Dual-Field Imaging for Joint Object Detection and 3D Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D reconstruction and visual object detection methods face contradicting requirements for successful object mapping and detection due to differing field of view needs, leading to inefficient and inaccurate processes.
Innovation Solution
Utilizing two image capturing units with different field of views, one with a narrow field of view for detailed detection and the other with a wide field of view for registration, synchronized to enable joint visual object detection and object mapping to a 3D model, with spatial relation-based object projection and guidance for optimal capture positioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a narrow field of view is used for visual object detection, then detection accuracy is improved, but object mapping and 3D reconstruction quality deteriorates
Solution Approach 1:
The patent divides the imaging task into two separate functions performed by two different image capturing units: one unit with a narrow field of view dedicated to visual object detection, and another unit with a wide field of view dedicated to object mapping and 3D reconstruction. This segmentation allows each unit to be optimized for its specific function, resolving the contradiction between detection accuracy and reconstruction quality
Solution Approach 2:
The system achieves multi-functionality by having two image capturing units that can independently perform different functions (detection and mapping) simultaneously. The processing device integrates results from both units, enabling the system to deliver both high-quality object detection and accurate 3D reconstruction without compromising either function
2Manufacturing precision
If a wide field of view is used for object mapping, then registration quality is improved, but object detection accuracy deteriorates
Solution Approach 1:
The patent segments the imaging system into two specialized units: one with a wide field of view optimized for capturing scene context and performing accurate object mapping and registration, and another with a narrow field of view optimized for detailed object detection. This eliminates the need to compromise detection accuracy when using a wide field of view for mapping
Solution Approach 2:
The processing device acts as an intermediary that receives and integrates data from both image capturing units. It combines the wide-field mapping information with narrow-field detection results, allowing the system to achieve both high registration quality and high detection accuracy through coordinated processing of both data streams
3Device complexity
If a single image capturing unit is used, then device complexity is reduced, but both object detection and mapping cannot be optimized simultaneously
Solution Approach 1:
The patent applies segmentation by dividing the imaging system into two specialized image capturing units with different field of view characteristics. This segmentation enables each unit to be optimized for its specific function (detection or mapping), achieving high quality in both tasks simultaneously, which would be impossible with a single unit
Solution Approach 2:
The system merges the capabilities of two specialized image capturing units into a coordinated dual-unit system. The processing device combines the narrow-field detection data and wide-field mapping data into a unified result, achieving optimized performance for both functions while managing the increased complexity through integrated processing
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There is provided mechanisms for joint visual object detection and object mapping to a 3D model. A method is performed by an image processing device (100). The method comprises obtaining (S106) a first sequence of digital images of a scene as captured by a first image capturing unit, and obtaining a second sequence of digital images of the scene as captured by a second image capturing unit. The second sequence of digital images is time-wise synchronized with the first sequence of digital images by being captured time-wise in parallel with the first sequence of digital images. The first image capturing unit has a narrower field of view than the field of view of the second image capturing unit. The first image capturing unit and the second image capturing unit have a known spatial relation. The method comprises performing (S108) joint visual object detection and object mapping to the 3D model.