Biometric ROI Detection Using Depth Maps for Face Authentication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric access control terminals face challenges in accurately detecting and exposing the right subject due to varying lighting conditions, wide fields of view, and misidentification of regions of interest, leading to suboptimal performance of authentication and identification algorithms.
Innovation Solution
The method involves obtaining a radiation image and a depth map, identifying a region of interest in the depth map, selecting a corresponding region in the radiation image, and adapting exposure to optimize detection of biometric traits, while removing stationary objects and focusing on moving subjects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If exposure is adjusted relative to a detected region of interest, then the brightness of the region is normalized, but other regions may become over- or underexposed
Solution Approach 1:
The patent applies local quality by adjusting exposure parameters specifically for the detected region of interest (face) rather than uniformly across the entire image. The system identifies the face location and modifies exposure settings locally to ensure proper brightness normalization for the face while accepting that other regions may have different exposure characteristics.
2Adaptability or versatility
If HDR imaging techniques are used to store numerous light intensity levels, then the full range of illumination is tested, but manipulation becomes cumbersome and slower
Solution Approach 1:
The patent extracts only the necessary exposure information from multiple illumination conditions rather than processing complete HDR images. By identifying the face region and selecting appropriate exposure values from different illumination scenarios, the system achieves adaptability to various lighting conditions without the computational burden of full HDR manipulation.
3Measurement precision
If the closest face is selected based on pixel size, then exposure can be adjusted, but the terminal can be misled by posters or banners depicting large faces
Solution Approach 1:
The patent introduces depth information as an intermediary to disambiguate between actual faces and false positives like posters. By combining pixel size analysis with depth data from depth sensors, the system can distinguish between a large face close to the camera and a large face image far away on a poster, significantly improving the reliability of region of interest identification.
4Ease of operation
If a wide field of view is used for good ergonomics, then the user does not need precise positioning, but detection of the right subject becomes difficult
Solution Approach 1:
The patent transitions from two-dimensional image analysis to three-dimensional spatial analysis by incorporating depth information. This additional dimension allows the system to easily distinguish between subjects at different distances within the wide field of view, making subject detection straightforward even when multiple people are present, while maintaining the ergonomic benefit of a wide field of view.
Data Source
Figure 1
Figure 2(a)~2(e)
Figure 3a~3b
AI summary
The present invention relates to a method for authenticating or identifying an individual, characterized in that it comprises the implementation, by means of data processing means (11) of a terminal (1), of the following steps: (a) Obtaining a radiation image and a depth map on each of which appears a biometric trait of said individual; (b) Identifying in said depth map a first region of interest capable of containing said biometric trait as the set of pixels of said depth map associated with a depth value within a predetermined range; (c) Selecting in said radiation image a second region of interest corresponding to said first region of interest identified in the depth map; (d) Detecting said biometric trait of the individual in said selected second region of interest of said radiation image;(e) Authentication or identification of said individual on the basis of the detected biometric trait;