Threshold Matrix Generation for Dithering Dot Uniformity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for generating threshold matrices in dithering processes fail to adequately suppress low-frequency components and result in nonuniformity in dot density, leading to visible periodic patterns in dithered images.
Innovation Solution
A method involving a threshold matrix generating device that divides the image area into discrete blocks, applies filtering and pattern optimization using average values, and rearranges dots based on error matrices and uniformity evaluations to create a threshold matrix that effectively suppresses low-frequency components and ensures uniform dot distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If conventional dithering methods are used with traditional threshold matrices, then the calculation time is short, but low-frequency components are not adequately suppressed and dot density nonuniformity occurs
Solution Approach 1:
The threshold matrix generation process is segmented into multiple iterative steps: initial dot pattern generation, spatial filtering application, dot rearrangement based on error diffusion, and uniformity evaluation. Each segment addresses specific aspects of the problem, allowing the system to achieve both speed and precision by processing different aspects in discrete stages rather than using a single complex algorithm.
Solution Approach 2:
The patent implements feedback mechanisms where the dot pattern is evaluated for uniformity after each rearrangement iteration. The error matrix calculated from the difference between target and actual dot patterns feeds back into the rearrangement process, continuously adjusting the pattern to reduce low-frequency components and improve density uniformity while maintaining computational efficiency.
2Manufacturing precision
If error diffusion method is used for quantization, then dot density uniformity can be improved, but the calculation process becomes complicated and time-consuming
Solution Approach 1:
The patent merges two previously separate approaches: dithering with threshold matrices and error diffusion methods. By combining the speed advantage of dithering with the uniformity advantage of error diffusion, the system achieves both computational efficiency and dot density uniformity. The threshold matrix provides the framework while error diffusion principles guide the dot rearrangement within that framework.
Solution Approach 2:
The patent performs preliminary generation of the threshold matrix and initial dot pattern before the main optimization process. This preliminary structure is then refined through iterative rearrangement rather than building the entire solution from scratch, reducing the overall computational complexity while maintaining the ability to achieve uniform dot distribution.
3Productivity
If threshold matrix is repetitively applied to multi-gradation image for dithering, then quantization speed is maintained, but nonuniformity in dot density creates visible periodic patterns
Solution Approach 1:
The patent applies different processing treatments to different regions of the threshold matrix. By dividing the matrix into discrete blocks and evaluating uniformity locally, the system can adjust dot placement in specific regions to eliminate periodic patterns while maintaining overall quantization speed. Each local region is optimized independently based on its specific characteristics.
Solution Approach 2:
The patent dynamically adjusts parameters such as the threshold values in the matrix and the dot placement patterns based on local uniformity evaluations. By changing these parameters iteratively during the generation process, the system eliminates the fixed periodic patterns that would otherwise be visible in the dithered output, while maintaining the speed advantage of threshold-based methods.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed is a threshold matrix generating method involving: a step for generating q(x, y, g) in which the number of dots in an initial dot pattern is varied; a step for calculating an error matrix (ERR(x, y, g)) of q(x, y, g); a step for calculating AVE(a, b) which represents the homogeneity of the number of dots in small sections into which the dot pattern has been divided; a step for determining the two pixels, of which the dot positions are to be switched, on the basis of ERR (x, y, g) and AVE(a, b); a step for switching the dot positions; a step for calculating the evaluation value (MSE(n)) of q(x, y, g) after the positional switch; a step for repeating the positional switch of the dots until a q(x, y, g) satisfying MSN(n)<MSN(n-1) is obtained; and a step for repeating the generation of q(x, y, g) and the positional switch of the dots with q(x, y, g) functioning as the initial dot pattern.