Image Binarization with Edge-Guided Reversal Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional binarization methods fail to produce high-quality binary images suitable for character recognition, especially when images contain reversed characters, low contrast between background and character colors, or a combination of character and non-character areas, leading to decreased recognition precision.
Innovation Solution
An image processing apparatus that creates an edge image and binary image, with a partial binarization unit identifying character areas and determining if characters are reversed, reversing and binarizing them as needed to improve recognition precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single binarization threshold is set for the entire surface of an image, then the image can be processed efficiently, but images with low contrast between background and character colors cannot be properly binarized
Solution Approach 1:
The image is divided into multiple blocks, and a binarization threshold is determined for each block based on its local characteristics. This allows the system to handle images with varying contrast regions effectively, as each block can be processed with an optimized threshold suitable for its specific content.
Solution Approach 2:
Instead of applying a uniform threshold across the entire image, the system adapts the binarization threshold to local regions. By calculating thresholds for individual blocks, the system preserves local image characteristics and handles low-contrast regions appropriately while maintaining overall processing efficiency.
2Manufacturing precision
If binarization is performed adaptively for each small block to cope with background color changes, then local image quality improves, but noise increases in blocks containing both character and non-character areas
Solution Approach 1:
Before performing binarization, the system extracts edge information from the image. This edge map is used to identify character areas in advance, allowing the binarization process to focus only on regions containing characters. This preliminary action prevents noise from being generated in non-character blocks while maintaining local quality in character regions.
Solution Approach 2:
The system extracts and separates character area information from the rest of the image using edge detection. By isolating character regions, the binarization process can be applied selectively only where needed, avoiding the generation of noise in blocks that contain only non-character content.
3Device complexity
If conventional binarization is used, then the processing is simple, but reversed characters cannot be recognized and character recognition precision deteriorates
Solution Approach 1:
The system performs edge extraction and character area identification before binarization. This preliminary analysis enables the system to detect reversed characters and apply appropriate processing to correct them, ensuring high recognition precision without significantly increasing overall system complexity.
Solution Approach 2:
The system uses edge information as feedback to guide the binarization process. By analyzing edge patterns, the system can identify reversed characters and adjust the binarization accordingly, creating a feedback loop that improves recognition accuracy while maintaining processing efficiency.
Data Source
AI summary
An image processing apparatus processes an input image, in which an edge image is created by executing an extracting operation on the input image, and a binary image is created by performing a thresholding operation on the input image. A first partial thresholding unit creates an output binary image by extracting a character area from the input image, which corresponds to a character area identified from the created edge image, and performs a thresholding operation on the extracted character area. A second partial thresholding unit creates an output binary image by extracting a character area from the input image, which corresponds to a character area identified from the created binary image, and performs a thresholding operation on the extracted character area. An output unit outputs the output binary image created by said first partial thresholding unit and the output binary image created by said second partial thresholding unit to a memory, wherein the first partial thresholding unit determines whether a character in the extracted character area is a reversed character, and if it is determined that the character is a reversed character, creates an output binary image by reversing and performing the thresholding operation on the character area.