Character String Region Identification Using Format String Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing devices face challenges in accurately identifying character string regions due to noise and positional deviations, leading to false detections, especially when characters span multiple lines.
Innovation Solution
An image processing device and method that utilize format character strings to restrict recognition candidates by assigning character types and relative positional relationships, calculating similarity between recognition results and format character strings, and determining groups that match the search condition to exclude false detections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If character recognition is performed on images with noise and positional deviations, then character string regions can be detected, but false detections occur and recognition accuracy decreases
Solution Approach 1:
The patent applies preliminary action by extracting character string regions and performing character recognition before the final determination step. The system pre-processes the image to identify candidate regions and performs initial recognition, then uses the format character string comparison to validate and filter out false detections, thereby improving accuracy while reducing false positives.
Solution Approach 2:
The patent implements feedback by comparing the recognized character strings against the format character strings and using this comparison result to determine whether the detected region is valid. The similarity calculation and determination based on format string matching provide a feedback mechanism that corrects and refines the initial recognition results, eliminating false detections and improving overall reliability.
2Quantity of substance
If character strings across multiple lines are identified, then more information can be captured, but it becomes difficult to remove falsely-detected character string regions
Solution Approach 1:
The patent applies segmentation by processing character strings line by line or in discrete regions. The system divides the character string region into manageable segments and processes each according to the format character string assignments, making it easier to identify and remove false detections while preserving valid multi-line information through structured comparison.
Solution Approach 2:
The patent uses feedback by comparing each character string region against its corresponding format character string. This comparison provides a validation mechanism that enables the system to distinguish between valid multi-line character strings and false detections, allowing for effective removal of erroneous regions while maintaining accurate capture of legitimate information.
3Measurement precision
If format character strings with character type assignments are used, then recognition candidates can be restricted and accuracy improved, but device complexity increases
Solution Approach 1:
The patent applies parameter changes by using format character strings that assign specific character types to different positions. Instead of processing all possible characters uniformly, the system changes the parameters (character type constraints) based on the format string assignments, which restricts the search space and improves recognition accuracy without requiring complex processing for every possible character combination.
Solution Approach 2:
The patent implements preliminary action by pre-defining the format character strings with character type assignments before the actual recognition process. This preliminary setup establishes the constraints and guidelines that will be used during recognition, simplifying the overall process by having the format strings prepared in advance rather than dynamically determining all constraints during processing.
Data Source
AI summary
An image processing method for identifying a region in an input image by character recognition, the region coinciding with a predetermined search condition, includes receiving the search condition, the search condition including assignments of plural format character strings, each format character string including an assignment of a character type or a specific character for each character of a recognition target, extracting a character string region becoming a candidate from the input image, calculating a similarity between a character recognition result and the plural format character strings with respect to each group of plural character string regions, the character recognition result being of each character string region included in each group, and determining the group coinciding with the search condition among the groups of plural character string regions according to the calculated similarity.