Face Matching Using Preferential Ranges to Reduce Computational Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face matching technologies suffer from low efficiency, resulting in significant feedback delays due to the need for extensive comparisons of face images.
Innovation Solution
A face matching method and apparatus that prioritize comparisons by determining preferential matching ranges based on attributes of the first face information, filtering and comparing second face information within these ranges to enhance matching efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If face matching compares first face information with all second face information in the database, then matching accuracy is improved, but matching efficiency deteriorates and feedback delay increases
Solution Approach 1:
The patent segments the face matching process into two distinct stages: coarse matching and fine matching. Coarse matching uses simplified features (age, gender, hairstyle) to quickly filter the database into a preferential matching range, while fine matching performs comprehensive comparison only on the filtered subset. This segmentation resolves the contradiction by maintaining high accuracy through fine matching while improving efficiency through coarse filtering.
Solution Approach 2:
The patent performs preliminary filtering actions before the main matching operation. By pre-extracting attributes (age, gender, hairstyle) and pre-defining preferential matching ranges based on these attributes, the system prepares the data in advance to enable rapid filtering. This preliminary action reduces the search space before comprehensive matching occurs, thereby improving efficiency without sacrificing accuracy.
2Reliability
If face matching compares first face information with all second face information in the database, then matching completeness is improved, but feedback delay increases
Solution Approach 1:
The patent segments the matching process to perform completeness checking efficiently. The coarse matching stage quickly identifies all potential candidates within preferential ranges, ensuring no true matches are missed, while the fine matching stage comprehensively verifies each candidate. This segmented approach maintains matching completeness while reducing overall feedback delay through efficient filtering.
Solution Approach 2:
The system performs preliminary attribute extraction and range determination before comprehensive matching. By pre-processing the data to identify preferential matching ranges based on key attributes, the system prepares everything needed for complete matching in advance, thereby maintaining completeness while minimizing the time required for the actual matching operation.
3Measurement precision
If face matching performs comprehensive comparison with all second face information, then matching precision is improved, but computational load increases
Solution Approach 1:
The patent divides the matching process into coarse and fine stages with different precision requirements. Coarse matching uses low-complexity attribute comparison (age, gender, hairstyle) to filter candidates, while fine matching applies high-precision comprehensive comparison only to the filtered subset. This segmentation maintains matching precision where needed while dramatically reducing overall computational load.
Solution Approach 2:
The system performs preliminary filtering based on simple attributes before executing computationally intensive comprehensive comparisons. By pre-determining preferential matching ranges using easily computable features, the system reduces the number of high-complexity operations required, thereby maintaining precision while reducing computational load.
Data Source
AI summary
Examples of the present disclosure provide a face matching method and a face matching apparatus, and a storage medium. The face matching method includes: obtaining a first attribute of first face information which is to be matched; determining one or more preferential matching ranges based on the first attribute; and comparing the first face information with second face information in the one or more preferential matching ranges.