Background Color Detection Using Merged Frequency Distributions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods fail to accurately detect and remove background color from selected regions in images, especially when the region corresponding to the original has not been determined and the selected region contains a small amount of background color.
Innovation Solution
An image processing method that calculates frequency distributions of the read and selected regions, detects the background color level by comparing candidates from both regions, and removes the background color based on the detected level, using a scoring system to accurately determine the background color even when the selected region lacks a large amount of background color.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the background-color level is detected in accordance with the density of the selected region, then the processing can be performed without determining the original region, but the background-color level cannot be detected accurately when the selected region does not contain a large amount of background color
Solution Approach 1:
The patent merges the frequency distribution information from both the selected region and the entire read region to detect the background-color level. By combining these two distributions, the system can accurately identify the background color even when the selected region alone contains insufficient background color data, thus resolving the contradiction between ease of operation and measurement precision.
Solution Approach 2:
The patent uses the frequency distribution of the entire read region as an intermediary to assist in detecting the background-color level of the selected region. This intermediary distribution provides additional information that compensates for the insufficient background color data in the selected region, enabling accurate detection without requiring original region determination.
2Device complexity
If a histogram is generated using only the density signal of the selected region, then the processing is simplified, but the histogram does not include a large amount of color that should be detected as the background color
Solution Approach 1:
The patent combines the frequency distribution (histogram) of the selected region with the frequency distribution of the entire read region. This merging ensures that the resulting histogram includes a comprehensive representation of all colors, particularly background colors, while maintaining processing efficiency. The combination resolves the contradiction by preserving both simplicity and reliability.
3Productivity
If the region corresponding to the original is not determined, then the processing speed is improved, but the background color detection accuracy deteriorates when the selected region lacks sufficient background color
Solution Approach 1:
The patent performs preliminary calculation of the frequency distribution for the entire read region before detecting the background-color level of the selected region. This preliminary action prepares additional reference data that can be quickly utilized during background color detection, maintaining high processing speed while ensuring accurate detection even when the original region is not determined.
Solution Approach 2:
The frequency distribution of the entire read region serves as an intermediary that bridges the gap between fast processing (without original region determination) and accurate background color detection. This intermediary provides the necessary information to achieve both high productivity and measurement precision simultaneously.
Data Source
AI summary
In a method for detecting a background-color level in accordance with a selected region, if the selected region does not contain a large amount of background color, the background-color level cannot be detected accurately. Thus, an image signal of a region read by an image reader is input, a part of the read region is selected, a background-color region is detected in accordance with the image signal of the input read region and an image signal of the selected region, and a background color contained in the selected region is removed in accordance with the level of an image signal of the detected background-color region.