User Verification Feature Transformation for Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user verification systems face challenges in accurately identifying users in varying environments and occlusion conditions, such as low illuminance or the presence of accessories, which can hinder recognition accuracy.
Innovation Solution
A processor-implemented method using a trained neural network, specifically a convolutional neural network (CNN), transforms input feature data and enrolled feature data based on detected characteristics, such as illuminance and accessory presence, to generate transformation functions that maintain target dimensional components and exclude remaining components, thereby enhancing verification robustness across different conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional facial recognition is used without transformation, then the system is simple and fast, but verification accuracy deteriorates in varying environments and occlusion conditions
Solution Approach 1:
The system performs preliminary detection of environmental characteristics (illuminance, occlusion) and landmark characteristics (facial features, accessories) before verification. Transformation functions are pre-determined based on these detected characteristics, allowing the system to adaptively prepare the appropriate transformation in advance, thereby improving verification accuracy without excessive complexity
Solution Approach 2:
The system changes parameters by applying different transformation functions to feature data based on detected characteristics. When environmental or landmark characteristics are detected, corresponding transformation functions modify the feature space parameters (maintaining target dimensional components, excluding remaining components), enabling accurate verification across varying conditions
2Reliability
If transformation functions are applied to feature data, then verification robustness improves across different conditions, but processing time and computational complexity increase
Solution Approach 1:
The system applies local quality by selectively transforming only the necessary dimensional components of feature data. The transformation function maintains target dimensional components (those important for verification) while excluding remaining components, thereby improving robustness without unnecessarily processing all feature dimensions, reducing overall processing time
3Measurement precision
If characteristic detection is performed before verification, then verification accuracy in challenging environments improves, but the number of processing steps increases
Solution Approach 1:
The system merges characteristic detection and verification processes by integrating them into a unified workflow. The detection of environmental and landmark characteristics is combined with the verification process itself, where detected characteristics directly inform the selection and application of transformation functions, reducing the perceived number of separate processing steps
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A processor-implemented verification method includes: detecting a characteristic of an input image; acquiring input feature transformation data and enrolled feature transformation data by respectively transforming input feature data and enrolled feature data based on the detected characteristic, wherein the input feature data is extracted from the input image using a feature extraction model; and verifying a user corresponding to the input image based on a result of comparison between the input feature transformation data and the enrolled feature transformation data.