Optoelectronic Code Reader Using Machine Learning Classifier
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional code readers face significant challenges in achieving high reading rates due to information loss from binarization and edge detection, leading to errors in decoding barcodes and two-dimensional codes, especially with issues like small bar widths, noise, and contrast problems.
Innovation Solution
An optoelectronic code reader equipped with a machine learning classifier that uses supervised learning based on data from a classic decoder, trained with gray level information to improve decoding accuracy and adapt to specific reading situations, capable of handling low resolution, blurred codes, and noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If binarization and edge detection are used in conventional decoders, then decoding speed is improved, but information loss occurs leading to reading errors
Solution Approach 1:
The patent changes the parameter representation from binary (0/1) to continuous gray level values (0-255). Instead of converting gray level images to binary edge position arrays, the system uses the full gray level information directly as input features for the neural network classifier, preserving all original information while enabling accurate decoding through machine learning.
2Ease of operation
If edge position arrays are generated from gray level profiles, then decoding can be performed, but reading errors increase due to information loss
Solution Approach 1:
The patent introduces a neural network classifier as an intermediary between the gray level profile acquisition and the final decoding. This classifier processes the full gray level information and outputs decoded results, replacing the conventional edge detection and binary classification chain. The neural network acts as a learned mediator that maintains information integrity while performing the decoding function.
3Adaptability or versatility
If multiple code readers are provided in a reading tunnel, then objects in any orientation can be read, but device complexity increases
Solution Approach 1:
The patent makes a single code reader universal by equipping it with a neural network classifier trained to recognize codes in various orientations. Instead of requiring multiple specialized readers for different orientations, one reader with multi-functional classification capability can handle all orientation cases, reducing system complexity while maintaining versatility.
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 solution significantly enhances reading rates by utilizing the full input information from gray level profiles, reducing errors, and improving decoding accuracy, even in challenging conditions, while maintaining high reliability and adaptability.
Implementation Method 1
at least one light receiving element for generating image data from reception light
Data Source
AI summary
An optoelectronic code reader (10) having at least one light receiving element (24) for generating image data from reception light and an evaluation unit (26) with a classifier (30) being implemented in the evaluation unit (26) for assigning code information to code regions (20) of the image date, wherein the classifier (30) is configured for machine learning and is trained by means of supervised learning based on codes read by a classic decoder (28) which does not make use of methods of machine learning.