License Plate Detection via HSV Color Augmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current deep learning-based image detection and recognition algorithms face challenges in effectively detecting and recognizing license plates of different colors, particularly yellow and green, due to limitations in sample data and ineffective data augmentation methods, which hinder the training of high-quality models.
Innovation Solution
A real-time license plate detection and recognition method utilizing color augmentation, where images of blue license plates are transformed into yellow and green license plates through HSV color space manipulation, enabling the training of deep convolutional neural networks for improved detection and recognition capabilities, and incorporating view correction transformations for accurate frontal view images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional data augmentation methods are used to process limited license plate samples, then the quantity of training samples is increased, but the model cannot detect and recognize license plates of different colors (yellow and green)
Solution Approach 1:
The patent applies color space transformation (RGB to HSV) and manipulates the Hue channel to systematically generate license plate images of different colors. By adjusting the Hue value within specific ranges, the method transforms blue license plates into yellow and green variants, enabling the model to learn color-invariant features and improve detection across multiple license plate colors.
Solution Approach 2:
The patent changes the color parameters of license plate images by transforming from RGB to HSV color space and modifying the Hue parameter. This parameter-based approach allows precise control over color transformation while preserving other image features, enabling effective data augmentation that maintains structural integrity while varying color characteristics.
2Measurement precision
If manual inspection is performed to achieve higher annotation accuracy, then the quality of training data is improved, but the time and cost required for data processing increases significantly
Solution Approach 1:
The patent uses image copying and transformation techniques to generate synthetic training samples from existing annotated blue license plate images. By applying color space transformation and Hue parameter adjustment, the system creates yellow and green license plate variants automatically, eliminating the need for manual annotation of each color variant while preserving annotation accuracy.
Solution Approach 2:
The patent performs color augmentation as a preliminary step before model training, systematically generating diverse color variants of license plates in advance. This preliminary data preparation ensures that the model is exposed to multiple color variations during training without requiring manual inspection or annotation during the actual training process.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method enhances the ability of license plate detection and recognition algorithms to handle diverse colors, maintaining objectivity and increasing training opportunities, thereby improving detection and recognition performance in complex environments.
Implementation Method 1
converting RGB color space to HSV color space
Data Source
AI summary
The present invention discloses a real-time license plate detection and recognition method and device based on color augmentation, the method comprises: step 1: obtaining car images containing license plates as a license plate detection training set, inferring license plate detection results through the trained license plate detection model; step 2: performing view correction transformation on the detected license plates to obtain frontal view images of the license plates; step 3, using the obtained frontal view images of the license plates as a training set for license plate recognition, using a license plate recognition model based on deep neural network for license plate recognition to obtain license plate recognition results; step 4: displaying the license plate detection results and the license plate recognition results on the original test images or outputting them as needed, completing the detection and recognition of the license plates in the images. The implementation method of the present invention is simple, has strong portability, improves the accuracy of the license plate detection model and the license plate recognition model, and enhances the generalization ability of the model.