ID Card Image Masking via Linked Item Name-Value Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing apparatuses struggle to accurately mask specific regions of personal information on personal identification cards, as the regions to be masked vary among users, and manual masking is cumbersome.
Innovation Solution
An image processing apparatus with an image reading portion and control portion that employs machine learning models to detect and link regions of item names and values, followed by a region discrimination process using dictionary data to identify and mask target items, all performed locally without cloud-based data transfer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual masking is used, then masking accuracy can be controlled, but operation complexity and time consumption increase
Solution Approach 1:
The system performs automatic region detection and masking without requiring manual intervention. The control portion autonomously detects character regions, identifies item names and values, determines masking targets, and applies masking operations, thereby eliminating the need for manual masking while maintaining accuracy through automated processing
Solution Approach 2:
The patent replaces manual mechanical masking operations with automated image processing techniques. Machine learning models and algorithms automatically detect regions, identify content, and apply masking, substituting human manual operations with computational processes that achieve both accuracy and efficiency
2Power
If cloud-based processing is used, then processing power is increased, but data security risks increase
Solution Approach 1:
The system extracts and processes only the necessary image data locally on the device, extracting essential features and performing masking operations without transferring the complete personal information data to cloud servers. This minimizes data exposure while maintaining processing capability through local computational resources
Solution Approach 2:
The patent introduces local processing as an intermediary between data input and output, creating a secure processing environment that prevents direct cloud connectivity. The control portion acts as an intermediary that handles all processing locally, eliminating the need for cloud-based processing and associated data leakage risks
3Productivity
If region detection algorithms are simplified, then processing speed is improved, but detection accuracy decreases
Solution Approach 1:
The patent segments the region detection process into multiple specialized stages: character region detection, item name region identification, item value region identification, and masking target determination. Each stage focuses on a specific aspect, allowing the system to maintain high accuracy through specialized processing while overall processing speed is optimized by dividing the complex task into manageable segments
Solution Approach 2:
The system performs preliminary character region detection and item name/value identification before final masking target determination. These preliminary actions prepare and organize the data in advance, enabling faster subsequent processing while maintaining accuracy through structured multi-stage analysis that builds upon previous results
Data Source
AI summary
A control portion performs a first region detection process to detect a first region expected to individually include either a region of an item name or a region of an item value, a second region detection process to detect a second region expected to include both a region of an item name and a region of an item value corresponding to the same item, and a process of linking together a plurality of first regions included in the same second region. The control portion further performs a process of detecting a hand-written region in which characters are printed characters.