Method and device for detecting height-limiting rod, and automatic driving system
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting height-limiting rods in automatic driving systems face challenges due to reliance on visual image identification requiring comprehensive databases and classifiers, and insufficient reliability when using sensors like laser/millimeter-wave radar, especially for thin rods.
Innovation Solution
A method and device using a binocular stereo camera that filters invalid points in a disparity map, projects valid points onto a 3D coordinate system, filters invalid blobs, and determines the location of height-limiting rods through X-axis, Y-axis, and Z-axis coordinate values, enabling accurate detection and distance calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If visual image identification is used to detect height-limiting rods, then detection can be performed, but it requires comprehensive sample databases and excellent classifiers making it impossible to reliably detect the rods
Solution Approach 1:
The patent replaces visual image identification (optical system) with a binocular stereo vision system that uses disparity maps and geometric projection. Instead of relying on image classification algorithms and databases, the system uses stereoscopic depth information and mathematical projection to detect height-limiting rods, thereby improving reliability while reducing dependence on complex visual databases.
Solution Approach 2:
The patent introduces disparity maps as an intermediary representation between the binocular camera input and the final detection result. The disparity map encodes depth information that serves as a bridge, allowing the system to detect height-limiting rods through geometric relationships rather than direct visual pattern recognition, thus improving reliability without requiring comprehensive visual databases.
2Reliability
If laser or millimeter-wave radar is used to detect thin height-limiting rods, then detection can be performed, but the reliability is insufficient
Solution Approach 1:
The patent transitions from one-dimensional radar detection to three-dimensional detection using binocular stereo vision. By capturing depth information through disparity maps and projecting onto a 3D coordinate system, the system can detect thin height-limiting rods with higher precision and reliability, as the additional spatial dimensions provide more information about the rod's presence and position.
Solution Approach 2:
The patent segments the detection process into distinct stages: filtering invalid points in the disparity map, projecting valid points onto a 3D coordinate system, filtering invalid blobs, and back-projecting to determine the rectangular region. This segmentation allows the system to systematically process the detection of thin rods, improving both precision and reliability by eliminating invalid data at each stage.
3Difficulty of detecting and measuring
If comprehensive sample databases and excellent classifiers are used for visual identification, then detection capability is improved, but the system becomes impossible to implement due to complexity
Solution Approach 1:
The patent replaces the complex visual classification system (requiring comprehensive databases and excellent classifiers) with a geometric projection system based on binocular stereo vision. The detection capability is improved through disparity-based depth estimation and 3D projection, while the system complexity is reduced by eliminating the need for large-scale visual databases and complex classification algorithms.
Data Source
AI summary
The present disclosure provides a method and a device for detecting a height-limiting rod, and an automatic driving system, for use in a binocular stereo camera. The method includes: filtering out invalid points in a disparity map corresponding to a target image in accordance with a predetermined detection range, so as to acquire a valid point disparity map; projecting the valid point disparity map along an X-axis direction in a three-dimensional coordinate system onto a plane where a Y axis and a Z axis are located, and counting the quantity of valid points for each row of pixels with respect to each disparity value, so as to acquire a projection image; filtering out invalid projection points and invalid blobs in the projection image, so as to acquire a valid blob image; and back-projecting the valid blob image to the three-dimensional coordinate system, and determining a rectangular region where the height-limiting rod is located in accordance with X-axis, Y-axis and Z-axis coordinate values.