3D Object Detection Control Using Adaptive Point Cloud Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection techniques using cameras for three-dimensional object detection require longer times to complete the entire process, particularly when combining methods like phase shifting and spatial coding, which demand multiple imaging sessions.
Innovation Solution
A control apparatus that executes object detection using a first point cloud generation process via the stereo block matching method and a second point cloud generation process via the phase shifting method, where the second process is initiated after the first and stopped if the first point cloud meets predetermined success conditions, allowing for earlier completion of the overall process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the phase shifting method or spatial coding method is used for depth measurement, then high-accuracy point clouds are generated, but the imaging process requires multiple imaging sessions and takes longer time
Solution Approach 1:
The patent implements dynamic switching between two depth measurement methods based on real-time evaluation. The system first attempts the rapid stereo block matching method, evaluates the quality of generated point clouds, and only switches to the phase shifting or spatial coding method when necessary. This dynamic adaptation allows the system to optimize between speed and accuracy based on actual measurement conditions.
Solution Approach 2:
The patent segments the depth measurement process into two distinct stages: a first stage using stereo block matching for rapid initial measurement, and a second stage using phase shifting or spatial coding for high-precision measurement only when needed. This segmentation allows the system to complete most measurements quickly while reserving computationally intensive methods for cases requiring higher accuracy.
2Productivity
If the stereo block matching method is used for depth measurement, then the imaging process is completed in shorter time, but the position accuracy of the point cloud is lower
Solution Approach 1:
The patent incorporates a feedback mechanism that evaluates the quality and accuracy of point clouds generated by the stereo block matching method. Based on this evaluation, the system determines whether to switch to the phase shifting or spatial coding method for improved accuracy. This feedback loop ensures that the system maintains measurement accuracy while maximizing processing speed.
Solution Approach 2:
The patent applies partial action by using the simpler stereo block matching method for the majority of measurements where high accuracy is not critical, and reserves the more accurate but time-consuming phase shifting or spatial coding method for cases where higher precision is required. This approach achieves sufficient accuracy for most applications while significantly reducing overall processing time.
3Measurement precision
If two depth measuring methods are combined to increase measurable height range and achieve high-accuracy measurement, then measurement accuracy and range are improved, but the entire process takes longer time to complete
Solution Approach 1:
The patent performs preliminary measurement using the rapid stereo block matching method before attempting high-precision measurement with phase shifting or spatial coding. This preliminary action allows the system to quickly assess whether high-precision measurement is necessary, thereby avoiding unnecessary time consumption while ensuring measurement accuracy when required.
Solution Approach 2:
The patent changes the measurement parameters dynamically by switching between different depth measurement methods based on the measured object's characteristics and required accuracy. The system adjusts which method to use by evaluating parameters such as point cloud quality, measurement range requirements, and time constraints, thereby optimizing the balance between accuracy and processing time.
Data Source
AI summary
A control apparatus includes a processor that executes a first point cloud generation process including a first imaging process of acquiring a first image according to a first depth measuring method and a first analysis process of generating a first point cloud and a second point cloud generation process including a second imaging process of acquiring a second image according to a second depth measuring method and a second analysis process of generating a second point cloud, and detects the object using the first point cloud or the second point cloud. The first point cloud generation process completes in a shorter time than the second point cloud generation process, and the processor starts the second point cloud generation process after the first imaging process and discontinues the second point cloud generation process if the first point cloud satisfies a predetermined condition of success.