Image Processing Apparatus Line Noise Removal Text Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods fail to effectively remove line noise from images without distorting the edge shape of characters, especially when line noise overlaps with textures like halftone dots, leading to poor edge recognition.
Innovation Solution
An image processing apparatus that includes a line noise detection section, a segmentation process section to identify text regions, and a line noise removal process section which excludes pixels within a certain distance from text regions to replace line noise pixels, ensuring that replacement pixels are chosen from areas outside the text region to maintain edge integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If line noise removal is performed by replacing noise pixels with interpolated values from neighboring pixels, then line noise is removed, but edge shapes of characters become distorted
Solution Approach 1:
The patent segments the image into different regions (text regions and non-text regions) using a segmentation process. By identifying which regions contain text and which contain halftone dots, the system can apply different processing strategies to each region, preserving edge shapes in text regions while removing line noise in non-text regions.
Solution Approach 2:
The patent applies different line noise removal strategies to different regions of the image based on their local characteristics. In non-text regions (halftone dot regions), line noise is removed using pixel replacement. In text regions, edge preservation is prioritized. This local differentiation allows the system to maintain high edge shape accuracy while effectively removing line noise where appropriate.
2Object-affected harmful factors
If pixels in text regions are used for replacement to remove line noise, then line noise is removed, but poor edge shapes occur in characters
Solution Approach 1:
The patent performs preliminary segmentation to identify text regions before executing the line noise removal process. This preliminary action allows the system to exclude text region pixels from being used as replacement sources, preventing the introduction of poor edge shapes while still enabling effective line noise removal from non-text regions.
3Device complexity
If all line noise pixels are replaced without considering text regions, then line noise removal is simplified, but edge shapes of characters are distorted
Solution Approach 1:
The patent divides the image processing into distinct stages: first segmenting the image into text and non-text regions, then applying line noise removal only to non-text regions. This segmentation approach adds minimal complexity while significantly improving edge shape preservation, as the system only performs replacement operations where they are safe to do so.
Data Source
AI summary
An image processing apparatus includes: a line noise detection section for detecting line noise pixels included in input image data; a segmentation process section for determining a pixel in a text region; and a line noise removal process section for (i) determining a replacement-target line noise pixel by excepting, from the line noise pixels, the pixel in the text region and a pixel within a first distance from the pixel in the text region and (ii) replacing the replacement-target line noise pixel with another pixel other than the line noise pixels, which another pixel is included in the input image data.