Fused X-ray Image Material Identification via Spectral Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current security screening processes using penetrating images, such as X-rays, face challenges in distinguishing between materials of similar density and shape, leading to unreliable and labor-intensive human interpretation, especially when differentiating between threatening and non-threatening objects.
Innovation Solution
The method involves fusing multiple penetrating image data sets captured using different spectra or detector responses to identify materials by associating them with atomic numbers or material descriptors, utilizing techniques like pixel value estimation, mismatch analysis, and regularization to create a material identifier image that minimizes mismatch and roughness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human interpretation of penetrating images is used to identify materials, then flexibility and adaptability are maintained, but reliability and productivity deteriorate due to human error and labor intensity
Solution Approach 1:
The patent replaces the mechanical human interpretation process with an automated computational system that uses pixel value analysis, mismatch calculation, and regularization algorithms to identify materials, thereby eliminating human error while maintaining systematic processing capability
Solution Approach 2:
The system enables self-service material identification by automatically analyzing penetrating image data, calculating mismatches between observed and expected pixel values, and determining material composition without requiring human intervention, thus improving reliability while reducing operational complexity
2Measurement precision
If multiple penetrating image data sets with different spectra are used to identify materials, then measurement precision improves through material differentiation, but device complexity increases due to multiple data acquisition systems
Solution Approach 1:
The patent implements a universal processing framework that can handle multiple penetrating image data sets with different spectra using the same mismatch calculation and regularization algorithms, allowing one system to perform multiple material identification functions without proportionally increasing complexity
Solution Approach 2:
The system changes spectral parameters by analyzing pixel values across multiple energy spectra, using these parameter variations to calculate mismatches and differentiate materials based on their unique attenuation characteristics, thereby improving measurement precision through controlled parameter variation
3Productivity
If automated processing algorithms are implemented to identify materials, then productivity improves through faster analysis, but measurement precision may worsen due to algorithmic limitations
Solution Approach 1:
The patent incorporates feedback mechanisms where the automated algorithm iteratively refines material identification by comparing observed pixel values with expected values, calculating mismatches, and adjusting regularization parameters to improve accuracy while maintaining high processing speed through efficient computational loops
Solution Approach 2:
The system applies partial action by focusing computational resources on regions of interest within the penetrating images, using regularization to constrain the solution space, thereby achieving high precision material identification at accelerated speeds without requiring exhaustive analysis of all image data
4Loss of information
If material identification is performed without spectral differentiation, then device complexity is reduced, but loss of information occurs regarding material composition
Solution Approach 1:
The patent extracts material composition information by isolating and analyzing spectral characteristics from penetrating image data, using mismatch calculations to separate material-specific attenuation patterns from overall density information, thereby retrieving composition data without requiring complex additional sensing systems
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
This approach enables reliable and efficient identification of materials within occluded objects, improving the accuracy and speed of security screening by providing both material and shape information, reducing human error and increasing the reliability of the screening process.
Implementation Method 1
The capture of images of a given object using penetrating energy (such as X-rays or the like) is well known in the art
Implementation Method 2
images often comprise areas that are relatively darker or lighter (or which otherwise contrast with respect to one another) as a function of the density, path length, and composition of the constituent materials
Data Source
AI summary
First image data (which comprises a penetrating image of an object formed using a first spectrum) and second image data (which also comprises a penetrating image of this same object formed using a second, different spectrum) is retrieved from memory and fused to facilitate identifying at least one material that comprises at least a part of this object. The aforementioned first spectrum can comprise, for example, a spectrum of x-ray energies having a high typical energy while the second spectrum can comprise a spectrum of x-ray energies with a relatively lower typical energy. By one approach, this process can associate materials as comprise the object with corresponding atomic numbers and hence corresponding elements (such as, for example, uranium, plutonium, and so forth).