Traffic Light Identification Using Multi-Scale CNN Image Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for autonomous vehicles to detect traffic lights are inaccurate due to failure to consider lighting changes and weather conditions, leading to false and missed detections.
Innovation Solution
A method and apparatus using a pre-trained convolutional neural network that zooms images and analyzes location and category information to identify traffic lights, trained with a machine learning method based on a preset image set, regression, and classification loss functions, to improve detection accuracy across various lighting and weather conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If color space conversion method is used for traffic light detection, then the detection process is simple, but the detection accuracy is low due to false and missed detections under varying lighting and weather conditions
Solution Approach 1:
The patent transforms the input image into multiple color spaces (RGB, HSV, LAB, YCbCr) and extracts features from each space. By analyzing traffic light characteristics across different color representations, the system adapts to varying lighting and weather conditions that affect color perception, thereby improving detection accuracy without excessive complexity
Solution Approach 2:
The patent combines multiple color space analyses with shape-based detection methods. By fusing features from different color spaces and incorporating geometric shape information, the system creates a composite detection approach that overcomes the limitations of single-method detection under diverse environmental conditions
2Measurement precision
If multiple color spaces and feature fusion are used to improve detection accuracy, then the detection precision improves, but the computational complexity increases
Solution Approach 1:
The patent divides the detection process into distinct stages: color space transformation, feature extraction from each space, shape analysis, and result fusion. By segmenting the complex detection task into manageable modules, the system achieves high accuracy through comprehensive analysis while maintaining organized computational flow that can be optimized
3Ease of manufacture
If existing color-based detection methods are used, then the implementation is straightforward, but the system fails to adapt to lighting changes and weather conditions
Solution Approach 1:
The patent implements a universal detection framework that processes images through multiple color spaces (RGB, HSV, LAB, YCbCr) and integrates shape-based detection. This multi-functional approach enables the system to adapt to various lighting conditions (daylight, nighttime, twilight) and weather scenarios by selecting and combining the most effective detection pathways for each situation
Data Source
AI summary
The disclosure discloses a method and apparatus for identifying a traffic light. An embodiment of the method comprises: zooming a to-be-processed image acquired by an image acquisition device by at least one preset ratio to obtain at least one zoomed image; inputting the at least one zoomed image into a pre-trained convolutional neural network to obtain location information and category information of a traffic light corresponding to each zoomed image of the at least one zoomed image, wherein the convolutional neural network is used for retrieving location information and category information of a traffic light displayed in an image; and analyzing the obtained location information and category information to generate at least one candidate traffic light identification result, and fusing the generated candidate traffic light identification result to generate a traffic light identification result corresponding to the to-be-processed image. The embodiment improves the accuracy in identifying a traffic light.