Man Overboard Detection Using Stereo Imaging and 3D Trajectory Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current man overboard detection systems on ocean-going vessels are inadequate for automatic or semi-automatic detection, especially in adverse conditions such as night or high seas, with high false alarm rates and limited effectiveness in adverse weather and vessel movement.
Innovation Solution
A system utilizing at least two opposed imaging devices mounted on the lowest deck of a vessel, capturing video streams within an overlapping region of view to determine object size, trajectory, and location, transforming image space into world space to identify man overboard incidents through conformance analysis and multiple cues like expected fall paths and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If crude imaging technology is used for man overboard detection, then the system can be simpler and less expensive, but the detection reliability and accuracy deteriorate
Solution Approach 1:
The system divides the detection task into multiple segments by using multiple imaging devices positioned at different locations (lowest deck and upper deck) and processing images from different camera perspectives. This segmentation allows each camera to focus on specific detection zones, improving overall reliability while managing complexity through modular processing of individual camera feeds before integrating results.
Solution Approach 2:
The system transitions from two-dimensional image analysis to three-dimensional spatial understanding by using imaging devices at multiple vertical levels (lowest deck and upper deck). This multi-level arrangement creates overlapping view volumes that enable the system to determine object size, trajectory, and location in three-dimensional space, significantly improving detection reliability without requiring a single overly complex camera system.
2Area of stationary object
If imaging devices are positioned at the lowest deck to maximize detection range, then the detection area increases, but the resolution and detail of detected objects deteriorate
Solution Approach 1:
The system resolves this contradiction by adding a vertical dimension to the detection architecture. Imaging devices are positioned at both the lowest deck and an upper deck, creating overlapping view volumes that extend in multiple dimensions. This multi-level arrangement allows the system to maintain detection area coverage while improving precision through stereo vision and trajectory analysis across different viewing angles and heights.
Solution Approach 2:
The system merges the detection capabilities of multiple imaging devices positioned at different decks by processing images from both levels simultaneously. The overlapping region of view volumes from lowest deck and upper deck cameras is specifically analyzed to identify and track objects, combining the advantages of both positions to achieve both wide detection area and high measurement precision.
3Productivity
If the system operates automatically without human intervention, then productivity increases, but false alarm rates increase due to inability to distinguish between similar objects
Solution Approach 1:
The system replaces manual visual assessment with automated computer vision algorithms that process images from multiple imaging devices. The automated system uses pattern recognition, trajectory analysis, and conformance checking to automatically identify man overboard incidents while filtering out false alarms through intelligent decision-making algorithms that simulate human judgment patterns.
Solution Approach 2:
The system incorporates feedback mechanisms where detected objects are continuously monitored and verified against expected patterns. The automated analysis compares detected objects against criteria for human figures, fall trajectories, and environmental context, adjusting detection sensitivity based on feedback from multiple imaging sources to reduce false alarms while maintaining high productivity.
4Adaptability or versatility
If the detection system covers all possible fall scenarios from any deck, then adaptability improves, but the complexity of processing and analyzing all possibilities increases
Solution Approach 1:
The system segments the detection problem by focusing imaging devices on specific detection zones corresponding to different deck levels and fall scenarios. The lowest deck cameras monitor falls from lower levels while upper deck cameras monitor upper level falls, allowing the system to handle multiple scenarios through organized modular processing rather than monolithic analysis of all possibilities.
Solution Approach 2:
The system manages processing complexity by organizing detection coverage across multiple vertical levels rather than attempting uniform coverage at a single level. The overlapping view volumes from different decks create a three-dimensional detection framework that systematically covers all fall scenarios while distributing processing tasks across multiple camera feeds and analysis channels.
Data Source
AI summary
A system and method for detecting a man overboard incident on structures such as cruise vessels and oil rigs. The system includes at least two opposed imaging devices which record video streams of a detection cuboid within an overlapping region of view volumes for the imaging devices. The imaging devices are located at the lowest deck of the structure and monitor a fall that passes through the cuboid. Identified objects within the video streams are paired, their conformance is determined, and real world information such as size, trajectory, and location is determined.