Character Recognition Using Minimum Intensity Curve
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional character recognition technologies face challenges in accurately detecting character segmenting positions, especially when characters are affected by noise or blurriness, leading to incorrect border detection and reduced recognition performance.
Innovation Solution
A character recognition apparatus and method that uses a minimum intensity curve to accurately determine character border positions by calculating the space between characters, excluding false detections due to noise or blots, without relying on predetermined thresholds, by scanning pixel rows perpendicular to the character string and comparing with standard character widths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a predetermined threshold is used to detect character segmenting positions, then the detection process is simple and fast, but the accuracy deteriorates when characters are blurred or affected by noise
Solution Approach 1:
The patent changes the parameter used for detection from a fixed threshold value to a dynamic minimum intensity value derived from the projection data itself. By calculating the minimum intensity value in each pixel row and using it to detect character borders, the system adapts to varying character conditions (sharp or blurred) without requiring manual threshold specification, thereby maintaining both speed and accuracy.
2Device complexity
If a fixed threshold is specified for character segmentation, then the device complexity is low, but the reliability deteriorates under varying character conditions
Solution Approach 1:
The system performs self-service by automatically determining the detection parameter (minimum intensity value) from the input image data itself, without requiring external specification or manual adjustment. The character segmentation process uses the projection data's own characteristics to identify character borders, making the system self-adapting and reliable across different character conditions.
3Ease of manufacture
If projection data with predetermined thresholds is used for segmentation, then the processing is straightforward, but false detections occur due to noise or blots
Solution Approach 1:
The patent extracts the minimum intensity value from the projection data as a distinctive feature that characterizes character borders. By focusing on this extracted feature rather than using a general threshold, the method isolates the relevant information for segmentation while filtering out noise and blot interference that would otherwise cause false detections.
Data Source
AI summary
A character recognition apparatus may include an imaging element configured to read a character string placed on an information recording medium; an image memory configured to store image data of the character string; and a character segmenting unit configured to segment a character constituting the character string. The character segmenting unit may include a minimum intensity curve creating unit configured to detect a minimum intensity value among light intensity values, and create a minimum intensity curve of the image data according to the minimum intensity value of each pixel row; a character segmenting position detecting unit configured to calculate a space between the characters neighboring in the created minimum intensity curve, in order to detect a character segmenting position between the characters; and a character segmenting process unit configured to segment each character according to the detected character segmenting position between the characters.