Light Spot Clustering Using Distance Measures for Complex ADAS Scenes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current camera-based Advanced Driver Assistance Systems (ADAS) for detecting headlights and tail lights rely on handpicked rules for pairing and clustering, which are not optimal and inflexible, especially in scenes with a high amount of light spots.
Innovation Solution
A device and method for clustering light spots using a distance measure that quantifies dissimilarity between light spots, allowing for flexible clustering by comparing each light spot to potential clusters based on features such as blob area, angle, center of gravity, color, intensity, and Euclidean distance, utilizing a Bayes classifier for feature selection and distance calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If handpicked rules are used for pairing and clustering light spots, then the device complexity is reduced, but the clustering accuracy and flexibility deteriorate
Solution Approach 1:
The patent transforms fixed handpicked rules into adaptive distance measures by changing parameters dynamically. Multiple features (blob area, angle, center of gravity, color, intensity, Euclidean distance) are combined with learned weights to create flexible distance calculations that adapt to different scene conditions, resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The patent replaces manual rule-based clustering with an automated machine learning system. The Bayes classifier and distance measure calculations automatically learn optimal clustering parameters from data, substituting mechanical rule-setting with intelligent automated decision-making that improves accuracy without proportionally increasing complexity.
2Speed
If traditional pairing methods are used for light spots, then the processing speed is maintained, but the adaptability to complex scenes deteriorates
Solution Approach 1:
The patent performs preliminary feature extraction and distance measure preparation for all light spot pairs before actual clustering. By pre-calculating features like blob area, angle, center of gravity, color, and intensity, the system prepares data structures that enable fast clustering decisions during execution, maintaining speed while improving adaptability.
Solution Approach 2:
The patent introduces dynamic adaptability through distance measures that can adjust their importance weights based on scene conditions. The clustering algorithm dynamically selects and weights different features (color, intensity, position) according to the specific traffic scene, allowing the system to adapt to varying lighting conditions, vehicle types, and environmental factors while maintaining processing efficiency.
Data Source
Figure 1~2
Figure 3
Figure 4A~4B
AI summary
A device for clustering light spots, the device is configured to receive an image captured by a camera, detect light spots in the image, calculate a distance measure for each of a plurality of pairs of light spots detected in the image, and group the detected light spots to clusters based on the calculated distance measures.