Image Analysis System for Suspicious Behavior Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis systems struggle to effectively analyze suspicious behavior in wide areas and identify unregistered individuals, as they rely on continuous camera coverage and pre-registered information, making it difficult to detect suspicious individuals in large or partially covered areas.
Innovation Solution
An image analysis system that extracts subject-identification-information from images, generates subject-image-capture-information, and creates appearance records to identify monitored subjects displaying suspicious behavior, even if they are not registered in advance, by comparing extracted information with stored records and matching defined rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single camera is used for behavior analysis, then the system is simple, but the coverage area is limited and blind spots exist
Solution Approach 1:
The system segments the monitoring task by using multiple cameras with different coverage areas (wide-angle camera for overall area, normal camera for detailed behavior). Each camera captures different portions of the monitoring area, and the system integrates information from both cameras to achieve complete coverage without requiring a single complex camera system.
2Area of stationary object
If multiple cameras are used to cover wider area, then the coverage area increases, but the system complexity and cost increase
Solution Approach 1:
The system applies local quality by using different camera types in different locations based on their specific functions. The wide-angle camera is positioned for overall area coverage, while normal cameras are positioned for detailed behavior capture. Each camera is optimized for its specific role, achieving efficient coverage without unnecessary complexity.
3Measurement precision
If facial image recognition is used for suspicious person detection, then identification accuracy is improved, but pre-registration information is required which limits applicability
Solution Approach 1:
The system performs preliminary action by pre-registering only basic facial images without requiring detailed behavioral patterns or multiple angle images. This preliminary registration is sufficient for the system to later detect suspicious behavior through image comparison and behavior analysis, enabling both high accuracy and broad applicability to unregistered individuals.
4Measurement precision
If continuous images are used for behavior analysis, then behavior tracking accuracy is improved, but the system cannot handle discontinuous or partial coverage images
Solution Approach 1:
The system uses an intermediary approach by introducing a behavior analysis unit that bridges the gap between discontinuous images from different cameras. This unit integrates images from multiple sources, compensates for missing segments, and reconstructs complete behavior patterns even when individual camera coverage is partial or discontinuous, maintaining tracking accuracy while handling versatile image inputs.
Data Source
AI summary
A monitor target shooting information generation means (81) extracts, from each of a plurality of images, as information to be used for estimating the identify of a monitor target, monitor target identification information that is identification information of the monitor target, and the monitor target shooting information generation means (81) then generates multiple pieces of monitor target shooting information each including both the extracted monitor target identification information and a shooting time at which the monitor target was shot. An appearance history generation means (82) generates, from the generated multiple pieces of monitor target shooting information, an appearance history of the monitor target that has been estimated to be identical. A determination means (83) determines the monitor target the appearance history of which matches a specified rule.