Agricultural Field Boundary Detection via Spatiotemporal Superpixel Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in accurately determining field boundaries of agricultural fields from available images, leading to incorrect yield predictions and decision-making due to inaccuracies in data representation.
Innovation Solution
The system processes images of agricultural fields using spatiotemporal segmentation and filtering combined with regionalization techniques to define accurate field boundaries. This involves accessing a dataset of time-series images, generating a data structure, segmenting images into segments, aggregating data into super pixels, merging super pixels into spatially contiguous regions, and defining field boundaries based on these regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing methods are used to determine field boundaries, then the process is simpler and faster, but the accuracy of boundary detection is insufficient leading to incorrect yield predictions
Solution Approach 1:
The patent applies segmentation by dividing the agricultural field images into multiple segments based on spatial contiguity and temporal characteristics. This allows the system to process complex boundary detection by breaking down the image data into manageable segments that can be analyzed individually and then reassembled into accurate field boundary definitions
Solution Approach 2:
The patent introduces a temporal dimension to the image processing by analyzing time-series images alongside spatial data. This spatiotemporal approach adds a fourth dimension (time) to the traditional 2D spatial analysis, enabling the system to detect boundaries more accurately by observing changes and patterns across multiple time points
2Measurement precision
If more detailed image analysis is performed to improve boundary accuracy, then yield prediction accuracy improves, but processing time increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing the time-series images to create a data structure that organizes image data spatially and temporally before boundary detection. This preliminary organization of data into segments and super-pixels enables faster and more accurate boundary detection without requiring excessive processing time during the actual analysis phase
3Reliability
If images are processed to define accurate field boundaries, then data representation accuracy improves, but computational resources required increase
Solution Approach 1:
The patent merges similar or adjacent super-pixels into spatially contiguous regions that represent complete fields. This merging process consolidates computational results by combining the analysis of multiple segments into unified field representations, reducing the total computational resources needed while maintaining high data representation accuracy through the systematic combination of segmented data
Data Source
AI summary
Systems and methods are provided for use in processing images related to boundaries. One example computer-implemented method includes accessing a data set of images, where the images define a time series of images over a time period and where each of the images includes a target agricultural field, and generating a data structure for the target agricultural field including at least a portion of the images. The method also includes segmenting the images in the data structure into a plurality of segments and aggregating data for the images included in each of the plurality of segments into super pixels. The method then further includes merging the super pixels into spatially contiguous regions and defining a field boundary of the target agricultural field, based on the spatially contiguous regions.