Grain Analysis Image Processing for Adherent MOG Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image processing systems in combine harvesters face challenges in accurately monitoring the percentage of Material Other Than Grain (MOG) due to adherent MOG particles on the camera window, which cause signal noise and reduce the accuracy of harvesting machine control.
Innovation Solution
A method that involves electronically processing images to detect invariant reflectance pixels across a series of images, identifying adherent MOG by monitoring reflectance variance, and generating an invariant pixel image to exclude noise from processing, thereby improving the accuracy of MOG assessment and reducing signal noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a camera is used to capture images of crop continuously through a window in the crop movement path, then the ability to monitor MOG content is improved, but signal noise is introduced due to adherent MOG particles on the window
Solution Approach 1:
The patent extracts and removes the harmful adherent MOG particles from the image processing by identifying pixels with invariant reflectance values across multiple images and excluding them from MOG content calculations. This separates the useful measurement data from the harmful noise caused by adherent particles.
Solution Approach 2:
The system continuously monitors reflectance values of pixels across a series of images and uses this feedback to identify which pixels correspond to adherent MOG. By comparing reflectance values across multiple time points, the system adapts its filtering to remove noise while preserving real MOG content information.
2Measurement precision
If the camera captures images at regular intervals to assess average MOG levels, then the ability to detect MOG is improved, but transient features in individual images are misinterpreted as errors
Solution Approach 1:
The patent performs preliminary analysis by capturing a series of images at regular intervals before final MOG assessment. By pre-processing the images to identify invariant pixels across multiple time points, the system prepares the data to distinguish between transient features and actual adherent MOG, preventing misinterpretation of temporary variations as errors.
3Productivity
If MOG particles adhere to the camera window, then the camera continues to capture images, but the accuracy of image-based monitoring is reduced
Solution Approach 1:
The system uses the image data itself to identify and correct the problem of adherent MOG. By analyzing reflectance patterns across multiple images, the system automatically detects which pixels are affected by adherent particles and excludes them from processing, allowing continuous monitoring without manual intervention or window cleaning.
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 method effectively identifies and eliminates the impact of adherent MOG, enhancing the accuracy of MOG content measurement and improving the control of harvesting machines by reducing signal noise and maintaining the reliability of image-based monitoring systems.
Implementation Method 1
monitoring the reflectance of one or more of the corresponding pixels
Data Source
Figure 1
Figure 2a~2d
Figure 3~4
AI summary
A method is provided for analysing images (24a) of bulk material (12), such as harvested crop material using a camera (11) for taking a series of multipixel images (24a) of samples of the bulk material (12). The taken images are processed electronically for determining therefrom properties of the bulk material. The invention provides for detecting one or more errors in the series of pixel images (24a), comprising the substeps of: a) monitoring the reflectance of one or more of the pixels during sequential processing of each image (24a) and, if the reflectance value of a said pixel is invariant or substantially invariant in more than a predetermined number of consecutive said images, b) identifying an error.