3D Image View Synthesis Using Depth Map Biasing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating images from 3D image data face challenges in providing complete representations of scenes, especially in virtual reality applications, where incomplete data leads to reduced image quality and an immersive experience, particularly when users move beyond captured viewpoints.
Innovation Solution
An apparatus and method that utilize a receiver for 3D image data, a target view vector, and a reference source to generate a rendering view vector, biasing it away from areas with no data, allowing for improved image generation by extrapolating from available data and applying spatially varying blurring to maintain image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If view images are generated for new viewing directions using view point shifting processing, then the adaptability of the display system is improved, but the image quality deteriorates due to dependence on depth information accuracy
Solution Approach 1:
The patent introduces depth maps as an intermediary element that mediates between the limited input images and the desired output views. By using depth maps to guide the view synthesis process, the system can generate images for directions not directly captured while maintaining quality, as the depth information provides structural guidance for accurate pixel mapping and rendering.
Solution Approach 2:
The patent changes the parameter space by incorporating multiple depth maps corresponding to different viewpoints, rather than relying on a single depth map. This allows the system to select or combine depth information appropriate for the target viewing direction, improving both the adaptability to new views and the accuracy of the generated images.
2Adaptability or versatility
If more than two view images are used for autostereoscopic displays, then the viewing experience is improved, but the complexity of data processing increases
Solution Approach 1:
The patent segments the complex task of generating multiple autostereoscopic views into manageable components: acquiring a base image, generating multiple depth maps from different virtual camera positions, and synthesizing multiple view images through view point shifting. This segmentation allows each component to be processed independently and efficiently, reducing overall computational complexity.
Solution Approach 2:
The patent performs preliminary depth map generation for multiple viewpoints before the actual view synthesis. By pre-computing depth maps for various virtual camera positions, the system prepares necessary data structures in advance, which significantly speeds up the real-time view generation process and reduces processing complexity during operation.
3Ease of manufacture
If 3D image data provides incomplete representation of the scene, then the data acquisition process is simplified, but the completeness of scene representation deteriorates
Solution Approach 1:
The patent creates virtual copies of the scene by generating depth maps for virtual camera positions that did not physically exist. These synthetic depth maps serve as copies of the real scene structure, allowing the system to represent and render views from positions and directions not captured by the physical camera, thereby completing the scene representation without additional physical data acquisition.
Solution Approach 2:
The patent adds the depth dimension to the 2D input image by generating multiple depth maps representing different virtual camera positions. This dimensional enrichment transforms incomplete 2D image data into a more complete 3D scene representation, enabling comprehensive view synthesis while maintaining simplicity in the original data acquisition process.
Data Source
AI summary
An apparatus for generating an image comprises a receiver (101) which receives 3D image data providing an incomplete representation of a scene. A receiver (107) receives a target view vector indicative of a target viewpoint in the scene for the image and a reference source (109) provides a reference view vector indicative of a reference viewpoint for the scene. A modifier (111) generates a rendering view vector indicative of a rendering viewpoint as a function of the target viewpoint and the reference viewpoint for the scene. An image generator (105) generates the image in response to the rendering view vector and the 3D image data.