Image Feature Migration With Adjustable Fusion Weight Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image feature migration techniques do not allow users to control the migration effect in the migration result image, such as the intensity of face makeup transfer from a reference image to a target image.
Innovation Solution
An image processing method that involves obtaining target and reference features, using a migration matrix to control the fusion of these features at a designated weight, and generating a migration result image based on the fused features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a fixed migration result image is obtained from a pre-trained image migration model, then the migration process is simple and automated, but the user cannot control the migration effect or adjust the intensity of feature transfer
Solution Approach 1:
The patent applies the Dynamics principle by introducing a dynamic weight parameter that allows users to adjust the migration effect intensity. The system transitions from a fixed, static migration process to a dynamic one where the user can control the degree of feature transfer by modifying the weight parameter, enabling flexible adjustment between complete migration and partial migration effects.
Solution Approach 2:
The patent implements Parameter changes by incorporating a weight parameter that modifies the fusion between target features and reference features. By changing this parameter, the system can control the intensity of migration, allowing users to adjust the contribution of reference features to the final migration result image, thus achieving controllable migration effects.
2Manufacturing precision
If the reference feature is completely migrated to the target image, then the migration effect is strong and noticeable, but the user cannot achieve partial migration or subtle integration effects
Solution Approach 1:
The patent applies Parameter changes by introducing a weight parameter that controls the fusion ratio between target and reference features. This parameter enables precise control over migration intensity, allowing the system to produce subtle integration effects when the weight is low and strong migration effects when the weight is high, thus achieving both precision and versatility.
Solution Approach 2:
The system becomes dynamic through the adjustable weight parameter, transitioning from a fixed migration intensity to a flexible, controllable one. This allows the user to adapt the migration effect to different needs, achieving partial migration for subtle changes or complete migration for strong effects, thereby enhancing both precision and adaptability.
3Productivity
If a pre-trained image migration model is used directly, then the implementation is straightforward and fast, but the migration effect lacks customization and user control
Solution Approach 1:
The patent maintains the efficiency of pre-trained models while adding customization through a weight parameter. The system remains fast by using the pre-trained model for feature extraction and migration, but gains adaptability by allowing users to adjust the weight parameter to control migration intensity, thus achieving both high productivity and customization.
Solution Approach 2:
The system transitions from a static, fixed-effect migration process to a dynamic one where the weight parameter enables real-time adjustment of migration intensity. This maintains the speed advantage of automated processing while adding the flexibility for users to customize migration effects according to their specific needs.
Data Source
AI summary
An image processing method including: obtaining a target feature and a reference feature, the target feature is a feature obtained by performing feature extraction on a target object in a target image, and the reference feature is a feature obtained by performing feature extraction on a reference object in a reference image; obtaining a migration matrix, the migration matrix is used for controlling the reference feature to be fused with the target feature at a designated weight; fusing the target feature and the reference feature according to the migration matrix to obtain a fused feature; and generating a migration result image according to the fused feature and the target image. An image processing apparatus, an electronic device, a storage medium and a program product are also disclosed.