detection systems fail to provide real-time detection over wide areas using optical sensors due to limitations in revisit time and coverage area, leading to high false alarm rates and inability to monitor densely built-up metropolitan areas effectively.
Innovation Solution
The system employs a method of capturing rapid sequences of partially overlapping images (multi-exposures) with short revisit times to detect and track moving objects, using airborne cameras and on-board processing to reduce data transmission and enhance detection accuracy, incorporating photogrammetry and computer vision for 3D scene reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If continuous monitoring is performed over a wide area, then coverage area is improved, but revisit time increases beyond the desirable 0.1-0.5 sec range
Solution Approach 1:
The system divides the wide area into multiple sub-areas or zones that can be monitored sequentially. By segmenting the coverage area, the system can revisit each segment within the desirable 0.1-0.5 sec time window while still providing comprehensive coverage of the entire wide area over longer periods.
Solution Approach 2:
The system implements periodic monitoring cycles where different regions are revisited at different intervals. High-priority areas with shorter revisit times (0.1-0.5 sec) are monitored more frequently, while other areas are monitored at longer intervals, creating a periodic action pattern that balances coverage area and revisit time requirements.
2Measurement precision
If short revisit times of 0.1-0.5 sec are maintained, then motion detection accuracy is improved, but coverage area is limited
Solution Approach 1:
The monitoring system segments the wide area into multiple regions that can be monitored with short revisit times. Each segment receives the full benefit of 0.1-0.5 sec revisit intervals for accurate motion detection, while the combination of multiple segments achieves wide area coverage that would be impossible with a single continuous scan.
Solution Approach 2:
The system dynamically adjusts monitoring parameters based on detected activity. Areas with detected motion or high interest receive increased monitoring frequency with short revisit times for accurate tracking, while quiescent areas are monitored at lower frequencies, allowing the system to maintain high motion detection accuracy across wide areas by dynamically allocating resources.
3Reliability
If multiple overlapping images are captured in rapid succession, then motion detection reliability is improved, but data transmission volume increases
Solution Approach 1:
The system extracts only the essential motion information from multiple overlapping images rather than transmitting the complete image sequences. By taking out only the relevant motion vectors, displacement data, and detected object information, the system maintains motion detection reliability while dramatically reducing the data transmission volume to manageable levels.
Solution Approach 2:
Instead of transmitting multiple complete image copies, the system creates simplified representations or copies of the motion information. These copies contain the essential motion detection data derived from comparing multiple images, preserving reliability while reducing transmission requirements by transmitting processed motion data rather than raw image sequences.
Data Source
AI summary
A method for moving object detection, comprising generating a time series of multi-exposures of scenes, each multi-exposure of a scene comprising a sequence of at least two at least partially overlapping images of that scene captured in rapid succession, wherein the time series of multi-exposures periodically revisits substantially the same scenes, detecting moving objects within each multi-exposure by comparing its sequence of overlapping images, and tracking objects by comparing moving objects detected within multi-exposures of substantially the same scenes.