Image Processing Apparatus Purpose Inference via Appearance Extent
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current surveillance technologies lack the ability to accurately infer the purpose of individuals in captured moving images, leading to degraded decision and extraction accuracy, especially when individuals are in disguise or when inference is based solely on appearance frequency.
Innovation Solution
An image processing apparatus that acquires moving images, generates external appearance information (such as clothing type, personal items, pose, and movement) and appearance extent information, and infers the purpose of individuals present in the image using these data points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If decision is made only based on appearance of a person, then the processing is simple, but decision accuracy is degraded
Solution Approach 1:
The patent combines multiple types of information (appearance information, movement information, and temporal information) into a comprehensive analysis system. This merging of different data sources enables accurate purpose inference while maintaining systematic processing efficiency.
Solution Approach 2:
The system changes the parameters used for analysis from simple appearance-based features to multiple dimensions including movement patterns, temporal frequency, and behavioral characteristics. This parameter expansion significantly improves decision accuracy without excessive complexity increase.
2Device complexity
If extraction is performed only based on appearance frequency, then the processing is simple, but extraction accuracy is degraded
Solution Approach 1:
The patent introduces dynamic analysis of movement patterns and temporal variations in addition to static appearance frequency. This dynamic approach enables the system to distinguish between genuine targets and false positives, significantly improving extraction accuracy.
Solution Approach 2:
The system uses movement information and temporal patterns as intermediary indicators to bridge the gap between simple appearance frequency and accurate target extraction. These intermediaries provide additional validation layers without requiring complex processing.
3Device complexity
If only uniform recognition is performed, then the processing is simple, but the ability to decide various purposes is limited
Solution Approach 1:
The patent creates a universal analysis framework that can infer multiple types of purposes (delivery, construction, soliciting, etc.) using the same core technology. This multi-functional system handles diverse scenarios without requiring separate processing pipelines for each purpose.
Solution Approach 2:
The system adds temporal and behavioral dimensions to the analysis, moving beyond static uniform recognition. By incorporating movement patterns, appearance frequency, and behavioral characteristics, the system gains the ability to infer various purposes accurately.
Data Source
AI summary
An image processing apparatus (10) according to the present invention includes an acquisition unit (11) that acquires a moving image capturing a target area, an analysis unit (12) that generates, based on the moving image, external appearance information indicating at least one of a type of clothing, a type of a personal item, a pose, and a movement of a target person being included in the moving image, and appearance extent information indicating an extent to which the target person appears in the moving image, and an inference unit (13) that infers a purpose for which the target person is present in the target area, based on the external appearance information and the appearance extent information.