Vehicle Pedestrian Detection via Radar-Guided Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle safety systems face challenges in accurately and efficiently recognizing pedestrians, particularly when multiple objects are present, leading to delayed recognition and increased processing loads, which can result in ineffective safe driving support.
Innovation Solution
An external environment recognition device that uses a combination of radar and camera data to set a predicted course for the vehicle, determine collision risks, select priority objects, and perform targeted image processing to quickly identify pedestrians, thereby reducing processing loads and ensuring timely recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image processing is performed on all detected objects to improve pedestrian recognition accuracy, then recognition accuracy is improved, but processing load is increased
Solution Approach 1:
The patent extracts and prioritizes specific objects that are most likely to be pedestrians based on radar data (position, velocity, course) before applying image processing. This selective extraction approach processes only high-probability candidate objects rather than all detected objects, thereby maintaining recognition accuracy while reducing overall processing load.
Solution Approach 2:
The system performs preliminary collision possibility determination using radar data before initiating image processing. By pre-filtering objects based on their likelihood of being pedestrians (using velocity, position, and predicted course information), the system prepares a reduced set of candidate objects for subsequent image processing, thus reducing processing load while preserving accuracy for critical cases.
2Productivity
If image processing is prioritized for distant immobile pedestrians to reduce processing load, then processing load is reduced, but recognition delay occurs for crossing pedestrians
Solution Approach 1:
The patent dynamically adjusts the prioritization of image processing targets based on real-time collision risk assessment. Objects with higher collision possibility (such as crossing pedestrians with velocity vectors pointing toward the vehicle) are dynamically prioritized over distant immobile objects. This dynamic reordering ensures that processing load is reduced while critical time-sensitive detections are not delayed.
Solution Approach 2:
The system changes the prioritization parameters for image processing based on collision risk metrics. Instead of static prioritization (e.g., always processing distant objects first), the system adjusts priority levels according to dynamic parameters such as relative velocity, predicted course intersection, and time-to-collision, ensuring that high-risk objects receive timely processing attention.
3Measurement precision
If complex image processing logic is implemented to handle various pedestrian patterns, then recognition accuracy is improved, but device complexity is increased
Solution Approach 1:
The patent segments the pedestrian detection process into distinct stages: initial radar-based object detection, collision possibility determination, candidate selection, and targeted image processing. This segmentation allows each stage to handle specific tasks with simpler logic, avoiding the need for a single complex image processing system to handle all patterns, thereby reducing overall device complexity while maintaining accuracy.
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
The system effectively prioritizes pedestrian detection, reducing recognition delays and ensuring safety by selectively focusing image processing on high-risk, proximity pedestrians, and adaptively managing image processing regions to minimize load while maintaining safety.
Implementation Method 1
a distance measuring sensor, such as a radar
Implementation Method 2
an image sensor, such as a camera
Data Source
AI summary
An external environment recognition device for a vehicle and a vehicle system capable of assuring safety and reducing the processing load in a compatible manner are provided. An external environment recognition device 100 for a vehicle includes: first collision determination means 103 for computing a risk of collision of a host vehicle with a detected object in front of the host vehicle on the basis of information of a predicted course of the host vehicle and the detected object; and second collision determination means 104 for determining whether the detected object enters the predicted course from the outside of the predicted course or not. This device further includes object selection means 105 for selecting the detected object having a risk at least a first threshold and the detected object determined to enter the predicted course as selection candidate objects from among the detected objects, and for selecting the selection candidate object having a minimum relative distance to the host vehicle as a pedestrian determination request object from among the selected selection candidate objects. This device further includes pedestrian determination means 106 for determining whether the selected pedestrian determination request object is a pedestrian or not using image information.