Windshield Camera Alignment and Shadow Filtering for Lane Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lane change aid and side object detection systems generate false positive and false negative detections due to their reliance on statistical methodologies and frame-by-frame analysis, which can lead to driver annoyance and increased risk of accidents, as they fail to accurately differentiate between vehicles and shadows, and struggle with vehicles moving at the same speed as the equipped vehicle.
Innovation Solution
An object detection system that uses edge detection algorithms to identify vehicles by processing a subset of image data focused on a target zone, adjusting for camera misalignment and vehicle turns, and distinguishing between vehicles and shadows, while reducing processing requirements through filtering mechanisms to minimize false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If statistical methodologies are used to analyze all pixels in captured images to detect vehicles, then detection coverage is improved, but false positive detections increase and processing complexity increases
Solution Approach 1:
The patent segments the image processing task by dividing the image into multiple zones (first zone, second zone, third zone) with different processing strategies. The first zone contains fewer pixels and is processed with full analysis, while the second and third zones contain more pixels and are processed with reduced analysis, thereby reducing overall processing complexity while maintaining detection coverage.
Solution Approach 2:
The patent applies different processing qualities to different regions of the image. High-quality processing (full pixel analysis) is applied to the first zone where vehicle detection is most critical, while lower-quality processing (reduced pixel analysis) is applied to the second and third zones, optimizing the balance between detection accuracy and processing complexity.
2Speed
If frame by frame flow algorithms are used to track each pixel to detect vehicle movement, then movement detection capability is improved, but processing time increases and vehicles moving at same speed as equipped vehicle cannot be detected
Solution Approach 1:
The patent extracts only the necessary information for movement detection by analyzing specific zones rather than tracking every pixel in the entire image. This extraction approach allows the system to detect vehicle movement while significantly reducing processing time and enabling detection of vehicles moving at the same speed as the equipped vehicle.
3Reliability
If all pixels are continuously analyzed to detect vehicles in adjacent lanes, then detection completeness is improved, but false positive detections increase due to inability to distinguish vehicles from shadows
Solution Approach 1:
The patent segments the analysis into different zones where the first zone receives comprehensive processing to distinguish vehicles from shadows, while the second and third zones receive reduced processing. This segmentation maintains detection completeness in critical areas while reducing false positives in areas where full analysis is less critical.
Data Source
AI summary
A vehicular imaging system includes a camera operable to capture image data. The camera is configured for attachment at an upper windshield area of an in-cabin side of a windshield of a vehicle. An image processor is operable for processing image data captured by the camera. Captured image data is processed by the image processor for a collision avoidance system of the equipped vehicle. Captured image data is processed by the image processor for a lane departure warning system of the equipped vehicle. Responsive at least in part to processing by the image processor of captured image data, shadows viewed by the camera are detected. The vehicular imaging system determines misalignment of the camera responsive at least in part to processing by the image processor of captured image data.