Character String Region Detection in Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying the standard region from a target object image is challenging due to varying formats of target objects with standard character strings, especially when the image is unclear or not photographed from the front side, leading to inaccurate character recognition.
Innovation Solution
An image processing system comprising a detection module to identify character string regions, a character recognition module to calculate recognition scores for each character, and a standard region determination module to determine if the character string region is a standard region based on the calculated scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If character recognition is applied to the target object image to identify the standard region, then the standard region can be identified from the image, but accurate character recognition cannot be executed when the image is unclear or the object is not photographed from the front side
Solution Approach 1:
The patent segments the character recognition process into two independent stages: first detecting character string regions using template matching, then performing character recognition only on detected regions. This segmentation allows the system to handle unclear images by separating the region detection task (which works with lower precision) from the character recognition task (which requires high precision), thereby resolving the contradiction between measurement precision and reliability
Solution Approach 2:
The patent applies preliminary action by first detecting character string regions using template matching before performing character recognition. This preliminary region detection step prepares the data structure and identifies potential areas of interest, enabling the subsequent character recognition to focus only on relevant regions. This preliminary preparation ensures that even in unclear images, the system can reliably identify where standard characters should appear
2Device complexity
If template matching is used to identify the standard region, then the identification process can be simplified, but it becomes difficult to handle target objects with various formats
Solution Approach 1:
The patent implements universality by designing a dual-module system where template matching serves as a universal region detector that can identify character string locations across different document formats, while character recognition acts as a universal verifier that can validate standard characters regardless of their specific position or format. This multi-functional approach allows the system to handle various target object formats without increasing overall process complexity
Solution Approach 2:
The patent introduces character recognition as an intermediary step between template matching and standard region identification. Template matching first detects potential character string regions, then character recognition verifies whether these regions contain standard characters. This intermediary layer acts as a bridge that reconciles the simplicity of template matching with the adaptability needed for various formats, allowing the system to maintain low complexity while achieving high versatility
3Measurement precision
If character recognition is applied to the entire target object image, then the standard region can be identified, but the processing time and computational resources increase significantly
Solution Approach 1:
The patent segments the image processing into two distinct phases: region detection phase using template matching, and character recognition phase applied only to detected regions. This segmentation dramatically reduces processing time by avoiding full-image character recognition, while maintaining identification accuracy because character recognition is concentrated on the specific regions where standard characters are expected to appear
Solution Approach 2:
The patent applies partial action by performing character recognition only on detected character string regions rather than the entire image. This partial processing approach is sufficient to identify standard regions accurately, as character recognition is only needed where character strings are present. The system avoids the excessive computation of full-image recognition while maintaining the necessary precision for standard region identification
Data Source
AI summary
An image processing system, comprising at least one processor configured to: detect a character string region including any character string from a target object image relating to a target object including a standard character string; apply character recognition to the character string region to calculate a score relating to a result of the character recognition for each character included in the character string region; and determine whether the character string region is a standard region including the standard character string based on the score calculated for the each character.