Security CT Object Recognition via 2D Views and 3D Back-Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing security inspection CT technologies struggle to accurately identify target objects with complex three-dimensional shape features and material composition due to incomplete expression of shape information and limited integration of two-dimensional recognition results, leading to reduced information content and difficulty in real-time performance.
Innovation Solution
A method involving dimension reduction of three-dimensional CT data to generate multiple two-dimensional views, followed by two-dimensional identification and feature extraction, and then a dimension increase using back-projection to obtain a three-dimensional recognition result, integrating semantic descriptions across dimensions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only two-dimensional projection is performed orthogonal to traveling direction, then processing complexity is reduced, but shape information completeness deteriorates
Solution Approach 1:
The patent applies dimensionality change by performing projections not only in the direction orthogonal to the traveling direction but also in the traveling direction itself. This multi-directional projection approach recovers three-dimensional shape information that would be lost in single-direction projection, effectively addressing the information loss problem while maintaining computational feasibility through systematic dimensional expansion.
2Measurement precision
If three-dimensional data is directly used for identification, then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the three-dimensional recognition problem into multiple two-dimensional projection tasks. By dividing the 3D volume into 2D slices and performing independent identification on each slice, the computational complexity is reduced while preserving the essential shape information needed for accurate recognition, thus achieving a balance between accuracy and processing speed.
Solution Approach 2:
The patent transforms the three-dimensional identification problem into two-dimensional projection problems that can be processed more efficiently. By projecting 3D data onto 2D planes in multiple directions and synthesizing the results, the method achieves fast processing while maintaining recognition accuracy through the preserved geometric relationships in the projections.
3Productivity
If two-dimensional recognition results are used without integration, then processing speed is maintained, but information content is reduced
Solution Approach 1:
The patent merges multiple two-dimensional recognition results by performing logical operations (such as AND operation) on the identification outcomes from different projection directions. This combination process integrates the information from various 2D slices to reconstruct the three-dimensional object's characteristics, effectively recovering the information content that would be lost if only single 2D views were used.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A method and an apparatus of identifying at least one target object for a security inspection CT are provided. The method includes: performing a dimension reduction on three-dimensional CT data to generate a plurality of two-dimensional dimension-reduced views (S10); performing a target identification on a plurality of two-dimensional views to obtain a set of two-dimensional semantic descriptions of the at least one target object, where the plurality of two-dimensional views include the plurality of two-dimensional dimension-reduced views (S20); and performing a dimension increase on the set of two-dimensional semantic descriptions to obtain a three-dimensional recognition result of the at least one target object (S30).