Action Recognition Apparatus Using Segmented Identification for Person Authentication

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for detecting behavior patterns and recognizing moving objects struggle to distinguish between individuals with similar walking patterns, making it difficult to personalize working environments and perform successful person authentication, especially as the number of persons increases.

Innovation Solution

An action recognition apparatus and method that includes an input unit for image data, a moving-object detection unit, a state detection unit, and a learning unit to identify and learn specific actions and states of moving objects, associating them with semantic information for personalized device control and authentication.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If behavior pattern detection methods are used to identify persons, then person authentication can be performed, but individuals with similar walking patterns cannot be distinguished from each other

Engineering Contradiction:
Improveperson authentication reliabilityVSAvoidindividual identification precision
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent segments the identification process into multiple independent components: appearance-based identification (face recognition, body shape), action-based identification (walking pattern, gestures), and attribute-based identification (age, gender, height). By dividing the identification task into these separate segments, the system can distinguish between individuals with similar walking patterns by using other identification segments to differentiate them.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent merges multiple identification results from different units (appearance identification unit, action identification unit, attribute identification unit) to produce a comprehensive identification result. This combination of multiple identification methods allows the system to maintain high reliability in person authentication while achieving precise individual identification, even when some identification methods produce similar results for different individuals.

Inventive Principle:
Principle #5Merging (Combining)

2Adaptability or versatility

If the number of persons to be authenticated increases, then more persons can be served, but actions must be specified more strictly and persons must pay more attention to faithfully reproduce actions

Engineering Contradiction:
Improvesystem adaptability to multiple personsVSAvoidauthentication operation ease
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent creates a universal identification system that handles multiple persons simultaneously through multiple identification units that work in parallel. The appearance identification unit, action identification unit, and attribute identification unit can each independently identify different persons using different criteria, allowing the system to serve an increasing number of persons without requiring more strict action specifications. The system adapts to different individuals' natural behaviors rather than requiring standardized actions.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system performs automatic identification and authentication without requiring persons to consciously control or standardize their actions. The identification units automatically detect and analyze appearance, actions, and attributes, reducing the cognitive load and attention required from persons during authentication. This makes the authentication process easier to operate while maintaining high accuracy for multiple persons.

Inventive Principle:
Principle #25Self-service

3Reliability

If action patterns only are used for person authentication, then authentication can be performed, but successful authentication becomes difficult as the number of persons increases

Engineering Contradiction:
Improveauthentication reliabilityVSAvoidnumber of persons to be authenticated
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments the authentication process into multiple independent identification units that operate in parallel: appearance identification, action identification, and attribute identification. This segmentation allows the system to handle authentication for an increasing number of persons by distributing the identification task across multiple units, each processing different identification features. The segmentation prevents the bottleneck that would occur if only action pattern recognition were used for all persons.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent merges the results from multiple identification units to produce a comprehensive authentication decision. By combining appearance-based identification, action-based identification, and attribute-based identification, the system achieves high authentication reliability even when the number of persons increases. The merged results from multiple identification approaches provide redundant verification, maintaining high reliability across a larger population of authenticated persons.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentUS8094881B2Action recognition apparatus and method, moving-object recognition apparatus and method, device control apparatus and method, and program
Publication Date: 2012.01.10 CANON KK
  • US8094881B2 patent drawing
  • US8094881B2 patent drawing
  • US8094881B2 patent drawing

AI summary

An action recognition apparatus includes an input unit for inputting image data, a moving-object detection unit for detecting a moving object from the image data, a moving-object identification unit for identifying the detected moving object based on the image data, a state detection unit for detecting a state or an action of the moving object from the image data, and a learning unit for learning the detected state or action by associating the detected state or action with meaning information specific to the identified moving object.