Depth Map Correction via Multi-Camera Occlusion Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Stereo camera systems face challenges in generating accurate depth maps due to occlusion issues, where an object between the camera and another object results in invalid areas in the depth map, especially when the baseline between the camera pairs increases.
Innovation Solution
A depth map generation device that utilizes multiple image capture pairs and a depth map generator to correct occlusion regions by using valid areas from secondary depth maps, potentially adjusting disparities and performing geometric calibration to align baselines, thereby replacing invalid areas with valid depth information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If the baseline between the left eye image capturer and the right eye image capturer is increased, then the depth map generation device can capture a wider field of view, but the occlusion problem becomes more severe and the depth map includes more invalid areas
Solution Approach 1:
The patent divides the depth map generation process into multiple segments by using multiple image capture pairs. Each pair generates a partial depth map covering a specific region, and the system segments the overall scene into multiple sub-regions that can be independently processed and then combined to form a complete depth map with reduced occlusion artifacts.
Solution Approach 2:
The patent introduces an intermediary processing mechanism that uses valid depth information from non-occluded regions to correct occlusion regions. This intermediary process acts as a mediator that transfers valid depth data between different image capture pairs to fill in the invalid areas caused by occlusion.
2Measurement precision
If the baseline between the left eye image capturer and the right eye image capturer is increased, then the depth resolution is improved, but the occlusion region in the depth map increases
Solution Approach 1:
The patent merges the depth maps generated by multiple image capture pairs to create a comprehensive depth map. By combining the results from multiple pairs with different baselines, the system achieves high depth resolution while compensating for occlusion regions through data fusion and mutual correction between the paired depth maps.
Solution Approach 2:
The patent transitions from a single baseline dimension to multiple baseline dimensions by employing multiple image capture pairs. This dimensional expansion allows the system to resolve depth information from multiple perspectives, thereby improving depth resolution while mitigating the occlusion problem that plagues single-baseline systems.
3Measurement precision
If multiple image capture pairs are used to correct occlusion, then the depth map accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent designs the multiple image capture pairs to serve multiple functions simultaneously: each pair not only captures images for depth calculation but also provides valid depth information for correcting occlusion regions in other pairs' depth maps. This multi-functionality reduces the need for additional correction mechanisms and simplifies the overall system architecture.
Solution Approach 2:
The system implements a self-service mechanism where the depth maps generated by each image capture pair automatically contribute to correcting occlusion regions in other pairs' depth maps. The valid depth information from each pair serves itself and others, creating a self-correcting system that improves depth map accuracy without requiring external intervention or complex post-processing.
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
Effectively addresses the issue of occlusion in depth maps by using valid depth information from non-occluded regions, improving the accuracy and completeness of depth maps generated by the system.
Implementation Method 1
when a depth map generator coupled to the stereo camera utilizes a triangulation algorithm provided by the prior art to generate a depth map
Data Source
AI summary
A depth map generation device capable of correcting occlusion includes at least two image capture pairs and a depth map generator. The at least two image capture pairs is used for capturing a plurality of images. The depth map generator is coupled to the two image capture pairs for generating a first depth map and a second depth map according to the plurality of images, wherein when the first depth map includes a first occlusion region and a first non-occlusion region, the depth map generator corrects the first occlusion region according to the second depth map.