Point-Cloud Target Extraction for 3D CT Security Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Three-dimensional CT images often contain multiple objects with varying textures, leading to prolonged discrimination times and inaccurate results due to manual input operations by image discriminators.
Innovation Solution
A method and apparatus that utilize point cloud data to automatically identify and strip target objects from three-dimensional CT images by determining local contours and sizes based on prior rules, using technologies like direct volume rendering and super voxel clustering to enhance object discrimination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual discrimination of three-dimensional CT images is performed, then the image discriminator can identify target objects, but the discrimination time is prolonged and accuracy is reduced due to complex textures and multiple objects
Solution Approach 1:
The patent extracts the target object from the complex three-dimensional CT image containing multiple objects and textures. By automatically identifying and separating the target object area using point cloud data and prior rules, the system removes interfering elements, allowing the image discriminator to focus solely on the target object, thereby improving both accuracy and reducing time
Solution Approach 2:
The system enables self-service automation where the computer automatically performs target object identification and extraction without manual intervention. The automated process uses point cloud data processing and prior rules to complete the discrimination task independently, eliminating the time loss and accuracy reduction associated with manual operations
2Productivity
If automated target object extraction is implemented, then discrimination efficiency is improved, but the device complexity increases due to point cloud processing and prior rule requirements
Solution Approach 1:
The patent applies preliminary action by pre-establishing prior rules based on the fixed shape characteristics of target objects. These prior rules are prepared in advance to guide the automatic extraction process, enabling efficient target object identification without requiring complex real-time analysis, thus improving productivity while managing device complexity
Solution Approach 2:
The system introduces point cloud data as an intermediary between the raw three-dimensional CT image and the final target object extraction. The point cloud serves as a simplified representation that facilitates automated processing, acting as a mediator that reduces the complexity of direct image analysis while maintaining high discrimination efficiency
Data Source
Figure 1~2
Figure 3(a)~3(c)
Figure 4~5
AI summary
The present disclosure relates to the field of security inspection. Provided is a method for separating a target object in a three-dimensional CT image. The method comprises: on the basis of CT data, acquiring point cloud data, the CT data being obtained by a security inspection device performing computed tomography scanning on N objects to be inspected, and N being greater than or equal to 2; according to the point cloud data, determining a target object area in N said objects; and separating the target object area from a three-dimensional CT image. Further provided in the present disclosure are an apparatus for separating a target object in a three-dimensional CT image, a device, a storage medium and a program product.