Vehicle Peripheral Image Pedestrian Detection Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional rear-view cameras in vehicles fail to accurately detect the presence or position of a person due to various postures, partial body hiding, or fisheye lens distortion, leading to inefficiencies in pedestrian detection.
Innovation Solution
A moving object information providing device and method that acquires peripheral images, extracts images of a vehicle's upper and lower body parts, and determines the presence or position of moving objects using these images, reducing computational complexity and improving detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If part-based pedestrian detection method is adopted to cope with various environments, then detection accuracy is improved, but calculation amount increases significantly
Solution Approach 1:
The patent segments the pedestrian detection task into two distinct stages: first detecting the lower body (legs, feet) which is easier to identify, then using this information to guide detection of the upper body. This segmentation reduces the overall computational burden while maintaining detection accuracy across various environments and postures.
Solution Approach 2:
The patent performs preliminary detection of the lower body before attempting to detect the upper body. By first identifying the lower body region and using it as a reference, the system prepares the detection process in advance, reducing the complexity of subsequent upper body detection and improving overall efficiency.
2Device complexity
If conventional rear-view camera is used, then device simplicity is maintained, but detection accuracy of moving objects fails due to various postures and fisheye lens distortion
Solution Approach 1:
The patent divides the detection process into separate stages for lower body and upper body detection, allowing each stage to be optimized independently. This segmentation enables the system to handle various postures and fisheye distortion more effectively while maintaining a relatively simple camera-based device architecture.
Solution Approach 2:
By performing preliminary lower body detection first, the system establishes a reference framework that helps correct for fisheye lens distortion and accounts for various postures before attempting upper body detection, thereby improving overall accuracy without adding complex hardware.
Data Source
AI summary
A device for providing moving body information for a vehicle, according to one embodiment, includes an image acquisition unit acquiring a peripheral image of the vehicle, an image extraction unit extracting at least one of a first image of an upper body of at least one moving object or a second image of a lower body of the moving object from the peripheral image of the vehicle, and a moving object information determination unit determining at least one of presence/absence or a position of the moving object using at least one of the extracted first image or second image, wherein the image extraction unit extracts as the first image at least one of a ‘1-1’-st image of an entirety of the upper body or a ‘1-2’-nd image of a part of the upper body, and extracts as the second image at least one of a ‘2-1’-st image of an entirety of the lower body or a ‘2-2’-nd image of a part of the lower body.