Infrared Image Noise Filtering Through Pixel-Level Modulation and Demodulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional filtering techniques for infrared signals are susceptible to interference and noise, leading to non-optimal performance in real-world scenarios due to environmental factors and limitations in existing filtering mechanisms.
Innovation Solution
Implementing modulation and demodulation techniques at the pixel level using Computational Pixel Imager (CPI) technology to modulate optical scenes at frequencies higher than the excess noise knee frequency, followed by demodulation and digital signal processing to filter out unwanted noise, such as 1/fn noise, using a read-out integrated circuit (ROIC) for real-time noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional filtering techniques are used for infrared signals, then the filtering mechanism is simple to implement, but the signal clarity and performance deteriorate due to interference and noise from environmental factors
Solution Approach 1:
The patent modulates the infrared signal by changing its frequency parameter to a frequency above the excess noise knee frequency. This parameter transformation shifts the signal into a frequency range where noise is minimized, allowing for effective noise reduction while maintaining signal integrity. The modulation converts the original low-frequency signal into a high-frequency modulated signal that can be processed with simpler filtering mechanisms.
Solution Approach 2:
The patent employs periodic modulation of the infrared signal at a specific frequency above the excess noise knee frequency. This periodic action creates a modulated signal that can be distinguished from noise through synchronous detection, enabling effective noise filtering while preserving the original signal information. The periodic nature of the modulation allows for the use of narrowband filtering techniques that are computationally efficient.
2Measurement precision
If modulation and demodulation techniques are implemented at the pixel level, then excess noise is significantly reduced and image clarity is enhanced, but the device complexity increases due to the need for CPI technology and ROIC
Solution Approach 1:
The patent implements modulation and demodulation operations at the pixel level, dividing the noise reduction task into independent per-pixel operations. This segmentation allows each pixel to be processed individually through the CPI technology, enabling parallel processing that reduces overall system complexity while achieving superior noise reduction performance. The read-out integrated circuit (ROIC) applies the demodulation operation to each pixel's signal independently.
Solution Approach 2:
The patent introduces an intermediary modulation frequency (above the excess noise knee frequency) that acts as a mediator between the original infrared signal and the final processed output. This intermediary frequency domain representation allows noise to be separated from the signal through the modulation-demodulation process, enabling effective noise reduction while using relatively simple filtering operations in the intermediate frequency domain.
3Reliability
If the input optical signal is modulated at frequencies higher than the excess noise knee frequency, then noise filtering effectiveness is improved, but the loss of information may increase due to potential signal distortion during modulation
Solution Approach 1:
The patent employs synchronous demodulation that uses the original modulation waveform as a reference feedback signal. This feedback mechanism ensures that the demodulation process accurately recovers the original signal by comparing the modulated signal with the known modulation pattern. The feedback approach prevents information loss by using the reference modulation signal to guide the extraction of the original infrared signal from the modulated version, even in the presence of noise.
Data Source
AI summary
A system, method and device for filtering noise from image data is disclosed. The method includes: (a) obtaining image data based at least in part on a modulated image signal, wherein the modulated image signal is obtained based on modulating an input infrared signal, (b) demodulating the image data to obtain a filtered image signal in which at least part of 1/fn noise is filtered from the image data, and (c) providing the filtered image signal.