Satellite Image Terrain Boundary Detection Using Hierarchical Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems face challenges in accurately refining terrain boundaries in satellite images due to variance in resolution and sensor capture, leading to inaccurate results from noise in low-resolution image data.
Innovation Solution
A machine-implemented method that obtains pixel data from satellite images, determines multiple image features based on texture and color statistics, generates initial rules, creates image models, and applies them to generate probability maps and binary label masks to accurately segment different terrains, while smoothing the maps to remove obstacles like cloud or snow coverage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional processing methods are used on low-resolution satellite imagery, then processing speed is maintained, but terrain boundary detection accuracy deteriorates due to noise in the image data
Solution Approach 1:
The patent segments the image processing into multiple hierarchical levels (coarse to fine) and divides terrain classification into multiple categories (water, land, forest, mountain). This segmentation allows the system to handle low-resolution images effectively by processing them in stages, improving accuracy without requiring all processing to occur at once, thus managing complexity.
Solution Approach 2:
The patent applies preliminary actions including noise filtering, thresholding, and initial terrain classification before final boundary determination. By performing these preparatory processing steps in advance, the system prepares the image data to reduce noise and highlight important features, thereby improving detection accuracy in low-resolution images.
2Measurement precision
If multiple image features are extracted and processed through hierarchical refinement, then terrain boundary accuracy is improved, but processing time increases
Solution Approach 1:
The patent implements a hierarchical refinement process that segments terrain classification into multiple levels (coarse classification first, then fine classification). This allows the system to quickly identify major terrain types in low-resolution images and then apply more detailed analysis only where necessary, improving accuracy while managing processing time through staged computation.
Solution Approach 2:
The patent applies partial action by extracting and processing only the most discriminative image features (texture, color statistics) rather than all possible features. This selective feature extraction maintains processing efficiency while providing sufficient information for accurate terrain boundary detection in low-resolution images.
3Measurement precision
If noise filtering and smoothing are applied to low-resolution images, then terrain boundary detection accuracy is improved, but image detail is lost
Solution Approach 1:
The patent applies local quality by implementing adaptive noise filtering and smoothing that adjusts to local image characteristics. Different regions of the image receive different levels of processing based on their content and resolution, allowing noise reduction in areas where it helps boundary detection while preserving important local details where they exist.
Solution Approach 2:
The patent performs preliminary noise filtering and thresholding operations before final terrain classification. By applying these processing steps in advance at appropriate intensity levels, the system reduces noise that would interfere with boundary detection while preserving essential image details through controlled, staged processing rather than aggressive single-pass filtering.
Data Source
AI summary
A system and method for processing a satellite image to determine a terrain boundary. Pixel data corresponding to a satellite image is obtained. For each pixel within a subset of the pixel data, multiple image features are determined corresponding to texture and color statistics for the pixel and neighboring pixels, based on a feature selection process for selecting discriminative features. An initial set of rules is generated for determining a terrain boundary. Plural image models for applying to the pixel data are created based on the initial set of rules. The plural image models are applied to the pixel data to generate at least one probability map, which represents output of the plural image models and corresponds to different terrains within the satellite image. A binary label mask is generated for the satellite image based on the probability map, and mask indicates segmentations corresponding to the different terrains.