Image Transformation Grid Homography Matrix Storage

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image transformation processing methods, such as those in Japanese Patent No. 4657367, achieve high-speed processing but compromise calculation accuracy, while software-based methods using homography matrices for higher accuracy result in increased data storage requirements.

Innovation Solution

An image processing device that stores coordinate values for a pre-transformed image, selects nearby grid points in the transformed image, and calculates corresponding pre-transformed coordinates using an inverse transformation matrix, reducing the number of parameters needed and achieving high-speed and accurate transformations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If linear interpolation is used for coordinate transformation, then processing speed is improved, but calculation accuracy deteriorates

Engineering Contradiction:

Improveprocessing speedVSAvoidcoordinate calculation accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The image is divided into multiple grid regions, and for each region, a dedicated homography matrix is calculated and stored. This segmentation allows the use of accurate projective transformation within each local region while maintaining overall processing efficiency through regional decomposition.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent changes the transformation parameter representation from simple linear coefficients to homography matrix parameters (8 parameters per grid). This parameter change enables more accurate projective transformation while the grid-based organization maintains manageable data storage requirements.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If homography matrix is used for each grid, then coordinate calculation accuracy is improved, but data storage requirements increase

Engineering Contradiction:

Improvecoordinate calculation accuracyVSAvoiddata storage amount

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

By dividing the image into grids and storing homography matrices only at grid apexes rather than for every pixel, the patent reduces data storage requirements. Each grid shares the homography matrix information from its apex points, eliminating redundant storage.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent calculates and stores homography matrices only at critical grid apex positions rather than for all pixels. This partial action approach provides sufficient accuracy for the entire grid region while significantly reducing the total data storage requirement compared to full-per-pixel transformation data.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS10325345B2Device for performing image transformation processing and method thereof

Publication Date: 2019.06.18 CANON KK

AI summary

An image processing device that transforms an image includes a storage unit configured to store coordinate values in a coordinate system of a pre-transformed image, each of the coordinate value corresponding to a grid point in a coordinate system of a transformed image, a selection unit configured to select a grid point in the vicinity of a target pixel in the transformed image, a calculation unit configured to refer to the pre-transformed coordinate value, corresponding to the selected grid point and stored in the storage unit, and to calculate a pre-transformed coordinate value corresponding to the target pixel, using the referred coordinate value, and an output unit configured to output a pixel value of the coordinate value calculated in the pre-transformed image, as a pixel value of the target pixel in the transformed image.