Multi-Spectrum Face Identification Against 3D Mask Spoofing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition systems using Wi-Fi enabled cameras can be tricked by 3-D masks or printed images, leading to misidentification.
Innovation Solution
A facial recognition system that utilizes electromagnetic radiation in both visible and infrared spectrums to confirm a match by comparing images from both spectrums, ensuring that both images match a stored pair of images in a database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If facial recognition is performed using only visible spectrum images, then the system is simple to operate and database management is easy, but the system is vulnerable to misidentification by 3-D masks or printed images
Solution Approach 1:
The patent extends facial recognition from a single spectrum (visible light) to multiple spectra (visible and infrared). By capturing images in both visible and infrared spectrums and requiring matches in both dimensions, the system prevents spoofing attacks while maintaining operational simplicity through automated multi-spectrum comparison.
2Reliability
If multiple spectrum images are captured and compared, then misidentification is prevented, but the quantity of data to be processed and stored increases
Solution Approach 1:
The patent divides the facial recognition data into separate spectrum-specific components (visible spectrum images and infrared spectrum images). Each spectrum's images are stored and processed independently, allowing the system to manage large volumes of multi-spectrum data through organized segmentation rather than handling it as a single undifferentiated mass.
3Reliability
If both visible and infrared images must match for identification, then authentication reliability is improved, but the time required for processing increases
Solution Approach 1:
The patent performs preliminary actions by capturing and storing both visible and infrared spectrum images in advance during database creation. When authentication is needed, the system retrieves pre-stored multi-spectrum images for comparison, eliminating the need for time-consuming real-time multi-spectrum capture and reducing processing delays.
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
Prevents misidentification by verifying the authenticity of the face using unique characteristics in both spectrums, enhancing the reliability of facial recognition.
Implementation Method 1
obtain a first image in a first spectrum and a second image in a second spectrum
Data Source
AI summary
A camera of a monitoring system detects electromagnetic radiation in at least two different spectrums. The system includes a database of stored images, where the stored images are stored in pairs of images corresponding to each spectrum and to a common face. A controller identifies a match from among the stored images by comparing the first image obtained by the camera in the first spectrum to each of the first images in the database and by comparing the second image obtained from the camera in the second spectrum to the corresponding second image stored in the database. A match is identified when the first image and the second image from the camera match the first image and the corresponding second image for a pair of images in the database.