Face Matching Device Using Photographing Condition Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face matching devices experience reduced recognition accuracy when the input face image is photographed under conditions not registered in the database, leading to potential false recognition of individuals.
Innovation Solution
A matching device that detects the photographing conditions of an input face image and matches it with the closest registered conditions in the database, using condition detection, determination, and specification means to select the most similar registration conditions for accurate matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If face images are registered under multiple different photographing conditions, then recognition accuracy under varying conditions is improved, but device complexity and data management burden increase
Solution Approach 1:
The system performs preliminary registration of face images under multiple predetermined photographing conditions (frontal, left profile, right profile views at various angles). By pre-capturing images under these standardized conditions and storing them in the database, the system prepares matching templates in advance, enabling accurate recognition regardless of the input image's photographing condition without requiring complex real-time adaptation
Solution Approach 2:
The photographing conditions are segmented into distinct categories (frontal view, left profile view, right profile view, each with specific angle ranges). This segmentation allows the system to systematically organize and manage registration images by condition type, making it easier to handle and process multiple conditions without overwhelming complexity
2Reliability
If multiple registration images per person are stored, then matching accuracy under different conditions is improved, but processing time and computational load increase
Solution Approach 1:
The system pre-organizes multiple registration images per person according to their photographing conditions during the registration phase. By having these images prepared and categorized in advance, the matching process can efficiently retrieve appropriate templates without performing complex real-time analysis, thus maintaining high accuracy while reducing processing time
Solution Approach 2:
The system selects and applies specific registration images based on the local characteristics of the input image's photographing condition. Rather than processing all registration images uniformly, the system identifies the most appropriate matched image (e.g., selecting a left profile view registration image when the input is also a left profile view) and focuses processing on that specific pair, reducing overall computational load
Data Source
AI summary
A high-accuracy matching result is obtained when a condition of a photographed input image differs from a condition of a photographed registration image. A face matching device including the registration face image, in which a person is photographed, and a photographing condition which corresponds to the registration face image are registered in a registration face image database. The face matching device includes a condition detecting unit, a registration face image selecting unit, and a matching unit. The condition detecting unit detects a photographing condition in the input face image which includes the photographed person. The registration face image selecting unit determines and selects the closest of the photographing condition from the input face image with the photographing conditions of the registration face images based on the determined closeness of the photographing condition. The matching unit performs matching using the registration face image corresponding to the selected photographing condition.