Rear Vehicle Recognition Using Day-Night Camera Distance Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing rear collision warning systems face challenges in effectively detecting rear approaching vehicles due to limited detection range and interference issues with active sensors, leading to increased vehicle costs and potential collisions, especially when rear sensors are not adequately integrated with existing front sensors.
Innovation Solution
A method utilizing an existing rear-mounted camera and deep learning network to recognize rear vehicles by determining time zones and adjusting learning parameters based on day/night conditions, combined with image processing to calculate distances and speeds, without adding additional hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If active sensors such as RADAR and LiDAR are mounted on the rear of the vehicle to enhance detection range, then the detection capability is improved, but the front sensor of the rear vehicle experiences interference and performance deteriorates
Solution Approach 1:
The patent extracts the active sensing function from the rear-mounted RADAR/LiDAR and relocates it to the front-mounted sensors. The front camera and existing front sensors are utilized to detect rear vehicles by processing images and sensor data, thereby eliminating interference to rear sensors while maintaining detection capability.
Solution Approach 2:
The front-mounted camera and sensors are made multi-functional by enabling them to perform both forward collision avoidance and rear vehicle detection. Through image processing and coordinate transformation algorithms, the same hardware serves dual purposes, eliminating the need for separate rear active sensors.
2Measurement precision
If expensive active sensors such as RADARs and LiDARs are installed for rear collision control to improve detection range, then the detection capability is improved, but the vehicle price increases
Solution Approach 1:
The patent makes the existing front camera and sensors multi-functional, enabling them to perform rear vehicle detection in addition to their primary forward collision avoidance function. This eliminates the need for expensive additional rear-mounted active sensors like RADAR and LiDAR, thereby reducing vehicle cost while maintaining detection capability.
Solution Approach 2:
The patent creates a virtual model of the rear vehicle by processing and transforming image data from the front camera. Through coordinate transformation and depth estimation algorithms, the system generates accurate spatial information about rear vehicles using only passive optical sensors, replacing the need for expensive active sensing hardware.
3Device complexity
If general rear sensors are used to detect rear vehicles, then the system complexity is reduced, but the detection distance is relatively short and collision may occur before response
Solution Approach 1:
The patent replaces traditional mechanical/radar-based rear sensing systems with an optical computing approach. The front camera captures images of rear vehicles, and through image processing, coordinate transformation, and depth estimation algorithms, the system calculates accurate distance and position information, achieving long-range detection with simple hardware.
Solution Approach 2:
The patent transforms the detection problem from a spatial dimension issue to an information processing dimension. By capturing images from the front camera and using computational methods to extract depth and distance information, the system achieves extended detection range without adding physical sensors to the rear, effectively solving the short detection distance problem through dimensional transformation in data space.
Data Source
AI summary
Disclosed are a method of recognizing a vehicle based on an image and an apparatus and vehicle therefor. A method of recognizing a rear vehicle in a vehicle includes acquiring an image captured by a camera while driving, determining a time zone based on the acquired image, performing object recognition based on the determined time zone, determining image coordinates of the rear vehicle based on a result of the object recognition, and converting the determined image coordinates into a distance.