Surrounding Image Synthesis for Lane Departure Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional lane departure warning systems struggle to accurately determine lane departure and provide clear visual cues to drivers due to the lack of natural image overlay from multiple camera views, leading to difficulty in recognizing driving conditions and potential accidents from fatigue or drowsiness.
Innovation Solution
A method and apparatus that synthesizes images from multiple cameras installed around a vehicle to generate a top-view surrounding image, overlaying revised images using a mask image with region and weight information to create a natural view, allowing for accurate lane detection and departure warning by calculating distance and angle information between the vehicle and the lane.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple camera images are simply combined and displayed, then the system can capture comprehensive vehicle surroundings, but the images cannot be naturally overlaid and the current vehicle location and lane departure degree cannot be accurately determined
Solution Approach 1:
The patent transforms multiple 2D camera images from different viewpoints into a unified top-down bird's-eye view representation. This dimensional transformation allows images from front, rear, left, and right cameras to be naturally overlaid in a common coordinate system, enabling accurate determination of vehicle location and lane departure degree by providing a comprehensive overhead perspective that integrates all surrounding views.
Solution Approach 2:
The patent introduces a mask image as an intermediary element that contains region information and weight information for pixels. This mask serves as a mediator to control the overlay process, determining which regions from different camera images should be visible and with what transparency weights, thereby achieving natural image synthesis while accurately representing the vehicle's surroundings and lane position.
2Ease of operation
If conventional methods use only front image or simple left-right image combination, then the system complexity is reduced, but the driver has difficulty easily checking driving condition and lane departure degree
Solution Approach 1:
The patent creates a top-down bird's-eye view that provides a comprehensive overhead perspective of the vehicle surroundings, integrating front, rear, left, and right camera images. This dimensional transformation allows the driver to easily check driving conditions and lane departure degree in a single unified view, while preserving complete surrounding information that would be lost in conventional single-view or simple combined displays.
Data Source
AI summary
A method and an apparatus for sensing a lane departure using images that surround a vehicle. The lane departure sensing method includes receiving images that are taken during driving by a plurality of cameras installed in a car, through respective channels connected to the cameras; modifying the obtained images into top-view images to generate a plurality of modified images; compositing the plurality of modified images to generate images that surround a vehicle; extracting a lane from the images that surround the vehicle, and yielding the information on a distance between the vehicle and the extracted lane or the information on an angle between the travel direction of the vehicle and the lane, so as to determine a lane departure; and giving a warning if the lane departure is determined.