Color Point Cloud Generation via Trajectory Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for generating color point clouds using LiDAR and cameras face challenges due to the discrepancy in Field of View (FOV) between LiDAR and monocular cameras, leading to inaccuracies and inefficiencies in aggregating point clouds and images, especially with panoramic cameras being costly and monocular cameras having limited FOV.
Innovation Solution
A method and system that segment point clouds into segments based on a vehicle's trajectory, associate each segment with corresponding images using calibration parameters adjusted for different distances, and aggregate these segments with images to generate color point clouds, optimizing the processing and accuracy by considering the varying distances and views.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If panoramic cameras are used for color point cloud generation, then the field of view coverage is improved (360-degree FOV), but the cost increases significantly
Solution Approach 1:
The point cloud is segmented into multiple segments based on the vehicle trajectory, with each segment associated with specific start and end points. This segmentation allows monocular cameras to capture relevant portions of the scene at different positions, achieving comprehensive coverage without requiring a single 360-degree panoramic camera.
Solution Approach 2:
The patent introduces the trajectory dimension by segmenting point clouds based on vehicle movement path. This transforms the problem from a static single-view coverage issue to a dynamic multi-position coverage problem, allowing monocular cameras to achieve comprehensive coverage through multiple positions along the trajectory.
2Ease of manufacture
If monocular cameras are used for color point cloud generation, then the cost is reduced, but the field of view is limited and aggregation accuracy decreases
Solution Approach 1:
The patent segments the point cloud based on vehicle trajectory into multiple segments, each associated with start and end points. This allows monocular cameras to capture images at different positions along the trajectory, effectively expanding the covered field of view despite the camera's limited single-view FOV.
Solution Approach 2:
The system performs preliminary segmentation of the point cloud and identification of associated images before aggregation. By pre-associating point cloud segments with their corresponding images based on trajectory information, the system ensures accurate matching and aggregation even with limited camera FOV.
3Device complexity
If point clouds are aggregated with images without segmentation, then the processing is simpler, but the aggregation accuracy and efficiency decrease due to FOV mismatch
Solution Approach 1:
The patent segments the point cloud into multiple segments based on vehicle trajectory, with each segment associated with specific start and end points. This segmentation enables efficient matching with corresponding images by reducing the search space and improving aggregation accuracy, thereby enhancing overall processing efficiency despite increased initial complexity.
Solution Approach 2:
The system performs preliminary association of point cloud segments with images based on trajectory information before the aggregation process. This pre-processing step establishes correct correspondences in advance, making the subsequent aggregation more efficient and accurate compared to unsegmented approaches.
4Measurement precision
If point clouds are segmented and associated with images based on trajectory, then the aggregation accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the point cloud based on vehicle trajectory into manageable segments, each associated with specific images. This segmentation improves aggregation accuracy by ensuring correct image-point cloud matching, while the structured approach based on trajectory information helps manage computational complexity through organized processing.
Solution Approach 2:
The system performs preliminary association of point cloud segments with images based on trajectory information before aggregation. By establishing correct correspondences in advance using trajectory data, the system achieves high aggregation accuracy while avoiding the need for complex real-time matching during aggregation.
Data Source
AI summary
Embodiments of the disclosure provide methods and systems for generating a color point cloud. The method may include receiving a point cloud and a plurality of images with respect to a scene captured by a plurality of sensors associated with a vehicle as the vehicle moves along a trajectory. The method may include segmenting the point cloud into a plurality of segments each associated with a start point and an end point on the trajectory of the vehicle. The method may also include associating each segment of the point cloud with one or more of the images based on the start point and the end point. The method may further include generating color point cloud by aggregating each segment of the point cloud and the one or more of the images based on calibration parameter in different distances between the segment of the point cloud and the vehicle.