Driving License Validity Extraction via Neural Network Sorting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Extracting the annual inspection validity period of a driving license is challenging due to issues like blurs, shadows, and unclear images, as well as multiple records not in chronological order, which interfere with accurate targeting and processing.
Innovation Solution
A method and system using a processor and deep convolutional neural networks to extract candidate contour regions, recognize and sort character blocks, and determine the latest validity period by analyzing image files of driving licenses, employing algorithms like MSER and Otsu's threshold method for image processing and contour extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing methods are used to extract driving license information, then the process requires manual intervention, but the extraction accuracy decreases due to blurs, shadows, and unclear images
Solution Approach 1:
The patent introduces an intermediary system comprising image quality assessment module and preprocessing module that mediates between the raw image and the extraction process. The system assesses image quality metrics (blurriness, shadow intensity, clarity) and applies appropriate preprocessing operations to enhance the image before extraction, thereby improving accuracy without requiring manual intervention throughout the process
Solution Approach 2:
The system enables self-service extraction by implementing automated quality assessment and adaptive preprocessing that operates without human intervention. The extraction module automatically adjusts its parameters based on the assessed image quality, and the system self-corrects for common issues like blurs and shadows through algorithmic preprocessing, eliminating the need for manual image correction while maintaining high extraction accuracy
2Loss of information
If multiple annual inspection records are present on the driving license, then complete information is available, but the complexity of identifying the latest record increases
Solution Approach 1:
The patent applies preliminary action by implementing a sorting mechanism that automatically arranges multiple inspection records in chronological order before extraction. The system identifies temporal markers in the records, sorts them by date, and presents them in sequence, thereby simplifying the identification of the latest record. This preliminary sorting action reduces the complexity from requiring manual comparison of multiple unsorted records to simply selecting the last record in the sorted sequence
Solution Approach 2:
The system segments the multiple inspection records into individual extractable units, each with its own temporal metadata. By dividing the complex task of analyzing multiple records into separate, standardized segments, the system can process each record independently and automatically identify the latest one through systematic comparison of temporal markers, rather than treating all records as a single complex entity
3Extent of automation
If deep convolutional neural networks are used for extraction, then automation level increases, but the computational resources and processing time increase
Solution Approach 1:
The patent implements partial action by using the deep convolutional neural network only for the specific task of extracting the annual inspection validity period, rather than applying it to the entire driving license processing workflow. The system uses traditional image processing and rule-based methods for other extraction tasks, reserving the computationally intensive neural network for the most challenging and critical extraction task, thereby achieving high automation where needed while controlling overall computational resource consumption
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of automatically extracting a validity period of a driving license is disclosed. The method include: extracting candidate contour regions in an image file of a vehicle driving license and selecting from among the candidate contour regions at least one containing an annual inspection validity period; extracting character contours in each selected candidate contour region; acquiring a smallest circumscribed rectangle of each character contour, and dividing the acquired smallest circumscribed rectangles into individual single-character blocks; sorting these single-character blocks according to their positions in the contour region; recognizing a corresponding character of each single-character block; generating a character string representing the annual inspection validity period using the recognized characters based on the sorting; and extracting and taking time data in the character string as the annual inspection validity period of the contour region. A device, system and storage medium to which the method is applicable are also disclosed. By this application, the annual inspection validity period of a driving license can be automatically extracted.