RGB-Segmented Image Processing for Character-Preserving Color Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing color reduction methods for images with uneven shading, such as handwritten characters, result in blurred characters due to conversion to background colors, leading to low object reproducibility.
Innovation Solution
An image processing apparatus that determines background and character attributes using RGB histograms to convert pixels into representative colors, ensuring high object reproducibility by distinguishing between background and character pixels during color reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If conventional color reduction processing is applied to images with uneven shading, then data size is reduced, but object reproducibility deteriorates due to character blurring
Solution Approach 1:
The image is segmented into character regions and background regions using binary image processing. This segmentation allows different color reduction strategies to be applied to different regions, preventing character blurring while maintaining data compression. The binary image separates foreground (character) pixels from background pixels, enabling selective color assignment.
Solution Approach 2:
Different color reduction approaches are applied to different regions of the image. Character regions use one color (representative color of the character), while background regions use another color (representative color of the background). This local differentiation ensures that characters remain distinct and readable while achieving overall color reduction.
2Productivity
If color reduction converts all pixels to limited colors, then compression efficiency improves, but character distinctiveness deteriorates
Solution Approach 1:
Before performing color reduction, the patent first generates a binary image to identify character regions. This preliminary action ensures that when color reduction is applied, characters are already distinguished from the background, preventing subsequent blurring and maintaining character distinctiveness in the compressed image.
Solution Approach 2:
The patent changes the color parameters of pixels based on their region classification. Character region pixels are assigned one set of color values (representative character color), while background pixels are assigned different color values (representative background color). This parameter change approach maintains character distinctiveness while achieving compression.
Data Source
AI summary
An image processing apparatus capable of generating a color-reduced image with high object reproducibility is provided. The image processing apparatus includes a data processing unit configured or programmed to function as a generating unit generating a binary image for judging whether each of pixels constituting an image obtained is a background attribute or a character attribute, a first determining unit determining a first type representative color based on a histogram of RGB values of pixels judged to be the background attribute based on the binary image among all the pixels constituting the image, a second determining unit determining a second type representative color based on a histogram of RGB values of pixels judged to be the character attribute based on the binary image among all the pixels constituting the image, and a processing unit performing a color reduction processing on the image to generate a color-reduced image.