Adaptive Structured Light Patterns for Depth Map Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Structured light imaging devices face challenges in capturing optimal depth maps due to the need for trade-offs between spatial frequency patterns, which can result in suboptimal detail capture and depth range, especially when objects are at varying distances or have different colors, leading to issues with signal-to-noise ratio and motion in the scene.
Innovation Solution
The implementation of adaptive structured light patterns that dynamically adjust based on scene content, including spatial frequency, motion, and object colors, by generating disparity maps and adapting patterns to optimize depth map computation, such as using high, low, or medium frequency patterns and combining static and time-multiplexed patterns as needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single fixed spatial frequency pattern is used in fixed pattern structured light imaging devices, then the device can handle moving objects, but the depth range and detail capture are compromised due to the trade-off between spatial frequency and depth range
Solution Approach 1:
The patent applies dynamics by transitioning from a static single-frequency pattern to a dynamic multi-frequency pattern system. The device captures initial depth information using a first spatial frequency, then adapts by projecting a second spatial frequency pattern that is optimized based on the initial depth map. This dynamic adaptation allows the system to handle moving objects while achieving accurate depth measurement across varying distances.
Solution Approach 2:
The patent changes the spatial frequency parameter of the projected pattern based on initial depth measurements. The system determines a second spatial frequency that is appropriate for the specific scene geometry revealed by the first pattern, thereby optimizing depth measurement accuracy for the actual scene conditions rather than using a fixed frequency that must compromise between near and far objects.
2Measurement precision
If higher spatial frequency patterns are used, then finer details in the scene can be captured, but the depth range is reduced
Solution Approach 1:
The patent segments the depth measurement process into two stages: first using a lower spatial frequency pattern to establish coarse depth structure and range, then using a higher spatial frequency pattern optimized for the revealed scene geometry to capture fine details. This segmentation allows the system to benefit from both low-frequency depth range and high-frequency detail capture without the limitations of a single fixed frequency.
Solution Approach 2:
The patent performs a preliminary depth measurement using a first spatial frequency pattern before capturing the final high-precision depth map. This preliminary action reveals the scene geometry, which then informs the selection of an optimized second spatial frequency that can capture fine details while maintaining appropriate depth range for the specific scene.
3Adaptability or versatility
If lower spatial frequency patterns are used, then the depth range is increased, but the capture of finer details is reduced
Solution Approach 1:
The patent dynamically transitions from a low spatial frequency pattern used for establishing depth range to a high spatial frequency pattern optimized for the revealed scene. This dynamic frequency adaptation allows the system to first ensure adequate depth range coverage, then enhance detail capture in regions where the optimized frequency provides benefit.
Data Source
AI summary
A method of depth map optimization using an adaptive structured light pattern is provided that includes capturing, by a camera in a structured light imaging device, a first image of a scene into which a pre-determined structured light pattern is projected by a projector in the structured light imaging device, generating a first disparity map based on the captured first image and the structured light pattern, adapting the structured light pattern based on the first disparity map to generate an adaptive pattern, wherein at least one region of the structured light pattern is replaced by a different pattern, capturing, by the camera, a second image of the scene into which the adaptive pattern is projected by the projector, generating a second disparity map based on the captured second image and the adaptive pattern, and generating a depth image using the second disparity map.