RGB-to-NIR Vegetation Mapping for Accurate Coverage Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating vegetation coverage are either destructive and time-consuming or lack precision in non-destructive, near-real-time monitoring, particularly in agricultural settings, and existing imaging-based techniques struggle with plant feature processing and illumination variations.
Innovation Solution
A computer-implemented method using two neural networks to process RGB images, one to derive a near-infrared channel and another for segmentation, enhanced with infrared-dark channels to improve accuracy, allowing non-destructive, real-time vegetation coverage mapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional vegetation index calculations using visible light wavelengths are used, then the method is simple and quick, but the measurement precision is low because leaves cause particularly low reflectance in the visible light range
Solution Approach 1:
The patent transitions from analyzing only visible light wavelengths (450-750 nm) to incorporating near-infrared wavelengths (750-3000 nm), adding a new spectral dimension. This allows the system to capture NIR radiation that penetrates the vegetal canopy and reflects off leaf mesophyll, providing much higher reflectance signals for accurate vegetation coverage estimation without requiring complex multispectral hardware.
Solution Approach 2:
The patent uses a neural network to synthesize a pseudo-NIR channel from standard RGB image data, creating a computational copy of infrared information without physical infrared sensors. This copying approach enables NIR-based vegetation analysis using only conventional RGB cameras, significantly reducing device complexity while maintaining measurement precision.
2Reliability
If RGB images are used directly for vegetation analysis, then the device complexity is low, but the reliability is poor due to illumination variations and plant damage affecting detection accuracy
Solution Approach 1:
The patent introduces an infrared-dark channel as an intermediary component that mediates between the RGB input and the final vegetation coverage output. This channel captures illumination-invariant information by computing the difference between the NIR channel and the minimum RGB channel, effectively filtering out illumination variations and enhancing the reliability of vegetation detection under diverse lighting conditions.
Solution Approach 2:
The patent transforms the input color space from standard RGB to a customized R-G-B-NIR parameter space through neural network synthesis. This parameter transformation enables the system to operate in a spectral domain where vegetation signals are more prominent and less sensitive to illumination changes, improving reliability without requiring physical spectral transformation hardware.
3Measurement precision
If biomass measurement is performed through destructive sampling, then the measurement precision is high, but the loss of substance occurs as plant samples are removed from the field
Solution Approach 1:
The patent replaces the mechanical destructive sampling process with an optical imaging and computational analysis system. Instead of physically removing and weighing plant samples, the system uses RGB cameras to capture images and employs neural networks to synthesize NIR channels and calculate vegetation coverage indices, achieving comparable precision without any plant loss.
Solution Approach 2:
The patent creates a digital copy of the vegetation information through image capture and processing. The neural network synthesizes NIR channel data from RGB images, producing a virtual representation of vegetation coverage that mirrors the accuracy of physical biomass measurement while causing no damage to the plants.
4Measurement precision
If near-infrared channel derivation and segmentation using two neural networks is implemented, then the measurement precision and reliability are improved, but the use of energy and computational resources increases
Solution Approach 1:
The patent makes the first neural network serve multiple functions: it both synthesizes the NIR channel and implicitly performs feature extraction for vegetation detection. This multi-functionality reduces the need for separate dedicated processing modules, optimizing energy usage while maintaining high measurement precision through comprehensive spectral analysis.
Data Source
AI summary
Computer-implemented method and system (100) for estimating vegetation coverage in a real-world environment. The system receives an RGB image (91) of a real-world scenery (1) with one or more plant elements (10) of one or more plant species. At least one channel of the RGB image (91) is provided to a semantic regression neural network (120) which is trained to estimate at least a near-infrared channel (NIR) from the RGB image. The system obtains an estimate of the near-infrared channel (NIR) by applying the semantic regression neural network (120) to the at least one RGB channel (91). A multi-channel image (92) comprising at least one of the R-, G-, B-channels (R, G, B) of the RGB image and the estimated near-infrared channel (NIR), is provided as test input (TI1) to a semantic segmentation neural network (130) trained with multi-channel images to segment the test input (TI1) into pixels associated with plant elements and pixels not associated with plant elements. The system segments the test input (TI1) using the semantic segmentation neural network (130) resulting in a vegetation coverage map (93) indicating pixels of the test input associated with plant elements (10) and indicating pixels of the test input not associated with plant elements.