Depth Map Generation Using Multi-Pose Image Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mobile electronic devices face challenges in generating high-resolution depth maps due to computational expense and hardware limitations, often resulting in loss of detail when upconverting low-resolution depth maps.
Innovation Solution
The method involves obtaining multiple high-resolution input image frames, generating an initial high-resolution depth map, and iteratively refining it using an optimization process that includes pose estimation, depth map coordinate conversion, and loss minimization to produce a high-resolution depth map with improved accuracy and detail.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If low-resolution depth maps are upconverted to high-resolution, then depth map resolution is increased, but detail is lost
Solution Approach 1:
The patent performs preliminary action by capturing multiple high-resolution input images at different poses before depth map generation. These pre-captured images serve as the foundation for generating high-resolution depth maps without upconversion, preserving all original detail information from the source images.
Solution Approach 2:
The patent transitions from a single 2D depth map representation to a multi-dimensional approach by utilizing multiple high-resolution input images captured at different poses. This dimensional expansion provides redundant information that enables accurate depth estimation without requiring resolution upconversion.
2Measurement precision
If high-resolution depth maps are generated directly, then depth map quality is improved, but computational expense increases
Solution Approach 1:
The system performs preliminary action by capturing multiple high-resolution images at different poses before depth map generation. This pre-capture strategy provides rich source material that enables accurate depth estimation without requiring computationally intensive upconversion processes.
Solution Approach 2:
The patent creates multiple copies of the scene from different perspectives (multiple input images at different poses). These copies serve as the basis for generating high-resolution depth maps, eliminating the need for computationally expensive upconversion while preserving detail.
3Measurement precision
If multiple input images at different poses are used, then depth map accuracy is improved, but device complexity increases
Solution Approach 1:
The patent performs preliminary action by capturing multiple high-resolution images at different poses before depth map generation. This pre-capture strategy provides rich source material that enables accurate depth estimation without requiring upconversion, preserving all original detail information from the source images.
Data Source
AI summary
A method includes obtaining, using at least one processor, first and second input image frames, where the first and second input image frames are associated with first and second image planes, respectively. The method also includes obtaining, using the at least one processor, a depth map associated with the first input image frame. The method further includes producing another version of the depth map by performing one or more times: (a) projecting, using the at least one processor, the first input image frame to the second image plane in order to produce a projected image frame using (i) the depth map and (ii) information identifying a conversion from the first image plane to the second image plane and (b) adjusting, using the at least one processor, at least one of the depth map and the information identifying the conversion from the first image plane to the second image plane.