Dual Camera Depth Map Generation Using Defocus Spreading
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic devices using phase difference calculation technologies struggle to generate depth information from stereo images with repeating patterns or high spatial frequencies, as they fail to accurately calculate phase differences due to multiple candidate regions.
Innovation Solution
The method involves using a Depth from Defocus (DFD) operation in electronic devices equipped with dual cameras to select depth information based on the degree of spreading of points in images, allowing for depth map generation even when phase difference calculations are unreliable.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If phase difference calculation technology is used to generate depth information, then depth information can be obtained for most regions, but it fails for regions with repeating patterns or high spatial frequencies due to multiple candidate regions
Solution Approach 1:
The patent introduces an auxiliary image captured with a different focal length as an intermediary to resolve the ambiguity in phase difference calculation. When multiple candidate regions are found in the stereo image pair, the auxiliary image provides additional depth cues through its different point spread function characteristics, enabling the system to select the correct matching region and generate reliable depth information for challenging patterns.
Solution Approach 2:
The patent changes the optical parameters by using an auxiliary camera with a different focal length than the main stereo cameras. This parameter change results in a different point spread function that provides alternative depth information, allowing the system to resolve ambiguities that arise with standard phase difference calculation on repeating or high-frequency patterns.
2Adaptability or versatility
If multiple candidate regions are identified in the second image corresponding to a first region, then more potential matches are available, but it becomes impossible to determine the correct match for depth calculation
Solution Approach 1:
The auxiliary image acts as a mediator that provides additional constraints to resolve the multiple candidate region problem. By comparing the auxiliary image data with the stereo image pair, the system can eliminate incorrect candidates and identify the true matching region, thereby restoring depth calculation accuracy despite the initial ambiguity.
Solution Approach 2:
The system uses feedback from the auxiliary image to evaluate and select among multiple candidate regions. The auxiliary depth information serves as a feedback mechanism that validates or rejects candidate matches, enabling the system to converge on the correct matching region and generate accurate depth information.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of depth information calculation by utilizing DFD operations to handle images with high frequency regions or repeated patterns, improving the reliability of depth map generation in challenging image conditions.
Implementation Method 1
obtain a first image of one or more external objects using a first camera, obtain a second image of the one or more external objects using a second camera
Implementation Method 2
select depth information about the specified object among the pieces of depth information based on a degree of spreading of a point corresponding to the specified object included in at least one of the first image and the second image
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
An electronic device is provided. The electronic device includes a first camera, a second camera spaced apart from the first camera, and a processor. The processor is configured to obtain a first image of external objects using the first camera, obtain a second image of the external objects using the second camera, identify a specified object in which pieces of depth information are generated from among the external objects included in the first image and the second image based on phase difference comparison between the first image and the second image, then select depth information about the specified object among the pieces of depth information based on a degree of spreading of a point corresponding to the specified object included in at least one of the first image and the second image, and generate depth information about the external objects including the specified object using the selected depth information.