Vehicle Light Detection Probability Calculation Reducing Processing Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems for detecting vehicle headlights or tail lights from images face accuracy issues due to unclear color detection methods, leading to increased processing loads when trying to enhance detection accuracy.
Innovation Solution
An apparatus that calculates a target area probability by correlating light-source feature quantities, such as color and shape, with preset feature quantities of vehicle lights, reducing processing load by setting a target area within a specific portion of the light source area, and using a program to execute these calculations efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the sampling frequency for detecting color is raised to detect color at multiple points of the light source area, then the accuracy of detection is improved, but the processing load is excessively increased
Solution Approach 1:
The light source area is divided into multiple target areas, and color detection is performed separately in each target area. This segmentation allows the system to focus detection efforts on specific regions where vehicle lights are most likely to appear, improving detection accuracy without requiring exhaustive sampling of the entire light source area.
Solution Approach 2:
Different target areas within the light source area are assigned different detection characteristics. The system correlates light-source feature quantities (color and shape) with target-area feature quantities that are specifically prepared for vehicle light detection. This local quality approach enables accurate detection in critical areas while reducing processing in less critical areas.
2Measurement precision
If color is detected at multiple points of the light source area, then the accuracy of detecting vehicle light is improved, but the processing time is increased
Solution Approach 1:
Target areas and their corresponding feature quantities are prepared in advance before actual detection occurs. The system pre-establishes the correlation between light-source feature quantities and target-area feature quantities for vehicle light detection. This preliminary preparation eliminates the need for complex real-time calculations, reducing processing time while maintaining high detection accuracy.
Solution Approach 2:
The system extracts only the necessary color information from specific target areas rather than processing the entire light source area. By taking out only the relevant feature quantities from predetermined target areas, the system achieves accurate vehicle light detection with minimal processing time.
Data Source
AI summary
An on-vehicle apparatus detects other vehicle lights from an image picked up outside the vehicle. The apparatus acquires an image around the vehicle and extracts light sources from the image. The apparatus calculates a target area probability that is a probability of a light source in a target area being different vehicle light, in accordance with a probability that light-source feature quantities at least including either of a color of light source and a shape of light source coincide with target-area feature quantities, the target area indicating a preset site that is a part of an area indicating each of the light sources. The target-area feature quantities are prepared as feature quantities of vehicle light, being correlated to the position of a target area. The apparatus outputs information that a light source having the target area probability of not less than a preset threshold is different vehicle light.