X-ray Container Detection via Symmetry Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current X-ray inspection systems face challenges in automatically detecting containers, such as bottles, within luggage, which can contain hazardous liquids, leading to delays in security checks due to the need for manual intervention and visual inspections.
Innovation Solution
A method and apparatus for detecting containers in X-ray images using symmetry properties, where pre-processing involves edge detection and feature extraction to identify lines of symmetry, allowing for automated identification and segmentation of containers, even when partially obscured or overlapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual visual inspection is performed to detect containers in X-ray images, then detection accuracy is improved, but inspection time increases and productivity decreases
Solution Approach 1:
The system performs automated detection of containers and their contents using image processing algorithms that analyze X-ray images independently, without requiring manual visual inspection. The algorithm automatically identifies containers with symmetrical properties, determines their orientation, and infers content type, enabling the system to serve itself in the detection task.
Solution Approach 2:
The patent replaces the mechanical/visual inspection process with an automated image processing system that uses computational algorithms to detect containers, identify their orientation through symmetry analysis, and infer content. This substitution of manual visual inspection with automated computational analysis maintains detection accuracy while significantly improving inspection speed.
2Productivity
If automated detection algorithms are used to identify containers in X-ray images, then productivity is improved, but measurement precision deteriorates due to difficulty in detecting partially obscured or overlapping containers
Solution Approach 1:
The patent analyzes X-ray images from multiple radiation directions (different viewing angles) to detect containers. By examining the same object from multiple dimensions or perspectives, the system can reconstruct a more complete understanding of container orientation and content, improving detection accuracy for partially obscured or overlapping containers while maintaining automation.
Solution Approach 2:
The system performs pre-processing of X-ray images to enhance features before main detection, and uses symmetry analysis to preliminarily identify potential containers and their orientations. This preliminary analysis prepares the data for more accurate content inference, improving overall detection precision while maintaining automated operation.
3Measurement precision
If multiple radiation directions are used to improve container detection accuracy, then measurement precision is improved, but device complexity and inspection time increase
Solution Approach 1:
The patent uses a single X-ray inspection system that can acquire images from multiple radiation directions by rotating the object or the radiation source. This multi-functional capability allows the same device to perform both single-view and multi-view inspections, reducing the need for multiple specialized devices while maintaining improved detection accuracy.
4Measurement precision
If manual inspection is performed for containers in checked baggage, then detection accuracy is improved, but loss of time increases due to removal and manual checking
Solution Approach 1:
The system automatically detects containers in checked baggage and infers their content without requiring manual removal or visual inspection. The automated algorithm processes images and identifies potential hazards, enabling the system to perform the inspection task independently and eliminate time-consuming manual handling.
Solution Approach 2:
The patent enables continuous automated inspection of checked baggage without the need to stop the conveyor or manually remove items. The system rapidly processes images and identifies containers in real-time, allowing the inspection process to skip the time-consuming steps of manual removal and visual examination while maintaining detection accuracy.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables efficient and automated detection of containers with symmetrical properties in X-ray images, reducing manual intervention and enhancing security check efficiency by accurately identifying potential hazards without the need for manual inspection.
Implementation Method 1
an X-ray image generated by means of an X-ray inspection system
Implementation Method 2
The intensities of the X-rays not absorbed by the object to be inspected and its contents are measured
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The invention relates to a method for detecting an item (23; 53; 104; 402, 404; 501; 603) having at least one symmetry property inside an inspection object (102) on the basis of at least one transmission image (1a; 40a; 400a; 500a; 600a; 600b; 600c), having the steps of: (a) detecting edges (29, 31; 59, 65; 410a, 412a, 410b, 412b, 442, 446, 418a, 420a, 418b, 420b, 444, 448; 512a, 512b) of individual items included in the transmission image and generating a corresponding edge image (1b; 40b; 400b; 500b); and (b) detecting the item by determining a symmetry line (27; 61; 408, 416, 511; 601a; 601b; 601c) which can be assigned to an item (23; 53; 104; 402, 404; 501; 603), which is included in the transmission image (1a; 40a; 400a; 500a; 600a; 600b; 600c) and has at least one symmetry property, on the basis of pairs of image pixels of the edge image (1b; 40b; 400b; 500b) which are arranged symmetrically to one another with respect to the symmetry line (27; 61; 408, 416, 511; 601a; 601b; 601c); wherein, in step (b) for determining the symmetry line (27; 61; 408, 416, 511; 601a; 601b; 601c) in the edge image (1b; 40b; 400b; 500b), only those edge pixels with respect to which the symmetry line (27; 61; 408, 416, 511; 601a; 601b; 601c) is in an item (23; 53; 104; 402, 404; 501; 603) which is included in the transmission image (1a; 40a; 400a; 500a; 600a; 600b; 600c) and to which the edge belongs are taken into account. The invention also relates to an image processing apparatus (210) having at least one processor (211) which is set up to carry out the method.