Document Corner Validation for Auto-Capture Image Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Challenges arise in obtaining suitable image data of documents for remote or electronic transactions due to issues like out-of-frame, obstructed, or damaged areas, user unawareness of areas of interest, and the need for automation in analyzing image data without human intervention, especially with varied document types.
Innovation Solution
A document detector system using machine learning to determine document corners, perform validity checks, and provide live feedback for capturing images, ensuring the document is upright, readable, and free from glare, blur, or obstructions, and automating the capture process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual review of document images is performed to ensure quality, then image quality can be verified, but user frustration increases and resources are consumed
Solution Approach 1:
The system performs self-verification by automatically detecting document corners, validating their positions, and determining image quality without requiring manual review. The corner validation system checks whether detected corners meet predefined criteria (e.g., proper spacing, orientation) and automatically rejects or requests retakes, eliminating the need for human operators to verify each image.
Solution Approach 2:
The patent replaces manual mechanical review processes with automated computer vision algorithms. Machine learning models detect document corners and validate image quality, substituting human operators with an automated system that processes images rapidly without fatigue or frustration, thereby maintaining reliability while reducing time loss.
2Productivity
If automated corner detection is performed without validation, then processing speed increases, but detection accuracy decreases
Solution Approach 1:
The system implements feedback loops where detected corners are validated against predefined criteria. If corners fail validation (e.g., improper spacing, incorrect orientation), the system automatically requests a new image capture. This feedback mechanism ensures high detection accuracy by filtering out poor-quality detections while maintaining processing speed through automated validation rather than manual review.
Solution Approach 2:
The patent performs preliminary corner validation checks before final processing. By validating corner positions, spacing, and orientation early in the workflow, the system prevents inaccurate detections from proceeding to subsequent processing stages, thereby maintaining high measurement precision while preserving productivity through automated early filtering.
3Reliability
If multiple validity checks are performed on document images, then image quality improves, but processing complexity increases
Solution Approach 1:
The validation process is segmented into distinct, modular checks: corner detection, corner validation (spacing, orientation), glare detection, blur detection, and obstruction detection. Each check operates independently and can be applied selectively based on document type and capture conditions. This segmentation reduces overall complexity by breaking down the validation process into manageable, reusable components rather than a monolithic complex system.
4Measurement precision
If corner coordinates are used to generate polygons for validation, then document orientation can be determined, but computational requirements increase
Solution Approach 1:
The system performs partial validation by checking only the essential corner properties needed for orientation determination, rather than analyzing the entire document image. By focusing computational resources on validating corner coordinates and generating simplified polygons for orientation checks, the system achieves precise orientation determination with reduced computational requirements compared to full-image analysis.
Data Source
AI summary
The disclosure includes a system and method for receiving, using one or more processors, first image data representing a first image of a document; obtaining, using the one or more processors, a first set of corner coordinates representing the corners of the document in the first image; generating, using the one or more processors, a first polygon based on the first set of corner coordinates; performing, using the one or more processors, one or more corner-based validity checks based on the first set of corner coordinates; and performing, using the one or more processors, an auto-capture.