Semantic-Aware White Balance for Object-Specific Color Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional auto white balance methods struggle to handle both raw device RGB and standard RGB color spaces simultaneously and often fail to accurately adjust colors based on scene illumination, particularly for objects like skin tones, green vegetation, and sky, leading to unrealistic color casts in images.
Innovation Solution
An image processing system that identifies specific objects (like skin, vegetation, or sky) in an image and applies reference color distributions to minimize the difference between pixel values and a reference color distribution, using loss functions to determine optimal image processing settings for temperature and tint adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional auto white balance methods are used, then general color adjustment is performed, but accurate color adjustment for specific objects like skin tones cannot be achieved
Solution Approach 1:
The patent segments the image processing task by identifying and separating specific objects (skin, vegetation, sky) from the general scene. It applies object detection to locate target objects and then performs white balance adjustment specifically on pixels corresponding to these objects using reference color distributions, rather than adjusting the entire image uniformly.
Solution Approach 2:
The patent implements local quality by applying different white balance adjustments to different regions of the image based on object type. It uses reference color distributions specific to each object type (e.g., skin color distribution) to adjust only the relevant regions, preserving the natural colors of specific objects while leaving other areas unaffected.
2Reliability
If standard RGB color space processing is used, then compatibility is maintained, but accurate white balance for raw device RGB images cannot be achieved
Solution Approach 1:
The patent creates a universal white balance system that handles both raw device RGB and standard RGB color spaces through a common framework. It uses reference color distributions that can be applied across different color spaces, making the system multi-functional and compatible with various image types while maintaining accuracy for each.
3Adaptability or versatility
If generic reference color distributions are used, then broad applicability is achieved, but personalized color accuracy for individuals cannot be obtained
Solution Approach 1:
The patent implements a dynamic reference color distribution system that can adapt between generic and personalized modes. It allows the reference color distribution to be selected or customized based on the specific application needs, enabling the system to switch from broad applicability (generic distributions) to personalized accuracy (custom distributions) as required.
Data Source
AI summary
An image processing system auto white balances an image using an object in the image and a reference color distribution. Given an input image, a target object in the input image is identified. A reference color distribution for the object type of the target object from the input image is accessed. One or more image processing settings are determined that, when applied to the input image, minimize a difference in values between pixels of the target object and the reference color distribution. A white balanced image is generated by applying the one or more image processing settings to the input image, and the white balanced image is provided for presentation.