Dual Lookup Table Image Correction for Printers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems face challenges in simultaneously performing uniform corrections, such as color balance correction, and local corrections, like dodging processes, on images with limited hardware resources, particularly in printers with constrained CPU, RAM, and memory, leading to increased complexity and calculation overhead.
Innovation Solution
A configuration with two interpolation units connected in series, where one unit performs uniform correction and the other's interpolation intensity is controlled using control parameters, allowing for simultaneous uniform and local corrections with a simple hardware setup.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multiple correction operations (color balance correction, luminance gradation correction, saturation correction) are performed on each pixel of the entire image, then comprehensive image correction is achieved, but the calculation amount becomes extremely large (29 multiplications per pixel, 174 million multiplications for a 3000*2000 image)
Solution Approach 1:
The patent segments the image correction process into two distinct stages: (1) uniform correction applied to the entire image using a rotation matrix for color balance adjustment, and (2) local correction applied only to specific regions using a dodging process with spatially varying luminance adjustment. This segmentation reduces the overall calculation burden by separating global operations from localized operations.
Solution Approach 2:
The patent applies local quality by implementing a dodging process that adjusts luminance differently across spatial regions. A low-frequency image (blurred version of the original) is generated and used as a spatial brightness control factor, allowing different correction strengths in different areas. This enables targeted correction of specific regions (e.g., darkening foreground subjects while preserving background exposure) without processing every pixel with the full correction pipeline.
2Productivity
If a 3DLUT system is used for image processing, then processing speed is improved and calculation amount is reduced, but the ability to perform both uniform correction and local correction simultaneously becomes difficult to control
Solution Approach 1:
The patent segments the correction operations into distinct uniform and local components that can be independently controlled and optimized. The uniform color balance correction uses a simple rotation matrix operation, while the local dodging process uses a separately generated low-frequency image. This segmentation allows each component to be efficiently implemented without the complexity of a full 3DLUT system.
Solution Approach 2:
The patent changes parameters by using a low-frequency image (spatially varying luminance map) as a control parameter for the dodging process. This parameter varies across space but can be pre-computed, allowing efficient local correction without requiring complex 3DLUT structures. The rotation matrix parameters for color balance are also explicitly calculated and applied uniformly.
3Manufacturing precision
If hardware resources (CPU speed, RAM capacity, access speed) are increased to handle complex image processing, then image processing capability is improved, but printer cost increases and hardware resources remain limited compared to personal computers
Solution Approach 1:
The patent extracts only the essential correction operations needed for quality image processing: a rotation matrix for color balance and a dodging process for local luminance adjustment. By removing unnecessary complex operations and focusing on these two key corrections, the system achieves good image quality with minimal computational resources suitable for printer hardware.
Solution Approach 2:
The patent changes the approach by using pre-computed parameters (rotation matrix from highlight/shadow point analysis, low-frequency image from filtering) that can be stored and applied efficiently. This parameter-based approach reduces real-time computation requirements, enabling complex-looking corrections to be performed with simple hardware operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing apparatus includes means (102) configured to calculate a plurality of characteristic amounts from an input image, first generation means (104) configured to generate a first multidimensional lookup table based on the characteristic amounts, means (105) configured to calculate attribute information of a specific object from the input image, second generation means (106) configured to generate a second multidimensional lookup table based on the attribute information, control parameter generation means (103) configured to generate a control parameter from the input image, and interpolation means (107) configured to apply the first and second multidimensional lookup tables to the input image, the image processing apparatus includes control means (108) configured to control interpolation intensity of the second multidimensional lookup table using the control parameter.