Optical Sensor Neural Network Code Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optical sensors for barcode detection face limitations due to non-scalable digital filter structures, leading to unsatisfactory results in barcode detection.
Innovation Solution
The use of neural networks as differentiation units in the evaluation unit of optical sensors, allowing for scalable complexity and reliable detection of various codes by learning coefficients through training with known codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If digital filter structures with differential tasks are used, then edge detection capability is provided, but the complexity of the filter structures cannot be freely scaled and detection reliability is limited
Solution Approach 1:
The patent changes the fundamental parameter of the differentiation unit from fixed digital filter coefficients to trainable neural network weights. This allows the complexity to be freely scaled by adjusting the number of layers and neurons in the neural network, while achieving superior detection reliability through learned optimal parameters rather than fixed filter designs.
Solution Approach 2:
The patent replaces the mechanical digital filter structure with a neural network-based differentiation unit. This substitution enables the system to learn optimal differentiation patterns from training data, achieving higher reliability while allowing flexible complexity scaling that was not possible with conventional digital filters.
2Measurement precision
If digital filter structures are used, then differentiation of intensity signals is enabled, but the filter functions are limited and detection results are unsatisfactory
Solution Approach 1:
The patent applies preliminary action by training the neural network beforehand on a comprehensive dataset of code patterns. This pre-training enables the differentiation unit to achieve high measurement precision for edge detection across various code types and conditions, overcoming the limited adaptability of fixed digital filter functions.
Solution Approach 2:
The neural network-based differentiation unit achieves universality by being able to detect edges across multiple code types (barcodes, data matrices, etc.) and varying conditions. The single trained model replaces multiple specialized digital filters, providing both high precision and broad adaptability.
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
This approach enables significantly improved and reliable detection of codes, including barcodes, by quickly converging coefficients and accurately differentiating received signals to determine code edges.
Implementation Method 1
a transmitter (3) emitting light beams (2) and a receiver (4) receiving light beams (2), wherein the transmitter and receiver are designed to scan codes
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention relates to an optical sensor (1) for detecting codes, comprising a light beam emitting transmitter (2) and a light beam receiving receiver (4), which are configured for scanning codes, and an evaluation unit configured for determining code information of the codes depending on received signals from the receiver (4). The evaluation unit includes neural networks (11, 12). The invention further relates to a corresponding method.