Geospatial Void Inpainting via Frequency-Spatial Domain Transformation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing geospatial modeling systems face challenges in effectively filling voids in data, particularly in frequency domain data sets, which can lead to distorted models due to interference and hardware malfunctions, and current interpolation techniques often blur edge content or are computationally burdensome.

Innovation Solution

A geospatial modeling system that uses a processor to select and transform reference and test samples from frequency domain to spatial domain, applying inpainting functions based on different boundary geometries and models, such as functional partial differential equations or stochastic models, to iteratively propagate contour data and fill voids, with the option to change inpainting functions based on similarity thresholds and windowing schemes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If conventional interpolation techniques (sinc, polynomial, spline) are used to fill voids in data, then data completeness is improved, but edge content becomes blurred or computational burden increases

Engineering Contradiction:

Improvedata completenessVSAvoidedge content sharpness

Core Design Contradiction:

Loss of informationVSManufacturing precision

Solution Approach 1:

The patent transforms the data from frequency domain to spatial domain, changing the representation parameters. This transformation allows the use of contour propagation methods that preserve edge sharpness while filling voids, avoiding the blurring effect of conventional interpolation techniques in the frequency domain

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces conventional mathematical interpolation methods (sinc, polynomial, spline) with a contour propagation mechanism. Instead of using algebraic interpolation formulas, the system propagates contour lines from known data points into void regions, preserving edge geometry while filling missing data

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If higher order polynomial interpolation is used to improve accuracy, then reconstruction precision is improved, but computational overhead becomes burdensome

Engineering Contradiction:

Improvereconstruction accuracyVSAvoidcomputational overhead

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex higher-order polynomial calculations with a geometric contour propagation approach. By transforming to spatial domain and propagating contour lines, the system achieves high reconstruction accuracy without the computational burden of solving high-order polynomial equations or dealing with ill-conditioned matrices

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediate spatial domain representation as a mediator between the frequency domain data and the final reconstructed model. This intermediate transformation enables accurate void filling through contour propagation while avoiding direct computation of complex interpolation formulas

Inventive Principle:

Principle #24Intermediary (Mediator)

3Measurement precision

If global spline interpolation is applied over the entire model to improve accuracy, then reconstruction precision is improved, but implementation difficulty increases due to ill-conditioned matrices

Engineering Contradiction:

Improvereconstruction accuracyVSAvoidimplementation difficulty

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the data processing into distinct domains: frequency domain transformation to spatial domain, contour propagation in spatial domain, and final reconstruction. This segmentation avoids the need to solve global spline equations over the entire model, eliminating the ill-conditioned matrix problem while maintaining accuracy through localized contour propagation

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS7949203B2Geospatial modeling system providing void inpainting based upon selectable inpainting functions and related methods

Publication Date: 2011.05.24 HARRIS CORP

AI summary

A geospatial modeling system may include a geospatial model data storage device and a processor. The processor may cooperate with the geospatial model data storage device for selecting and transforming a reference sample of a geospatial model frequency domain data set into a corresponding reference sample geospatial model spatial domain data set, and inpainting data into at least one void of the geospatial model frequency domain data set based upon an initial selected inpainting function from among a plurality of different inpainting functions. The processor may further select and transform a test sample of the inpainted geospatial model frequency domain data set into a corresponding test sample geospatial model spatial domain data set, and compare the reference sample geospatial model spatial domain data set and the test sample geospatial model spatial domain data set to determine whether to repeat the inpainting using a different inpainting function from among the plurality thereof.