Satellite Image Field Delineation via Super-Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for delineating agricultural field boundaries using satellite imaging are inaccurate, outdated, and costly, failing to provide the high-precision, low-cost, and up-to-date data required for precision farming, especially for small fields, due to limitations in spatial resolution and frequent updates.
Innovation Solution
A method utilizing a computer system that processes multitemporal, multispectral satellite image sequences through pre-processing, super-resolution techniques, and artificial neural networks to generate high-resolution images and accurately classify pixel positions as part of an agricultural field or not, with the ability to retrain the network for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual hand-digitization is used to create field boundaries, then boundaries can be created, but the process is non-scalable and inaccurate due to resource and time intensiveness
Solution Approach 1:
The patent replaces manual mechanical digitization with an automated image processing system using satellite imagery and machine learning algorithms. The system automatically detects field boundaries by analyzing satellite images, eliminating the need for manual GPS tracking and hand-digitization, thereby achieving both high accuracy and scalability.
Solution Approach 2:
The system enables self-service by allowing field boundaries to be automatically generated without human intervention. The machine learning model processes satellite images autonomously to create accurate boundary delineations, making the process scalable to millions of fields without requiring human resources.
2Reliability
If existing large scale field boundary data is used, then data availability is improved, but the data becomes outdated and inaccurate over time
Solution Approach 1:
The patent implements periodic action by using multitemporal satellite imagery to repeatedly observe and update field boundaries at different time points during the growing season. This allows the system to detect changes in field boundaries over time and maintain up-to-date boundary information without requiring continuous manual updates.
Solution Approach 2:
The system performs preliminary action by detecting field boundaries early in the growing season and continuously monitoring for changes. This allows farmers and authorities to have accurate boundary information available before critical decisions such as subsidy distribution or input application, preventing the use of outdated data.
3Measurement precision
If high-resolution satellite imagery is used to improve field boundary detection, then accuracy is improved, but the cost increases significantly
Solution Approach 1:
The patent applies segmentation by dividing the satellite imagery processing into manageable segments or tiles. This allows the system to process high-resolution imagery efficiently by breaking it down into smaller regions, reducing the computational burden while maintaining the ability to detect fine boundary details across large areas.
Data Source
AI summary
A computer-based system (210) for delineating agricultural fields based on satellite images includes a first subsystem (201) configured to receive at least one multitemporal, multispectral satellite image sequence (101) and pre-processing the images in the at least one multitemporal, multispectral satellite image sequence to generate a pre-processed image sequence (303) of multitemporal multispectral images covering a specific geographical region; a second subsystem (202) configured to perform a super-resolution method on the images in the pre-processed image sequence to generate a high-resolution image sequence (403) of multitemporal multispectral images where corresponding pixel positions in images in the sequence relate to the same geographical ground position; and a third subsystem (203) including a delineating artificial neural network (501) trained to classify pixel positions in the high-resolution image sequence (403) as being associated with a geographical ground position that is part of an agricultural field (104) or not part of an agricultural field.