Characteristic Band Selection for Compact Active Hyperspectral Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional hyperspectral imaging instruments are bulky, require excessive computing resources due to redundant spectral bands, and lack efficient data screening, making them unsuitable for clinical use and prolonged analysis times.
Innovation Solution
A hyperspectral characteristic band selection method using a virtual dimension algorithm to determine endmembers, followed by band prioritization or averaging to select characteristic bands, reducing the number of bands while maintaining analysis accuracy, and an active hyperspectral imaging device without filters or spectrometers for a lightweight design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional hyperspectral imaging instruments use full-band irradiation halogen lamps and spectrometers, then complete spectral information can be obtained, but the instrument size becomes bulky and unsuitable for clinical use
Solution Approach 1:
The patent extracts only the necessary characteristic spectral bands (1100nm, 1200nm, 1300nm, 1450nm, 1550nm, 1650nm) from the full spectrum, eliminating the need for heavy spectrometers and filters. This selective extraction approach maintains essential spectral information while dramatically reducing instrument complexity and size for clinical portability.
Solution Approach 2:
The patent changes the illumination approach from full-band continuous spectrum to discrete wavelength bands using LED light sources. This parameter change from broad-spectrum halogen lamps to targeted LED wavelengths enables compact device design while preserving diagnostic spectral information.
2Measurement precision
If conventional instruments collect all spectral bands (100+ continuous bands), then comprehensive spectral data is obtained, but computing time and resource consumption increase significantly
Solution Approach 1:
The patent extracts only the 6 critical characteristic bands needed for skin disease diagnosis from the full 100+ spectral bands. This extraction eliminates redundant data processing while preserving diagnostically relevant information, significantly reducing computing time and resource requirements.
Solution Approach 2:
The patent applies partial action by collecting only the necessary 6 characteristic spectral bands rather than all 100+ bands. This partial data collection approach is sufficient for diagnostic purposes while dramatically reducing computational burden.
3Illumination intensity
If conventional instruments use ring-shaped halogen lamps with light pipes and electronic control modules, then predetermined irradiation intensity is achieved, but device complexity and weight increase
Solution Approach 1:
The patent replaces the mechanical/optical system of ring-shaped halogen lamps with light pipes and electronic controls with a simpler LED-based illumination system. LED modules directly provide controlled irradiation at specific wavelengths, eliminating the need for complex light guiding and control infrastructure.
Solution Approach 2:
The patent changes the light source parameter from broad-spectrum halogen to narrow-band LED, enabling direct control of irradiation intensity at each characteristic wavelength without requiring complex optical filtering and electronic modulation systems.
Data Source
AI summary
A hyperspectral characteristic band selection method and an active hyperspectral imaging device using the same. The method using a virtual dimension algorithm to calculate a number of endmembers of a hyperspectral image of a predetermined disease, and using between 1 time and 2 times the number of endmembers as a number of predetermined selected bands, and then selecting characteristic bands with a number matching the number of selected bands from a plurality of spectral bands in the image. The device comprises a base, light sources and a sensing portion, the light sources are respectively disposed on the base, bands emitted by the light sources are configured according to the characteristic bands calculated by the method. The sensing portion is disposed on the base and spaced apart from the light sources for receiving reflected light from an external target object irradiated by the light sources to obtain a hyperspectral image.