Biometric Input Quality Assessment via Embedding Space Translation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric identification systems face challenges in efficiently assessing the quality of biometric input data, leading to poor system performance when low-quality data is used.
Innovation Solution
The system utilizes embedding networks trained to generate embedding vectors, which are then processed using translator modules to translate these vectors into a common embedding space. The quality of the input data is assessed by calculating the mean distance of the translated embedding vectors, determining if it meets a threshold for sufficient quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional filtering algorithms are used to assess biometric input data quality, then the assessment process can be implemented, but it requires substantial testing, assessment, and manual adjustment which is time consuming and expensive
Solution Approach 1:
The system pre-trains multiple embedding networks and translator modules before deployment. During runtime, the quality assessment module can directly use these pre-trained components to assess biometric input data without requiring manual adjustment or extensive testing at the time of assessment, thus reducing time loss while maintaining high measurement precision
Solution Approach 2:
The patent introduces translator modules as intermediaries that translate embedding vectors from different embedding networks into a common embedding space. This intermediary mechanism enables automated quality assessment without requiring manual tuning of filtering algorithms, resolving the contradiction between assessment accuracy and time consumption
2Measurement precision
If traditional filtering algorithms are used to assess biometric input data quality, then the assessment can be performed, but it requires manual adjustment when system changes occur
Solution Approach 1:
The system employs embedding networks and translator modules that automatically adapt to system changes through their inherent machine learning capabilities. When the system changes (e.g., new embedding networks are introduced), the translator modules can be re-trained automatically without requiring manual adjustment or re-tuning, thus reducing device complexity while maintaining assessment precision
Solution Approach 2:
The patent utilizes parameter changes in the embedding networks and translator modules to adapt to system changes. By modifying the internal parameters of these neural network components through automated re-training, the system maintains high assessment accuracy without requiring complex manual adjustment procedures
3Reliability
If multiple embedding networks are used to generate embedding vectors, then the quality assessment can be more comprehensive, but it increases the complexity of the system
Solution Approach 1:
The patent merges multiple embedding networks and their respective embedding spaces into a unified quality assessment framework. By combining the outputs of multiple embedding networks and translating them into a common space, the system achieves comprehensive and reliable quality assessment while managing system complexity through the unifying translator module architecture
Data Source
AI summary
A user performs an enrollment process to utilize a biometric identification system. This includes acquisition of biometric input data. Accuracy of subsequent identification is improved by utilizing high quality input data during enrollment. Input data is processed using a plurality of embedding models to determine a plurality of embedding vectors. These embedding vectors are translated into a common embedding space. Input quality may be determined based on analysis of these embedding vectors. For example, if a mean distance of the translated embedding vectors is less than a threshold value, the input data may be deemed to be of sufficient quality for use to complete an enrollment process. This analysis may also be used for post-enrollment operation, such as during an identification process to determine query input data that is of insufficient quality.