Face Image Virtual Effect Placement with Occlusion Masking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies fail to accurately apply virtual special effects to face parts when they are occluded by other objects or body parts, leading to suboptimal results due to the inability to distinguish between visible and invisible areas in the face image.
Innovation Solution
The method involves face key point detection, fusion of virtual special effects with face parts, and the use of an occlusion mask to differentiate between visible and occluded areas, ensuring that virtual effects are only applied to non-occluded face regions, utilizing image semantic segmentation to generate a second fusion image that correctly places the effects on visible areas and hides them on occluded areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If virtual special effects are applied to all face parts based on face key point detection, then the coverage and completeness of special effect application is improved, but the accuracy of special effect placement deteriorates because effects are also applied to occluded areas
Solution Approach 1:
The patent segments the face image into visible and occluded regions using occlusion detection technology. This segmentation allows the system to selectively apply virtual special effects only to visible face parts, resolving the contradiction by dividing the application area into distinct zones based on occlusion status.
Solution Approach 2:
The patent implements local quality by applying different processing rules to different regions of the face image. Visible regions receive virtual special effects while occluded regions do not, creating spatially varying quality characteristics that improve overall accuracy without sacrificing necessary coverage.
2Measurement precision
If occlusion detection is added to distinguish visible and occluded areas, then the accuracy of special effect application is improved, but the processing complexity increases
Solution Approach 1:
The patent performs occlusion detection as a preliminary step before applying virtual special effects. By pre-identifying visible and occluded regions, the system avoids the need for complex real-time decision-making during effect application, thereby managing processing complexity while maintaining high accuracy.
Solution Approach 2:
The patent merges multiple functions into an integrated processing pipeline that combines face key point detection, occlusion detection, and virtual special effect application. This consolidation manages complexity by coordinating these functions within a unified system framework rather than treating them as separate complex subsystems.
3Productivity
If virtual special effects are applied without occlusion awareness, then the processing speed is maintained, but the quality of beautification results deteriorates due to incorrect placement on occluded areas
Solution Approach 1:
The patent performs occlusion detection in advance before applying virtual special effects, enabling the system to make accurate placement decisions without slowing down the overall processing pipeline. This preliminary identification of visible regions ensures high-quality results while maintaining efficient processing throughput.
Data Source
AI summary
Provided is a method for processing images. The method can include: acquiring a target face image, and performing face key point detection on the target face image to acquire a face key point detection result; determining a face visible area which is not subject to an occluder and a face invisible area which is subject to the occluder in the target face image; and acquiring a target fusion image by fusing a virtual special effect and a face part matched in the target face image based on the face key point detection result, the face visible area and the face invisible area.