Image Processing Apparatus Gradation Region Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately identifying gradation regions in images, as they are often confused with low-contrast texture and edge regions, and are affected by noise, leading to inefficient noise-reduction processing.
Innovation Solution
An image processing apparatus that calculates dispersion and average values in multiple pixel groups with the pixel of interest and its peripherals, arranged in different directions, to determine whether the pixel belongs to a gradation region based on specific threshold conditions, applying appropriate noise-reduction processing accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spatial frequency calculation by discrete cosine transformation is used to identify gradation sections, then gradation regions can be identified, but edge sections and flat sections are often misidentified due to noise and low-contrast textures
Solution Approach 1:
The image processing method segments the analysis by dividing pixel groups into different directional orientations (first direction and second direction perpendicular to the first). This segmentation allows independent calculation of first and second values for each direction, enabling more precise characterization of local image structures and reducing misidentification of edge and flat sections as gradation regions.
Solution Approach 2:
The invention introduces a directional dimension by analyzing pixel groups in multiple orientations (first direction and second direction). Instead of single-value analysis, the method calculates values in two perpendicular directions and compares their magnitudes, adding orientational information that helps distinguish true gradation regions from edge regions and low-contrast textures.
2Object-affected harmful factors
If noise-reduction processing is applied to all regions, then noise can be reduced, but processing time increases and gradation regions may be over-smoothed
Solution Approach 1:
The method applies local quality by determining noise-reduction processing parameters specifically for gradation regions based on the relationship between first and second values. Only pixels identified as belonging to gradation regions receive specific noise-reduction processing, while other regions (edge sections, flat sections) are processed differently or not at all, thereby reducing overall processing time and preventing over-smoothing of important image features.
Solution Approach 2:
The image processing system performs self-service by automatically identifying gradation regions through the directional value comparison method and autonomously applying appropriate noise-reduction processing only to those regions. This self-identification and selective processing eliminates the need for manual region specification and optimizes processing efficiency without requiring external intervention.
3Measurement precision
If multiple pixel groups in different directions are analyzed to improve gradation identification, then identification accuracy improves, but calculation complexity increases
Solution Approach 1:
The calculation complexity is managed through segmentation by dividing pixel groups into directional subsets (first direction and second direction). Each segmented group is processed independently to calculate its value, and the results are compared. This segmentation approach organizes the calculation workload systematically and reduces overall complexity compared to analyzing all pixels simultaneously without directional grouping.
Data Source
AI summary
An image processing apparatus includes: a dispersion calculation unit for calculating a dispersion of pixel values in each of a plurality of pixel groups which are each composed of a pixel of interest and peripheral pixels around the pixel of interest in an image and in which the pixel of interest and the peripheral pixels are arranged in directions different from one another; and a gradation determination unit for determining whether or not the pixel of interest belongs to a gradation region on the basis of a magnitude relationship among the dispersions calculated by the dispersion calculation unit.