Thermal Face ROI Detection for Occluded Crowd Temperature Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face detection algorithms for thermal images are computationally intensive and require significant processing power, memory, and user input, making them unsuitable for real-time ROI detection on low-end devices, and they struggle with occlusions and varying orientations in crowded environments.
Innovation Solution
A heuristic approach for ROI detection in thermal face images using a deep learning model and morphological operations, combined with a first and second heuristic method to identify and correct for occlusions, allowing for efficient and accurate ROI detection on low-power devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional face detection algorithms (dlib, openface, ViolaJones) are used on thermal images, then face detection can be performed, but processing speed is slow and computational resources are excessive for real-time applications on low-end devices
Solution Approach 1:
The patent segments the thermal image processing into distinct stages: initial face region localization using simplified thermal signature matching, followed by refined ROI detection using heuristic approaches. This segmentation allows the system to quickly eliminate non-face regions before applying more computationally intensive analysis only to relevant areas, thereby improving processing speed while maintaining detection accuracy.
Solution Approach 2:
The patent applies partial action by implementing a two-stage detection process where only promising regions identified in the first stage undergo full analysis in the second stage. This avoids applying computationally expensive algorithms to entire images or irrelevant regions, reducing overall computational load and enabling real-time processing on low-end devices while preserving detection precision for actual faces.
2Measurement precision
If sophisticated thermal image processing methods (edge detection, ellipse fitting) are applied, then face detection accuracy improves, but processing power, time, and memory consumption increase significantly
Solution Approach 1:
The patent performs preliminary thermal signature matching and face region localization before applying sophisticated processing methods. By pre-identifying potential face regions using computationally lighter techniques, the system prepares the data structure and reduces the search space, enabling subsequent accurate ROI detection with heuristic approaches to consume less processing power and memory while achieving the required precision.
3Measurement precision
If existing face detection algorithms are used, then face detection can be achieved, but they require additional user input specifying occluding objects (glasses, hats), making the system semi-automated and introducing delays
Solution Approach 1:
The patent implements self-service by developing heuristic approaches that automatically detect and handle occluded faces without requiring user input. The system autonomously identifies occlusion patterns, adjusts detection parameters, and locates ROIs on occluded faces using thermal signature analysis and spatial reasoning, thereby maintaining high automation levels and avoiding the delays associated with manual intervention.
4Measurement precision
If manual infrared temperature gun scanning is used, then temperature measurement can be performed, but the screener is at high risk due to close proximity with potential infectors and delays occur at crowded places
Solution Approach 1:
The patent introduces thermal imaging technology as an intermediary between the screener and the subject. Instead of direct contact or close proximity scanning with manual infrared guns, the system uses thermal cameras to capture thermal images from a distance, automatically detect faces and ROIs, and determine temperatures without requiring the screener to approach subjects closely. This intermediary approach eliminates physical contact risks while maintaining temperature measurement precision.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Embodiments herein provide a method and system for real time ROI detection in thermal face images based on a heuristic approach. The ROI of the thermal images, once detected, is then further used to detect temperature of a subject corresponding to the ROI. Unlike state of the art techniques, the heuristic approach is computationally less intensive and provides fast and accurate ROI detection even in case of occluded faces in a crowd with a single thermal image having a plurality of subject being scanned. The heuristics applied does not focus on face detection but directly on point of interest detection. Once the point of interest (ROI) is detected, it may be used for plurality of applications such as subject tracking and the like, not limited to subject or object temperature sensing since the method disclosed herein is easily implementable on low power devices.