Image Processing Device Pixel Grouping Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques fail to accurately group pixels belonging to the same object in images, leading to integration of different objects into the same group and vice versa, resulting in inefficient information usage and real-time processing limitations.
Innovation Solution
An image processing device that groups pixels based on differences in image data and average values within predetermined thresholds, allowing for real-time processing by comparing each pixel with adjacent pixels and their respective groups, effectively utilizing rich image data without compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If pixels are grouped based on color classification numbers obtained from quantized image data, then image processing can be performed with discrete information, but pixels belonging to the same object are not readily integrated into the same group
Solution Approach 1:
The patent changes the grouping parameter from color classification numbers (discrete quantized values) to difference values of image data between adjacent pixels. By using the difference value |Di,j-Di-1,j| and comparing it against a threshold, the system achieves more precise grouping that reflects actual object boundaries rather than arbitrary quantization classes.
Solution Approach 2:
The patent replaces the mechanical quantization process (which loses information) with a direct comparison of image data differences. Instead of converting continuous image data into discrete classification numbers, the system directly processes the difference values to determine grouping, preserving more information and achieving better grouping precision.
2Device complexity
If uniform classification methods are applied to all pixels, then processing is simplified, but different objects with similar color characteristics are integrated into the same group
Solution Approach 1:
The patent applies local quality by using different comparison criteria for different pixel positions. Specifically, it compares image data differences between adjacent pixels locally, and uses the result to determine grouping. This local comparison approach allows the system to adapt to local object boundaries and color transitions, improving object separation reliability while maintaining relatively simple processing.
3Productivity
If image data is quantized to discrete information for compression, then information amount is reduced for efficient processing, but the abundance of information cannot be used effectively
Solution Approach 1:
The patent extracts only the necessary information for grouping - the difference values between adjacent pixels - rather than processing all quantized image data. By focusing on the difference |Di,j-Di-1,j| and using it for threshold-based grouping, the system achieves efficient processing while preserving the essential information needed for accurate object segmentation, avoiding the information loss inherent in aggressive quantization.
Data Source
AI summary
An image processing device for dividing an image imaged by imaging means into multiple regions includes: processing means for grouping, when difference of the pieces of image data between a single pixel within the image and a pixel adjacent thereto is less than a predetermined threshold, the single pixel and the adjacent pixel, and dividing the image into multiple regions with finally obtained each group as each region of the image; and average-value calculating means for calculating the average value of the image data within the group including the single pixel; with the processing means comparing the image data of the single pixel, and the average value calculated at the average-value calculating means regarding the group to which the adjacent pixel belongs; and when the difference thereof is equal to or greater than a predetermined second threshold, doing not group the single pixel and the adjacent pixel.