Stereo Camera Dynamic Cropping for Curved Road Obstacle Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing stereo camera systems face challenges in accurately determining the image position for obstacle detection on curved roads and roads without white lines, leading to incomplete gradient determination and ineffective distance measurement.
Innovation Solution
A stereo camera device incorporating a wide-angle image cropping unit, a distant image cropping and enlargement unit, and a road shape determination unit, which adjusts the cropping position and range based on road shape and gradient information to enhance image positioning and distance measurement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If disparity calculation is performed by enlarging the central portion of the image, then the measurement range to distant objects is improved, but the traveling destination is not accurately captured when the road is curved and the traveling direction is out of the center
Solution Approach 1:
The image cropping position is dynamically adjusted based on the detected road shape. Instead of using a fixed central cropping position, the system determines the road's curvature and gradient, then shifts the cropping position along the road's direction to ensure the traveling destination remains within the enlarged measurement range. This dynamic adaptation resolves the contradiction between maintaining central cropping for distance measurement and capturing curved road destinations.
Solution Approach 2:
The system changes the cropping position parameter based on road shape parameters (curvature and gradient). By detecting the road's geometric characteristics and using them to adjust the cropping position, the system adapts the image processing parameters to match the actual driving conditions, ensuring accurate obstacle detection on both straight and curved roads.
2Adaptability or versatility
If image position is determined based on steering angle or white line detection, then the measurement range can be adjusted, but the method fails on roads without white lines or when the steering wheel has not yet turned before entering a curve
Solution Approach 1:
The system uses the camera's own captured images to determine road shape, rather than relying on external indicators like white lines or steering angle. By detecting the road's geometric features directly from the captured image data, the system becomes self-sufficient and can operate reliably on any road condition, including roads without white lines or before the driver reacts to a curve.
Solution Approach 2:
The system replaces mechanical indicators (steering wheel position) and visual markers (white lines) with image processing-based road shape detection. By using computational methods to analyze the road geometry from camera images, the system eliminates dependencies on physical or visual cues that may not be available, achieving universal adaptability across all road conditions.
3Ease of manufacture
If the cropping position is fixed at the center of the image, then the system is simple to implement, but the gradient of the road cannot be accurately determined and obstacles on curved roads cannot be detected early
Solution Approach 1:
The system performs preliminary road shape determination by analyzing the captured images to detect road curvature and gradient before proceeding with obstacle detection. This preliminary action of characterizing the road geometry allows subsequent obstacle detection to be performed in the correct spatial context, improving both accuracy and early detection capability while maintaining a straightforward implementation structure.
Data Source
AI summary
An objective of the present invention is, in a stereo camera device, to determine an accurate image position in a direction of progress to detect at an early stage an obstacle or a preceding vehicle on a road. Provided is a stereo camera device for measuring the distance to a solid object from images photographed with a plurality of cameras, said device characterized by: a wide-angle image cropping part for cropping a portion of the images; a distance image cropping part for cropping and enlarging a portion of the images; a road shape determination part for determining a road shape, including slope information, of a road being traveled; and determining, on the basis of the road shape in a prescribed distance, which has been derived with the road shape determination part, the cropping position and/or range of the distance image cropping part.