Weed Detection Sensor System Using Reference Spectral Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for detecting weed proportions in fields using herbicides are prone to errors due to changing light conditions and require significant computational effort, especially when considering plant growth stages, leading to inefficient herbicide application.
Innovation Solution
A method involving a static reference sensor and a movable viewing sensor that compares spectral information under the same conditions within the same field, excluding external perturbations and using a mapping unit to determine weed proportions based on differences in spectral data, allowing for accurate and efficient herbicide application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If reflectivity detection is used to determine weed proportion, then herbicide application can be optimized, but measurement precision deteriorates under strongly changing external light irradiation
Solution Approach 1:
The patent combines a viewing sensor aimed at the observation section with a reference sensor aimed at a reference section, merging their spectral information to determine weed proportion. This combination allows cancellation of common-mode lighting variations while preserving weed detection capability.
Solution Approach 2:
The reference sensor acts as an intermediary that measures the background spectral information including lighting conditions. By subtracting this reference spectrum from the viewing sensor spectrum, the system eliminates the harmful effect of varying light irradiation on measurement accuracy.
2Measurement precision
If look-up tables with high quality adaptation to lighting conditions and plant growth status are used, then measurement precision improves, but device complexity and computing effort increase
Solution Approach 1:
The patent extracts only the necessary spectral comparison function, removing the need for large look-up tables storing spectral information for all possible lighting and growth conditions. By using a reference sensor to capture current conditions, the system extracts only the relevant baseline data needed for comparison.
Solution Approach 2:
Instead of storing pre-computed tables for all parameter combinations (lighting, growth stage), the system dynamically adjusts the reference spectrum parameter to match current conditions. This allows the same simple comparison algorithm to work across varying conditions without requiring extensive pre-computed data structures.
3Area of stationary object
If spectral information is recorded at different times, then measurement coverage increases, but measurement precision deteriorates due to changes in lighting conditions and plant growth status
Solution Approach 1:
The system uses periodic synchronous measurement where both the viewing sensor and reference sensor take measurements simultaneously at regular intervals. This periodic simultaneous sampling ensures that both spectra reflect the same lighting and growth conditions, maintaining precision while covering the entire field over time.
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
This approach enhances the accuracy and efficiency of weed detection, reducing errors from lighting changes and plant growth variations, thereby optimizing herbicide application by ensuring precise dosage based on actual weed proportions.
Implementation Method 1
detect the intensity of at least two light wavelengths for the field of view of the sensor
Data Source
Figure 1~2
AI summary
A method for determining a weed percentage in a viewing section (10) of a field (20) is described. Actual spectral information is acquired by means of at least one optical viewing section sensor (12a, b) which is directed to the viewing section (10). Furthermore, reference spectral information is acquired by means of an optical reference sensor (32), which is directed to a reference section (30) of the same soil surface (22). A difference between the actual spectral information and the reference spectral information is determined. Finally, the difference is applied to the weed percentage by means of a predefined image. Furthermore, an agricultural control device (100) is described which is suitable for the implementation of the method.