Parallel Character Sequence Recognition Using Dual Feature Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image identification models for sequence recognition, such as license plate numbers and barcodes, suffer from low identification efficiency due to serial identification processes.
Innovation Solution
A sequence identification method utilizing a convolutional neural network (CNN) and fully connected layer to perform feature extraction and time sequence relationship extraction, followed by parallel character identification based on first and second image features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If serial identification manner is used for license plate numbers and barcodes, then the identification process is simple, but the identification efficiency is low
Solution Approach 1:
The patent segments the identification task into two independent parts: first extracting spatial features (first feature map) and then extracting temporal sequence features (second feature map). This segmentation allows parallel processing of spatial and temporal dimensions, improving identification efficiency while maintaining process manageability through modular architecture.
Solution Approach 2:
The patent transitions from traditional serial identification to a two-dimensional approach by introducing temporal sequence relationship extraction as a separate dimension alongside spatial feature extraction. This dimensional expansion enables parallel processing across both spatial and temporal dimensions, significantly improving identification efficiency.
2Productivity
If parallel character identification is implemented, then identification efficiency is improved, but the processing complexity increases
Solution Approach 1:
The patent performs preliminary feature extraction to generate the first feature map before the parallel identification process. This preliminary action prepares the data in advance, allowing the subsequent parallel character identification to proceed efficiently without increasing overall processing complexity, as the heavy lifting is done beforehand.
Solution Approach 2:
The patent introduces the second feature map as an intermediary that merges upper and lower information from the image. This intermediary component facilitates parallel character identification by providing enriched feature representations that capture both spatial and temporal relationships, enabling efficient parallel processing without excessive complexity.
Data Source
AI summary
A sequence identification method and apparatus, an image processing device and a storage medium, which belong to the field of image identification, are provided. The method includes: performing feature extraction on a to-be-identified target image through an image identification model to obtain a first feature map, where the first feature map includes a plurality of first image features; performing time sequence relationship extraction on the first feature map based on a convolutional neural network layer and a fully connected layer in the image identification model, to obtain a second feature map that merges upper and lower information included in the to-be-identified target image, where the second feature map includes a plurality of second image features; and performing character identification on the to-be-identified target image in parallel based on the plurality of first image features and the plurality of second image features to obtain a character sequence.