Image Difference Masking for Precise Local Diffusion Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image adjustment methods using diffusion models fail to accurately capture image details, leading to suboptimal adjustments that do not meet actual requirements, hindering the development of effective image processing services.
Innovation Solution
An image processing method that involves determining similarity between images, identifying difference information, generating a target mask image, and performing text expansion to adjust images based on difference description text, ensuring the adjusted image meets content requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If description information is adaptively adjusted while network parameters of diffusion model remain unchanged, then image adjustment flexibility is improved, but image adjustment accuracy deteriorates due to ignoring image details
Solution Approach 1:
The patent segments the image adjustment process into multiple stages: first finding a similar image from database, then identifying difference information between reference and similar images, generating target mask images for different difference types, and finally performing localized adjustments. This segmentation allows both flexibility in handling different adjustment scenarios and precision in maintaining image details through targeted modifications.
Solution Approach 2:
The patent performs preliminary actions by first searching for and selecting a similar image from the database that matches the reference image, then pre-identifying difference information and generating target mask images before executing the actual adjustment. This preliminary preparation ensures that the subsequent adjustment process can focus on specific differences with high precision while maintaining overall flexibility.
2Adaptability or versatility
If existing diffusion model is used for image adjustment, then general image processing capability is maintained, but detailed image content accuracy deteriorates
Solution Approach 1:
The patent applies local quality by generating different types of target mask images (first type for adding difference objects, second type for retaining similar image objects, third type for removing difference objects) and applying adjustments only to specific regions where differences exist. This localized approach maintains high accuracy for detailed image content while preserving the general capabilities of the diffusion model for overall image processing.
Data Source
AI summary
An image processing method includes obtaining a similarity between each image in a database and a reference image, determining an image having a largest similarity in the database as a similar image, determining difference information between the reference image and the similar image, determining a target mask image, in the reference image, for the difference information that highlights a difference object that is in the reference image relative to the similar image and that is determined according to the difference information, performing text expansion expression based on the difference object and the reference image to obtain a difference description text that describes a content difference between the reference image and the similar image, and locally adjusting the similar image according to the target mask image, the difference description text, and the reference image to obtain a target image conforming to a content requirement of the reference image.