Check Image Duplicate Detection via Pixel Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital check processing systems face challenges in accurately and efficiently detecting duplicate transactions, leading to increased costs and customer dissatisfaction due to false positives and false negatives, caused by incomplete image analysis and variations in check formats and capture devices.
Innovation Solution
A method and system that processes and compares check images by scaling them to a common resolution, selecting relevant pixel regions, and performing pixel-by-pixel subtraction to determine if images depict the same subject, reducing the need for second-level reviews and intelligent character recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If OCR technology and intelligent recognition are used to analyze check images, then duplicate detection accuracy is improved, but processing time and computational resources increase significantly
Solution Approach 1:
The patent segments the check image analysis into multiple levels: first-level automated analysis using simple image comparison algorithms for quick processing, and second-level OCR-based analysis only for images that fail the first level. This segmentation allows most images to be processed quickly while maintaining high accuracy for ambiguous cases.
Solution Approach 2:
The patent applies different analysis qualities to different regions and cases: simple pixel-based comparison for clear-cut cases, and more sophisticated OCR-based character recognition only for regions or cases where the simple method fails. This local quality approach optimizes resource allocation by applying heavy processing only where necessary.
2Reliability
If comprehensive image analysis is performed to accurately detect duplicates, then false positives are reduced, but processing costs and computational resources increase
Solution Approach 1:
The patent implements a dynamic, two-level processing system where the analysis depth adjusts based on the initial image comparison results. Images that pass the first-level comparison are processed quickly with simple algorithms, while only those that fail are subjected to the more complex second-level OCR analysis, optimizing both reliability and resource usage.
Solution Approach 2:
The patent introduces an intermediary first-level image comparison process that acts as a filter between the raw image input and the expensive OCR-based second-level analysis. This intermediary layer screens out most images that don't require deep analysis, reducing the overall system complexity and resource requirements while maintaining high detection reliability.
3Measurement precision
If pixel-by-pixel comparison is performed on full-resolution images, then comparison accuracy is improved, but processing speed decreases
Solution Approach 1:
The patent applies partial action by performing pixel-by-pixel comparison only on scaled-down versions of images for the first-level analysis, and only applying full-resolution detailed analysis to images that fail the initial comparison. This partial approach maintains sufficient accuracy for the majority of cases while dramatically improving processing speed.
Data Source
AI summary
A system and method for comparing digital images, such as checks images used by banks, includes receiving and processing the images to be compared, including scaling the images to a common resolution, as well as filtering them to remove spot noise, background pels, and other non-information carrying elements. One or more regions of each image are selected for comparison. The selected regions are compared to one another by subtracting the pels of one image from the other's pels. A determination is made of whether the two or more images are duplicates of one another, or depict a substantially identical subject, based on the results of the subtractions. Furthermore, the amount of filtering and scaling may be adjusted to enhance the effects of the system to take advantage of common characteristics that may be known or detected in a particular set of images to be compared.