Face Recognition Dictionary Segmentation by Demographics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face recognition systems struggle with high recognition accuracy due to the lack of categorization of similar patterns for each object, leading to poor performance in personal recognition tasks.
Innovation Solution
An image processing technique that classifies face feature information into categories based on individual information such as age and sex, using a management unit, condition setting unit, and determination unit to create a dictionary for high-accuracy recognition processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If face feature information is not categorized for each object, then the dictionary can be created with simpler structure, but the recognition accuracy becomes poor
Solution Approach 1:
The patent segments the face dictionary into multiple category dictionaries based on individual information (age, sex). Each category dictionary stores face feature information for a specific group, enabling more accurate recognition by comparing against appropriate categories. This segmentation resolves the contradiction by organizing complexity into manageable segments while improving recognition accuracy.
Solution Approach 2:
The patent applies local quality by creating different dictionary structures tailored to specific object characteristics (age groups, sex). Instead of a uniform dictionary structure, each category has optimized storage and comparison methods suited to its specific population, thereby improving recognition accuracy for each demographic while managing overall system complexity through localized optimization.
2Measurement precision
If face feature information is categorized by individual information, then recognition accuracy corresponding to individual information is improved, but the process of creating and managing the dictionary becomes more complex
Solution Approach 1:
The patent applies preliminary action by pre-establishing category dictionaries organized by individual information (age, sex) before actual recognition tasks. This preliminary categorization structure is built in advance, making the recognition process more accurate while the complexity of creation is performed once during system initialization rather than during each recognition operation.
Solution Approach 2:
The patent uses parameter changes by organizing the dictionary structure according to specific parameters (age ranges, sex categories). This parameter-based organization allows the system to adapt to different recognition scenarios by selecting appropriate category dictionaries, improving accuracy while managing complexity through systematic parameter-based classification.
3Reliability
If similar patterns are not categorized for each object, then the system operation is simpler, but the personal recognition performance becomes poor
Solution Approach 1:
The patent segments similar patterns into object-specific categories within the face dictionary. By dividing the recognition process into object-specific segments (age groups, sex categories), the system improves personal recognition performance while maintaining operational simplicity through automated category selection based on input parameters.
Data Source
AI summary
An image processing apparatus comprises, a management unit configured to classify a face feature information of a face region of an object extracted from image data into a predetermined category in accordance with a similarity determination, and manage the face feature information in a dictionary, a condition setting unit configured to set category determination conditions for classifying the face feature information into the category in accordance with individual information representing at least one of an age and sex of the object and a determination unit configured to determine, based on the category determination conditions set by the condition setting unit, a category to which the face feature information belongs in the dictionary.