Machine Learning Code Reading for Curved, Reflective Surfaces

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Optically readable codes embedded in the surfaces of plant and animal products, such as fruits and vegetables, suffer from low contrast, distortions, and reflections, making them difficult to read using consumer devices like smartphones.

Innovation Solution

A machine learning model is trained using transformed reference images to enhance contrast and reduce distortions and reflections, enabling accurate decoding of optically readable codes on diverse surfaces.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If optically readable codes are embedded into the surface of plant or animal products using a laser, then the codes can be applied directly to the product surface, but the contrast between the code and surrounding tissue is lower making them more difficult to read

Engineering Contradiction:
Improvecode applicationVSAvoidcode readability
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent applies preliminary image processing transformations to the captured code image before decoding. The system performs contrast enhancement, distortion correction, and reflection reduction as preprocessing steps to improve the readability of laser-etched codes on product surfaces.

Inventive Principle:
Principle #10Preliminary action

2Ease of manufacture

If optically readable codes are applied to curved surfaces such as fruits and vegetables, then the codes can be applied to the product surface, but the codes become distorted making them difficult to read

Engineering Contradiction:
Improvecode applicationVSAvoidcode distortion
Core Design Contradiction:
Ease of manufactureVSShape

Solution Approach 1:

The system performs distortion correction as a preliminary processing step before decoding. Image processing algorithms detect and correct the geometric distortions caused by applying codes to curved surfaces, restoring the code to its original rectangular form for accurate reading.

Inventive Principle:
Principle #10Preliminary action

3Ease of manufacture

If optically readable codes are applied to smooth surfaces, then the codes can be applied to the product surface, but reflections from ambient light occur during reading making reading more difficult

Engineering Contradiction:
Improvecode applicationVSAvoidlight reflections
Core Design Contradiction:
Ease of manufactureVSObject-affected harmful factors

Solution Approach 1:

The system performs reflection reduction as a preliminary image processing step. Algorithms detect and remove reflection artifacts from the captured code image, enhancing the visibility of the code elements by suppressing spurious light reflections from smooth surfaces.

Inventive Principle:
Principle #10Preliminary action

4Stability of the object's composition

If the surfaces of fruits and vegetables are uneven with specks and spots or bumps, then the natural surface characteristics are preserved, but the inconsistencies and bumps make codes difficult to read

Engineering Contradiction:
Improvesurface characteristicsVSAvoidcode readability
Core Design Contradiction:
Stability of the object's compositionVSMeasurement precision

Solution Approach 1:

The system performs noise reduction and surface artifact removal as preliminary processing steps. Image processing algorithms distinguish between code elements and surface irregularities such as specks, spots, and bumps, removing the latter while preserving the code structure for accurate decoding.

Inventive Principle:
Principle #10Preliminary action

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 model generates transformed images with improved code readability, reducing decoding errors and ensuring reliable code recognition on various objects.

Implementation Method 1

sensing means (camera 100) for optically detecting a code image reflected from the optically readable code

Methodology Applied
Scientific EffectReflection: Reflection

Data Source

PatentEP4453903B1Reading of optically readable codes
Publication Date: 2025.08.13 BAYER AG
  • EP4453903B1 patent drawingFigure 1
  • EP4453903B1 patent drawingFigure 2
  • EP4453903B1 patent drawingFigure 3~4

AI summary

The present invention relates to the technical field of the marking of articles by means of optically readable codes and of decoding the codes. Such a code is introduced into a surface of an article. The code is decoded on the basis of a transformed captured image of the code. The transformed captured image is generated from at least one captured image of the code using a machine learning model. The model is trained to generate a transformed captured image from at least one captured image and the readout of the optically readable code leads to less decoding errors than the readout of the code in the at least one captured image. The present invention relates to a method for training the machine learning model and to a method, a system and a computer program product for decoding a code using the trained machine learning model.