Hybrid Dither Error Diffusion Quantization for Image Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods, such as dither and error diffusion processes, suffer from texture repetition, dot delay, worm noise, and pattern noise, leading to degraded image quality during halftone printing, especially when switching between these methods based on gray levels.
Innovation Solution
A program that dynamically adjusts the noise use rate and error use rate for each pixel based on its gray level, combining the dither and error diffusion processes to suppress texture, dot delay, and pattern noise by using a dither matrix noise and accumulated error, allowing for bidirectional pixel selection and adaptive threshold calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the dither process is used for quantization, then the quantization process can be performed, but texture patterns are generated and image quality degrades
Solution Approach 1:
The patent combines the dither process and error diffusion process into a hybrid quantization method. The dither matrix noise provides the quantization function while the error diffusion component suppresses the texture patterns generated by pure dithering, achieving both quantization and texture suppression simultaneously
Solution Approach 2:
The patent creates a composite quantization approach by integrating two different quantization methods (dither and error diffusion) into a unified process. This composite method leverages the strengths of both approaches while mitigating their individual weaknesses, particularly the texture issue in dithering
2Object-affected harmful factors
If the error diffusion process is used for quantization, then texture patterns are suppressed, but dot delay and worm noise are generated and image quality degrades
Solution Approach 1:
The patent applies different quantization strategies to different regions or contexts within the image processing. By dynamically adjusting the balance between dither and error diffusion components based on local image characteristics, the method suppresses texture where needed while minimizing dot delay and worm noise in other areas
Solution Approach 2:
The patent modifies the parameters of the hybrid quantization process by adjusting the influence weights of dither matrix noise and accumulated error dynamically. This allows the system to adapt to different image regions and suppress various artifacts (texture, dot delay, worm noise) by changing the parameter balance appropriately
3Ease of manufacture
If dither process and error diffusion process are separately used depending on gray level ranges, then quantization can be performed, but boundary lines are formed and image quality degrades
Solution Approach 1:
The patent introduces dynamic switching between dither and error diffusion processes based on gray level ranges. Instead of static separation, the method dynamically adjusts the contribution of each process, using dither for highlight/shadow regions and error diffusion for mid-tone regions, thereby avoiding boundary lines while maintaining quantization effectiveness
Solution Approach 2:
The patent performs preliminary classification of pixels by gray level range before applying the hybrid quantization. By pre-identifying which regions should use dither versus error diffusion components, the method prevents boundary line formation at the transitions between different quantization approaches
Data Source
AI summary
A computer-readble medium stored with a program causing a computer to quantize a gray level causes the computer to perform a pixel selecting process; determining a noise use rate indicating the degree to which a dither matrix noise has an influence on the quantization and an error use rate indicating the degree to which an accumulated error has an influence on the quantization and determining the noise use rate and the error use rate corresponding to the corresponding pixel depending on the gray levels of the pixels sequentially selected in the pixel selecting process; and performing the quantization on the gray levels of the pixels sequentially selected in the pixel selecting process and performing the quantization using the dither matrix noise based on the noise use rate corresponding to the corresponding pixel and using the accumulated error based on the error use rate corresponding to the corresponding pixel.