Document Image De-skewing and Cropping via Corner Point Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for de-skewing and cropping images of rectangular target objects, such as checks, are error-prone, time-consuming, and difficult for non-trained users to perform accurately, especially in financial transaction contexts where precision is crucial.
Innovation Solution
A method that determines only the first and second dimensions of a rectangular target object by allowing users to select corner points on a user interface, calculating the de-skew angle and dimensions using coordinate points, and providing a visual cue for accurate alignment, thereby simplifying the process and reducing user error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional de-skewing and cropping methods are used, then image alignment precision can be improved, but the operation time and complexity increase significantly
Solution Approach 1:
The patent combines de-skewing and cropping operations into a single integrated process. The system determines the rectangular bounding box and de-skew angle simultaneously through one user interaction sequence (selecting two diagonal corner points), rather than requiring separate operations. This merging eliminates the need for iterative adjustment and multiple steps, resolving the contradiction by achieving both precision and speed through unified processing.
Solution Approach 2:
The system performs preliminary calculation of the de-skew angle and bounding box parameters based on the user's corner point selections before executing the actual de-skewing and cropping. By pre-calculating transformation parameters (rotation angle, crop coordinates) from the selected points, the system prepares all necessary data in advance, enabling immediate execution without iterative adjustment and reducing overall operation time while maintaining precision.
2Manufacturing precision
If manual de-skewing and cropping operations are performed, then image accuracy can be improved, but the ease of operation deteriorates due to complexity
Solution Approach 1:
The patent merges multiple complex operations (de-skewing, cropping, bounding box adjustment) into a single simplified user action of selecting two diagonal corner points. This consolidation reduces the number of steps from multiple manual adjustments to a single intuitive gesture, making the operation accessible to non-trained users while maintaining high accuracy through automated calculation of transformation parameters.
Solution Approach 2:
The system performs self-service by automatically calculating the de-skew angle, bounding box dimensions, and crop coordinates based on the user's corner point selections. The automated computation of transformation parameters eliminates the need for users to manually adjust multiple controls or understand complex alignment procedures, thereby improving ease of operation while preserving accuracy through precise mathematical calculations.
3Reliability
If multiple steps are used for de-skewing and cropping, then processing thoroughness can be improved, but device complexity increases
Solution Approach 1:
The patent merges de-skewing and cropping into a single integrated processing pipeline that computes all transformation parameters simultaneously from two corner points. The system calculates the rotation angle, bounding box coordinates, and crop region in one unified mathematical model, eliminating the need for separate processing steps and reducing overall process complexity while maintaining thoroughness through comprehensive parameter determination.
Solution Approach 2:
The system changes the approach from multiple stepwise parameter adjustments to a single parameter set determination based on corner point coordinates. By deriving the de-skew angle, bounding box dimensions, and crop coordinates simultaneously from the selected points using geometric calculations, the system reduces process complexity while ensuring processing thoroughness through consistent mathematical relationships among all parameters.
Data Source
AI summary
An image of a rectangular target is resolved. First and second dimensions for the rectangular target are determined from an initial image. A cropped and de-skewed final image for the rectangular target is produced responsive to the first and second dimensions.