Facial Recognition Feature Selection by Attribute

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image processing systems lack efficient methods for identifying specific individuals and updating their registered information, leading to potential misidentification and outdated data due to manual registration processes and the difficulty in maintaining accurate facial features over time.

Innovation Solution

An image processing apparatus and method that includes a subject information storage unit, a subject detecting unit, an attribute determining unit, a feature quantity extracting unit, and a similarity calculating unit to detect and analyze facial attributes, allowing for the selection and registration of suitable feature quantities based on determined attributes and similarity calculations, enabling precise individual identification and automatic updating of facial information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If manual user operations are used to register specific individual information, then the registration process is simple to operate, but the accuracy and timeliness of individual identification deteriorates

Engineering Contradiction:

Improveregistration operationVSAvoidindividual identification accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The system performs automatic individual identification and updates by itself without requiring manual user operations. The imaging apparatus automatically detects faces, determines attributes, extracts feature quantities, calculates similarity, and updates registration information, enabling the system to serve itself and eliminating dependence on user expertise for accurate identification.

Inventive Principle:

Principle #25Self-service

2Ease of operation

If manual user operations are used to register specific individual information, then the operation process is straightforward, but the timeliness of information updates deteriorates

Engineering Contradiction:

Improveregistration operationVSAvoidinformation update timing

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The system continuously performs face detection, attribute determination, and similarity calculation to automatically update individual information whenever conditions are met. This continuous automatic operation ensures that individual identification information is always current without requiring periodic manual updates, thereby eliminating time delays associated with manual registration processes.

Inventive Principle:

Principle #20Continuity of useful action

3Measurement precision

If all feature quantities are used for similarity calculation, then the identification precision may be improved, but the processing complexity and time increase

Engineering Contradiction:

Improvesimilarity calculation accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system determines face attributes (age, sex, facial expression) and selectively uses feature quantities corresponding to the determined attributes for similarity calculation. This localized approach means that not all feature quantities are processed uniformly, but only those relevant to the specific attributes detected, thereby reducing processing complexity while maintaining identification precision.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system changes the parameters used for similarity calculation based on the determined attributes. By dynamically selecting which feature quantities to use based on attribute determination results, the system adapts the calculation parameters to the specific case, reducing unnecessary computations while maintaining accuracy.

Inventive Principle:

Principle #35Parameter changes

4Measurement precision

If variable factors such as face orientation and facial expression are standardized, then the registration accuracy improves, but the adaptability to real-world variations deteriorates

Engineering Contradiction:

Improveregistration accuracyVSAvoidhandling of face variations

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system dynamically determines attributes including face orientation and facial expression during the identification process, rather than requiring static standardized registration. This dynamic approach allows the system to adapt to real-world variations in face orientation and expression, maintaining both registration accuracy and adaptability to actual usage conditions.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS8311293B2Image processing apparatus and associated methodology for facial recognition

Publication Date: 2012.11.13 SONY GROUP CORP

AI summary

An image processing apparatus includes: a subject information storage unit configured to store feature quantities and attributes relating to a plurality of subjects; a subject detecting unit configured to detect a subject included in an image; an attribute determining unit configured to determine the attributes of the detected subject; a feature quantity extracting unit configured to extract a feature quantity relating to the detected subject; and a similarity calculating unit configured to select one or a plurality of feature quantities from feature quantities relating to a plurality of subjects stored in the subject information storage unit based on the determined attributes to calculate similarity between a subject according to the selected feature quantities, and the detected subject, based on the selected feature quantities and the extracted feature quantity.