On-Vehicle Camera Foreign Matter Detection via Feature Point Accumulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing on-vehicle camera systems struggle to accurately detect and locate foreign matters like raindrops or dirt on the lens, which affects lane recognition and parking frame detection, especially in non-periodic or dirty road conditions, limiting their applicability and accuracy.
Innovation Solution
An on-vehicle control device that acquires images, divides the image-capturing range into regions, extracts feature points like traffic lane lines, accumulates points over time, and judges foreign matter adhesion based on cumulative scores, allowing for precise determination of adhesion positions and quick judgment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the length and distance of dashed traffic lane lines are recognized to detect foreign matter adhesion, then foreign matter detection is achieved, but processing becomes complicated
Solution Approach 1:
The patent extracts only the necessary information (presence/absence of feature points in divided regions) rather than analyzing complete lane line characteristics. By taking out only the essential detection elements and discarding unnecessary processing steps, the system achieves foreign matter detection without complicated processing of lane line length and distance measurements.
2Reliability
If only presence or absence of uniformity of dashed traffic lane lines is judged, then foreign matter adhesion is detected, but position information of adhered foreign matters is not obtained
Solution Approach 1:
The patent divides the image-capturing range into multiple regions and independently judges foreign matter adhesion in each region based on feature point distribution. This segmentation approach enables the system to detect not only the presence of foreign matter but also its specific position, as each region's judgment result corresponds to a particular spatial location in the captured image.
3Reliability
If the approach based on periodicity of white lines is used, then foreign matter detection is achieved, but it becomes inapplicable when white lines have no periodicity due to dirty road or faded lines
Solution Approach 1:
The patent uses feature points that are automatically extracted from the captured image without requiring pre-defined periodicity patterns or manual configuration. The system adapts to various road conditions (dirty roads, faded lines) by extracting feature points directly from the actual image content, making the detection method self-adjusting and universally applicable without relying on ideal periodic patterns.
Data Source
AI summary
An on-vehicle control device includes: an image acquiring unit that acquires a captured image from an image-capturing device, the image-capturing device capturing an image of surroundings of a vehicle; a judged range dividing unit that divides a predetermined judged range of an image-capturing range of the image-capturing device into a plurality of regions; a feature point extracting unit that extracts a feature point having predetermined characteristics in the surroundings of the vehicle from the image acquired by the image acquiring unit; a point accumulating unit that accumulate points for a region where there is the feature point extracted by the feature point extracting unit, among the plurality of regions, for a plurality of images acquired by the image acquiring unit in time series; and a judging unit that judges whether a foreign matter is adhered to the image-capturing device based on scores resulting from accumulating points of the plurality of regions.