Infrared Depth Map False Alert Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video surveillance systems face challenges in efficiently utilizing infrared emitters and accurately identifying objects and zones within a scene, leading to false alerts and inefficient data recording, particularly in low-light conditions and dynamic environments.
Innovation Solution
The implementation of a camera system that controls infrared emitters individually to create depth maps, allowing for more precise identification of objects and zones, and uses machine learning to categorize motion events and adjust alerts based on context, reducing false positives and improving data management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional video surveillance systems record all motion events, then comprehensive monitoring coverage is achieved, but data storage requirements and processing load increase significantly
Solution Approach 1:
The patent extracts and removes false positive motion events from the surveillance data stream by comparing infrared depth information with visible light image data. Motion events that appear in both wavelength channels are identified as genuine, while those appearing only in the visible channel are filtered out as false positives (such as insects or dust particles), thereby reducing storage requirements while maintaining reliable monitoring of actual events
Solution Approach 2:
The system implements feedback by continuously comparing motion detection results across multiple wavelength channels and using machine learning algorithms to learn from historical data patterns. This feedback mechanism enables the system to automatically adjust false positive identification criteria and improve filtering accuracy over time, maintaining comprehensive monitoring while reducing unnecessary data storage
2Illumination intensity
If multiple infrared emitters are activated simultaneously to illuminate the scene, then sufficient illumination is provided, but depth mapping precision and object identification accuracy decrease
Solution Approach 1:
The patent segments the infrared emitter activation process into sequential phases rather than simultaneous operation. Different groups of infrared LEDs are activated in separate time intervals, allowing the system to capture depth information from multiple angles and positions. This segmentation enables precise depth mapping while maintaining sufficient overall illumination through cumulative exposure across multiple emitter groups
Solution Approach 2:
The system employs periodic activation patterns for different infrared emitter groups, cycling through various subsets of emitters in a structured sequence. This periodic action allows the camera to capture multiple depth maps from different illumination perspectives, improving both depth mapping precision and object identification accuracy while maintaining adequate scene illumination through the cumulative effect of periodic emitter activation
3Illumination intensity
If infrared emitters are activated in low-light conditions to capture the scene, then visibility is improved, but false alerts increase due to inability to distinguish objects from background
Solution Approach 1:
The patent adds a depth dimension to the surveillance system by capturing infrared depth maps alongside visible light images. This dimensional enhancement allows the system to distinguish objects from background based on their spatial position and depth characteristics, not just intensity variations. Objects at different depths create distinct patterns in the depth map, enabling reliable differentiation between genuine targets and background elements even in low-light conditions
Solution Approach 2:
The system exploits the fact that different materials reflect infrared light differently, creating characteristic 'color' patterns in the infrared and depth channels. By analyzing these material-specific reflection patterns and comparing them across wavelength channels, the system can identify genuine objects versus background elements or false positives (such as insects or dust), thereby improving alert accuracy while maintaining visibility in low-light conditions
4Reliability
If zones are manually configured to exclude false alert regions, then false alerts are reduced, but system complexity and setup time increase
Solution Approach 1:
The patent implements self-service by enabling the system to automatically identify and configure false alert regions without manual user intervention. The machine learning algorithms analyze historical motion detection data, infrared depth information, and visible light images to automatically determine which regions consistently generate false positives. The system then automatically excludes these regions from monitoring or applies reduced sensitivity thresholds, eliminating the need for manual zone configuration while maintaining reliable false alert reduction
Solution Approach 2:
The system employs feedback mechanisms where motion detection results, alert accuracy metrics, and user corrections are continuously fed back into the machine learning model. This feedback enables the system to learn from actual performance data and automatically refine false alert region identification and exclusion strategies over time, reducing system complexity while maintaining high reliability in false alert reduction
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances user satisfaction by providing more accurate and efficient video monitoring, reducing unnecessary data recording and improving alert relevance through precise object identification and context-aware motion event categorization.
Implementation Method 1
surveillance cameras include infrared emitters in order to illuminate a scene when light from other sources is limited or absent
Implementation Method 2
a sensor array for capturing images of an illuminated scene
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Inter alia a method of estimating height and tilt angle of a camera system having a 2-dimensional array of image sensors and a plurality of IR illuminators in fixed locations relative to the array of image sensors is disclosed. The method is performed at a computing device having one or more processors, and memory storing one or more programs configured for execution by the one or more processors. Also disclosed are a corresponding computing device and a corresponding non-transitory computer readable storage medium.