Image Quality Assurance System for Post-Capture Document Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image quality assurance methods for check images are inefficient, particularly when dealing with large volumes of checks, as they require sequential examination of image quality flag data for defects, wasting processing time on images without defects and failing to consolidate common defects or detect temporal defects effectively.
Innovation Solution
An improved image quality assurance system that analyzes image metrics against preselected thresholds, generates image quality flags for defective images, and stores these flags in a separate file, allowing for faster identification of suspect images and consolidating defects across multiple renditions, while also supporting the detection of temporal defects and providing a summary of batch-level quality issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sequential examination of image quality flag data is used for each check image, then complete quality assurance is achieved, but processing time increases significantly for large volumes of checks
Solution Approach 1:
The system performs preliminary analysis by creating an image quality flag file that pre-identifies suspect images before detailed examination. This preliminary action allows the system to skip thorough analysis of known good images, significantly reducing processing time while maintaining complete quality assurance for flagged suspects.
Solution Approach 2:
The image quality assurance process is segmented into two stages: (1) rapid flagging of suspect images using preliminary metrics, and (2) detailed examination only of flagged images. This segmentation separates the bulk processing of all images from the intensive analysis phase, reducing overall processing time while ensuring thorough quality checks where needed.
2Measurement precision
If all image renditions are examined individually for defects, then comprehensive defect detection is achieved, but processing efficiency decreases
Solution Approach 1:
The system merges defect detection across multiple image renditions by consolidating quality flags in a centralized image quality flag file. This allows the system to detect common defects that appear across multiple renditions and temporal defects that evolve over time, improving detection accuracy without requiring separate examination of each rendition.
Solution Approach 2:
The image quality flag file serves multiple functions: it stores quality metrics for individual images, consolidates common defects across renditions, detects temporal defects, and provides batch-level quality summaries. This multi-functionality allows comprehensive defect detection through a single unified system rather than multiple separate examination processes.
3Reliability
If detailed analysis is performed on every image rendition, then all defects are detected, but processing time is wasted on images without defects
Solution Approach 1:
The system extracts only the suspect images identified by quality flags from the full set of image renditions for detailed analysis. By separating the flagged suspect images from the large volume of good images, the system performs comprehensive defect detection only where needed, eliminating wasted processing time on images without defects while maintaining complete reliability for identifying all actual defects.
Data Source
AI summary
An image quality assurance method for identifying image defects in a plurality of previously-imaged documents, where each previously-imaged document includes at least one image rendition stored in at least one image rendition file according to type of image rendition. The image renditions are analyzed to provide selected image metrics, the latter of which are compared against preselected image quality metric threshold values. An image quality flag is generated for any image rendition if at least one of the selected image metrics for that image rendition does not successfully compare against the preselected image quality metric threshold values. A record entry is created in an image quality flag file for those documents having at least one flagged image rendition, the record entry including information about the selected image metrics of the at least one flagged image rendition for that imaged document. Image defects in a plurality of imaged documents can be identified by examining the record entries in the image quality flag file.