Helideck Image Monitoring Reducing False Alarms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automatic image monitoring systems in helideck environments are prone to false alarms due to high-frequency events, while failing to detect low-frequency events that pose operational risks, leading to a lack of effective monitoring that differentiates between routine and risky situations.
Innovation Solution
The implementation of a Relative Quality Score (RQS) system, which compares time-spaced image frames to a reference frame using methodologies like Multi-scale Structural Similarity Index Measure (MS-SIM) and Laplacian variance ratio, calculates a median quality score to distinguish between high-frequency and low-frequency events, triggering alarms only for significant low-frequency occurrences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional automatic image monitoring systems compare current frames with previous frames to detect changes, then the system can detect events occurring in the helideck environment, but the system generates excessive false alarms due to high-frequency events such as birds flying, water on lens, and routine aircraft operations
Solution Approach 1:
The patent changes the temporal parameter of frame comparison by introducing a time threshold T. Instead of comparing every consecutive frame, the system only triggers alarms when significant differences persist for longer than threshold T. This parameter change filters out high-frequency transient events (birds, water droplets, routine operations) while maintaining detection of low-frequency critical events (obstructions, sabotage, prolonged abnormalities).
Solution Approach 2:
The system dynamically adjusts its response based on the duration of detected changes. By introducing a time-based differentiation mechanism, the system adapts its alarm behavior according to whether events are transient (short duration) or persistent (long duration). This dynamic approach allows the same monitoring system to handle both high-frequency routine events and low-frequency critical events appropriately.
2Measurement precision
If the monitoring system uses frame comparison algorithms to detect all changes, then it can identify potential risks, but it cannot differentiate between routine high-frequency events and critical low-frequency events
Solution Approach 1:
The patent introduces a temporal parameter (time threshold T) to the existing frame comparison algorithm. This single parameter addition enables the system to differentiate between event types based on their duration characteristics, transforming a simple change-detection system into an event-classification system without requiring complex machine learning models or multiple sensor types.
Solution Approach 2:
The system performs preliminary temporal analysis by monitoring the duration of detected changes before triggering an alarm. Instead of immediately alarming upon detecting any frame difference, the system first measures how long the difference persists, then makes a classification decision based on this preliminary measurement. This preliminary temporal assessment simplifies the overall system architecture compared to complex real-time classification algorithms.
Data Source
AI summary
The present invention discloses a method of monitoring images, particularly images from cameras on offshore platforms, where only low-frequency events trigger alarms and high-frequency events are ignored. Preferably, it implements the Multi-scale Structural Quality Score where the compared image frames are spaced in time, for example, by a space of 1 minute, 10 minutes, 15 minutes, 30 minutes, or 60 minutes. If a quality score lower than a threshold quality score is obtained, or if the quality score obtained remains within a predetermined range for a time greater than a predetermined time limit, an alarm may be triggered. Accordingly, only events that represent a significant risk to operational and personnel safety and that persist for a certain period will trigger alarms. Consequently, spurious alarms that would be caused by low relevance events are avoided.