Human Body Security Inspection Method with Dynamic Mode Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current human body security inspection methods using automatic target recognition (ATR) face a trade-off between low rates of missing alarms and high rates of false alarms, leading to degraded inspection efficiency and privacy concerns.
Innovation Solution
A human body security inspection method combining suspicious matter automatic target recognition, privacy protection images, and human-computer interaction, allowing operators to select from three inspection modes: exclusive ATR, ATR with privacy protection images, and ATR with human intervention for modifying recognition results, to enhance inspection efficiency and privacy protection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ATR technique is used to achieve high inspection efficiency and quick inspecting speed, then productivity is improved, but the rate of false alarm increases
Solution Approach 1:
The system dynamically adjusts the inspection mode between fully automatic ATR and manual review based on the confidence level of automatic recognition results. When ATR confidence is high, the system operates in automatic mode for high productivity; when confidence is low or suspicious, it transitions to manual review mode to reduce false alarms, thus dynamically optimizing the balance between efficiency and reliability
Solution Approach 2:
The system introduces an intermediary manual review mechanism that acts as a mediator between automatic ATR recognition and final alarm generation. Operators review suspicious cases identified by ATR, confirming or dismissing potential threats. This intermediary layer filters out false alarms while maintaining the high-speed automatic inspection capability for clear cases
2Speed
If ATR technique is used to achieve quick inspecting speed, then inspection speed is improved, but measurement precision of suspicious matter detection deteriorates
Solution Approach 1:
The system dynamically adjusts the inspection depth and review level based on the complexity and suspicion level of detected objects. Simple, high-confidence detections are processed quickly with minimal review, while complex or ambiguous cases trigger more thorough manual examination, thus adapting the detection precision to the specific case requirements while maintaining overall high inspection speed
3Reliability
If original scanning image is displayed to allow human intervention for modifying recognition results, then reliability is improved, but privacy protection is compromised
Solution Approach 1:
The system extracts and displays only the relevant suspicious regions or specific features of interest to operators during manual review, rather than displaying the complete original scanning image. This extraction approach provides operators with sufficient information to accurately verify recognition results while minimizing the exposure of unnecessary personal information and maintaining privacy protection
Solution Approach 2:
The system applies different quality levels of image display to different regions: full-resolution detailed display for suspicious areas requiring operator review, and blurred or obscured display for non-suspicious personal areas. This local differentiation ensures that operators receive high-quality visual information where needed for accurate detection while protecting privacy in areas where detailed information is not required
Data Source
AI summary
The present invention provides a human body security inspection method and system. The method comprises: retrieving in real-time scanning row or column image data of a personal to be inspected; transmitting in real-time the image data to an algorithm processing module and processing these image data by the module; automatically recognizing a suspicious matter by a suspicious matter automatic target recognition technique, after retrieving an image data of an entire scanning image of the personal; any of the following three inspection modes is selected, so as to perform a further processing on basis of the recognition result of the suspicious matter, (1) in a manner of automatic target recognition technique, (2) in a combination manner of the automatic target recognition technique and a privacy protection image; and (3) a combination manner of the automatic target recognition technique, a privacy protection image and human intervention.