Multispectral Thermal Imaging With ML Spectral Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing compact thermal imaging spectrometers lack spatial image information and require an active blackbody source, limiting their real-world applications, especially in resource-constrained settings.
Innovation Solution
A multispectral imaging system using a thermal imager with multiple image sensors and a spectral reconstruction module that applies a pretrained machine learning algorithm to generate a reconstructed spectrum, enhancing spatial resolution and eliminating the need for an active blackbody source.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing compact thermal spectrometers are used, then spectral measurement capability is achieved, but spatial image information is lost
Solution Approach 1:
The imaging system is divided into multiple image sensors, each dedicated to capturing a specific spectral band. This segmentation allows simultaneous acquisition of spatial and spectral information by assigning dedicated sensors to different wavelength ranges, thereby preserving both spatial image information and spectral measurement capability
Solution Approach 2:
The system transitions from traditional single-dimensional spectral measurement to multi-dimensional imaging by incorporating multiple image sensors that capture spatial information across different spectral bands simultaneously, adding the spatial dimension to the spectral measurement process
2Measurement precision
If traditional thermal spectrometers are used, then spectral analysis is performed, but an active blackbody source is required
Solution Approach 1:
The imaging system uses the target sample itself as the radiation source by capturing emitted thermal radiation across multiple spectral bands. This self-service approach eliminates the need for external active blackbody sources, simplifying the device while maintaining spectral analysis capability through passive thermal imaging
Solution Approach 2:
The requirement for an active blackbody source is extracted and removed from the system. The patent achieves spectral analysis by directly imaging the thermal radiation from targets without needing external calibration sources during operation, reducing device complexity
3Measurement precision
If multiple image sensors are used to capture multiple spectral bands, then spectral resolution is improved, but device complexity increases
Solution Approach 1:
Multiple image sensors are designed with identical structural configurations but optimized for different spectral bands, allowing them to perform the same imaging function across different wavelengths. This multi-functionality approach maintains spectral resolution while reducing overall device complexity through standardized sensor designs
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
The system provides improved spatial resolution and spectral accuracy, enabling effective identification and characterization of samples, such as minerals, without the need for an active blackbody source, and allows for simultaneous imaging of multiple samples.
Implementation Method 1
each component image associated with a unique spectral band and representing, at each pixel of the particular component image, an intensity of incident radiation, wherein the spectral band of each component image overlaps in part with at least one spectral band of another component image
Data Source
AI summary
A sample analysis method, comprising: obtaining a multispectral image (e.g., a thermal multispectral image) of a first sample of a sample class, said multispectral image corresponding to a first number (n) of component images, each component image associated with a unique spectral band and representing, at each pixel of the particular component image, an intensity of incident radiation, wherein the spectral band of each component image overlaps in part with at least one spectral band of another component image; and applying a sample image analyser to said multispectral image, wherein the sample image analyser implements a pretrained machine learning algorithm configured to generate a reconstructed spectrum comprising a second number (m) of spectral points, wherein the second number is larger than the first number (m>n), wherein the first number is two or greater (n≥2), and wherein the unique spectral bands are arranged to cover an operating band of the long-infrared spectrum, and related device and system.