Image Processing With Depth-Map Masks for AR Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for augmented reality fail to accurately determine the occlusion relation between virtual and target objects due to the use of fixed-size standard virtual models, leading to inaccuracies in rendering and reduced realism.
Innovation Solution
An image processing method that involves dividing a target object into a set portion, acquiring depth maps for virtual and standard models, adjusting mask maps based on these depth maps, and superimposing rendered portions to achieve accurate occlusion rendering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed-size standard virtual model is used to determine occlusion relation, then the processing complexity is reduced, but the accuracy of occlusion relation determination deteriorates
Solution Approach 1:
The patent applies the dynamics principle by transforming the fixed-size standard virtual model into a dynamic, adaptive model that automatically adjusts its size and shape to match the target object. The system performs size adaptation and shape adaptation processes that allow the virtual model to dynamically conform to various target objects, thereby maintaining high occlusion relation determination accuracy while managing processing complexity through automated adaptation mechanisms.
Solution Approach 2:
The patent implements parameter changes by modifying the size and shape parameters of the virtual model based on the target object's characteristics. The system changes the scale parameter to match the target object's size and adjusts the geometric parameters to conform to the target object's shape, enabling accurate occlusion relation determination for diverse objects while keeping the overall processing framework manageable.
2Device complexity
If a fixed-size standard virtual model is used, then the model simplicity is maintained, but the adaptability to various target objects deteriorates
Solution Approach 1:
The system transforms the static virtual model into a dynamic one that automatically adapts to different target objects. The adaptation process includes size adaptation that scales the model to match the target object's dimensions and shape adaptation that adjusts the model's geometry to conform to the target object's form, thereby achieving high versatility while maintaining relatively simple processing through automated procedures.
Solution Approach 2:
The patent applies universality by creating a virtual model that can serve multiple functions across different target objects. The adapted virtual model can determine occlusion relations for various types of objects (e.g., cups, bottles, containers) using the same base model, enhancing adaptability while keeping the system structure relatively simple through reusable components.
3Loss of time
If the virtual object does not conform to the target object, then the processing time is reduced, but the realism of the rendered image deteriorates
Solution Approach 1:
The patent applies preliminary action by performing size adaptation and shape adaptation of the virtual model before the occlusion relation determination and rendering processes. This pre-adaptation ensures that the virtual model is already conforming to the target object's characteristics when rendering occurs, thereby achieving high realism without significantly increasing processing time during the critical rendering phase.
Solution Approach 2:
The system uses dynamic adaptation to quickly adjust the virtual model's size and shape to match the target object before rendering. This dynamic conforming process occurs in advance, allowing the rendering stage to proceed efficiently with pre-adapted models, thus maintaining both realism and acceptable processing time.
Data Source
AI summary
An image processing method, a device, and a storage medium are provided. The image processing method includes: dividing a set portion of a target object to obtain an initial mask map; acquiring a first depth map of a virtual object and a second depth map of a standard virtual model relating to the target object; adjusting the initial mask map based on the first depth map and the second depth map to obtain a target mask map; rendering the set portion based on the target mask map to obtain a set portion map; and rendering the virtual object to obtain a virtual object map; and superimposing the set portion map and the virtual object map to obtain a target image.