Spectrometry Apparatus Using Random Transmittance Filters

The patent changes the transmittance function parameters of the optical filter from conventional band-pass characteristics to a random transmittance function with multiple peak values distributed throughout the target wavelength band. This parameter change allows a single filter to provide multiple spectral sampling points, effectively increasing the information content without adding more filter elements, thus resolving the contradiction between resolution and device complexity

The patent replaces the mechanical approach of increasing filter quantity with a digital signal processing system that uses L1 norm minimization algorithms. Instead of physically adding more filters to improve resolution, the system uses computational methods to reconstruct the original spectrum from measurements obtained through random transmittance filters, substituting mechanical complexity with digital processing capability

The patent converts the non-ideal transmittance function, which traditionally causes spectrum distortion, into a beneficial random transmittance pattern. By deliberately designing filters with random peak values and multiple transmission bands, the previously harmful distortion becomes a useful sampling pattern that provides diverse spectral information. The L1 norm minimization algorithm then exploits this randomness to accurately reconstruct the original spectrum, turning the manufacturing simplicity advantage into a source of improved measurement capability

The patent fundamentally changes the transmittance function parameters from conventional smooth band-pass characteristics to random patterns with multiple peak values. This parameter transformation allows low-cost nano-process filters to provide rich spectral information that was previously only available from expensive, precisely engineered filters, thereby resolving the contradiction between manufacturing cost and information preservation