Vehicle Tyre Wall Recognition System
Automated Tyre Inspection with Computer Vision
Project Introduction
German automotive workshops face strict legal requirements regarding tyre safety. According to German law, tyres must be replaced when the tread depth falls below 1.6mm. However, manually measuring tread depth and reading tyre specifications from sidewalls is a time-consuming and error-prone process that creates bottlenecks in busy workshops.
The purpose of this project was to develop an automated solution that would help workshop owners quickly and accurately assess tyre condition and streamline the tyre replacement process. The system needed to accomplish three critical tasks: measure tread depth with precision, read tyre specifications from sidewalls, and provide customers with the best purchasing options available in the market.
The solution combined cutting-edge computer vision, photogrammetry, and e-commerce integration to create a complete end-to-end tyre inspection and ordering system deployed on edge computing infrastructure for real-time performance in workshop environments.
The Challenge
Creating an automated tyre inspection system presented several unique and complex technical challenges:
Precise 3D Depth Measurement
Measuring tyre tread depth with sub-millimeter accuracy to determine compliance with the 1.6mm legal requirement. Tyre treads have complex curved patterns requiring sophisticated 3D reconstruction techniques.
Reading Text on Black Rubber
Extracting embossed text and numbers from tyre sidewalls where characters are raised on a black rubber surface. Traditional OCR fails due to minimal contrast and the three-dimensional nature of embossed text.
Curved Surface Recognition
Tyre sidewalls are cylindrical surfaces, causing text distortion and perspective challenges. Standard 2D image processing techniques struggle with curved text reading.
Point Cloud Processing
Creating accurate 3D point clouds from multiple camera angles and reconstructing the tyre surface with sufficient detail to measure tread depth across the entire tyre profile.
Real-time Performance
Workshop environment requires fast processing to avoid bottlenecks. The system needed to deliver results within seconds, not minutes, while maintaining accuracy.
Edge Deployment
Workshop environments may have limited internet connectivity. The solution needed to run on local edge computing infrastructure with minimal external dependencies.
The Solution: Three Core Modules
Module 1: Tread Depth Measurement
3D photogrammetry with multi-camera setup
Hybrid 3D Depth Camera and Laser System
To achieve the sub-millimeter accuracy required for legal compliance, I implemented a hybrid approach combining 3D depth cameras with laser triangulation technology:
Hardware Setup:
- ✓Three synchronized 3D depth cameras (stereo vision setup) positioned at different angles
- ✓Laser line projector for precise triangulation measurement
- ✓High-resolution RGB cameras for texture capture
- ✓Controlled IR illumination for consistent depth capture on black rubber
Why hybrid approach? 3D cameras provide fast full-surface capture, while laser triangulation provides ultra-precise depth measurements at critical groove locations. Combining both gives speed AND accuracy.
Point Cloud Generation with Open-Source Software
Leveraging open-source computer vision libraries, the system generates and processes dense 3D point clouds representing the tyre surface:
Open-Source Software Stack:
Open3D
Point cloud processing, visualization, and 3D reconstruction pipeline
PCL (Point Cloud Library)
Advanced filtering, segmentation, and surface reconstruction algorithms
OpenCV + Contrib
Stereo matching, camera calibration, and feature detection
Point Cloud Creation Pipeline:
- • Camera calibration and rectification (OpenCV)
- • Stereo correspondence matching across views
- • Triangulation to generate 3D coordinates
- • Merge point clouds from laser and cameras
- • Statistical outlier removal (PCL)
- • Voxel downsampling for optimization (Open3D)
Depth Extraction Method:
- • Segment tread blocks vs. grooves using RANSAC
- • Fit reference plane to tread surface peaks
- • Identify groove bottom points (local minima)
- • Calculate perpendicular distance to reference plane
- • Extract multiple cross-section profiles
- • Report minimum depth (most worn area)
Depth Calculation: The system establishes a reference plane from the highest tread points, then measures the perpendicular distance to the lowest groove points. This gives the actual tread depth in millimeters, which is compared against the 1.6mm legal threshold.
Laser Triangulation for Precision Measurement
While 3D cameras provide overall surface geometry, laser triangulation delivers the precision needed for legal compliance:
How Laser Triangulation Works:
- • Project laser line across tyre tread grooves at known angle
- • Camera observes laser line from different angle (typically 30-45°)
- • Groove depth causes laser line displacement/deformation
- • Calculate depth using triangulation geometry (0.01-0.1mm accuracy)
- • Scan multiple lines across tyre surface for complete profile
Accuracy validation: Laser measurements validated the depth values extracted from 3D camera point clouds, ensuring sub-millimeter precision required for 1.6mm compliance verification.
3D Mesh Reconstruction Techniques
Beyond point clouds, mesh reconstruction creates continuous surface representation for visualization and analysis:
Reconstruction Algorithms:
- • Poisson surface reconstruction (Open3D)
- • Ball-pivoting algorithm for detailed areas
- • Marching cubes for volumetric conversion
Analysis & Extraction:
- • Mesh smoothing preserving groove edges
- • Curvature analysis for wear patterns
- • Profile extraction along groove paths
Result: The system accurately measures tread depth at multiple points across the tyre, identifying the minimum depth and comparing it against the 1.6mm legal threshold. If the tread depth falls below the requirement, the workshop owner is immediately alerted.
Module 2: Sidewall Text Recognition
Advanced image processing for embossed text extraction
The Sidewall Reading Challenge
Tyre sidewalls contain critical information including size specifications (e.g., 205/55R16), load index, speed rating, and manufacturer details. This information tells workshop owners exactly which tyre to order. However, reading this text is extremely challenging because:
- •Text is embossed (raised) on black rubber, creating minimal visual contrast
- •Traditional OCR fails because it relies on color contrast, not depth information
- •Sidewall curvature distorts text perspective
- •Lighting conditions in workshops create inconsistent shadows
Image Inversion Technique
The breakthrough came from using image inversion combined with controlled lighting. By carefully positioning lights and inverting the captured images, embossed text could be made visible:
- ✓Angled lighting to create shadows around embossed characters
- ✓Image inversion to convert raised areas into visible features
- ✓Contrast enhancement to maximize text visibility
- ✓Multiple lighting angles to handle different text orientations
OpenCV Contour Detection and Canny Edge Techniques
After image inversion, I applied advanced OpenCV techniques to extract text:
Canny Edge Detection:
- • Multi-stage edge detection algorithm to identify character boundaries
- • Adaptive thresholding for varying lighting conditions
- • Edge linking to connect broken character segments
Contour Detection:
- • Hierarchical contour detection to identify character shapes
- • Contour filtering based on size and aspect ratio
- • Grouping contours into text regions
- • Character segmentation for individual digit/letter extraction
Image Processing Pipeline:
- • Gaussian blur for noise reduction
- • Morphological operations (dilation/erosion) to enhance text
- • Perspective correction for curved sidewall text
- • Binary thresholding for clean text extraction
Text Recognition and Validation
Once text regions were extracted, I implemented recognition and validation:
- ▸Template matching for common tyre specification patterns
- ▸Regular expressions to validate tyre size format (e.g., 205/55R16)
- ▸Character recognition using trained models
- ▸Cross-validation with tyre specification databases
Module 3: Real-time Price Comparison
Finding the best tyre deals for customers
Multi-Marketplace Integration
Once the tyre specifications are extracted from the sidewall, the system automatically queries multiple online tyre marketplaces:
- ✓Real-time API integration with major German tyre retailers
- ✓Parallel queries to multiple marketplaces for speed
- ✓Price aggregation and comparison across vendors
- ✓Stock availability checking
- ✓Delivery time estimates
Customer Value Proposition
The system presents workshop customers with the best available offers, including pricing, delivery options, and vendor reputation, enabling informed purchasing decisions. Workshop owners can either order directly through the integrated system or use the price information to negotiate with their existing suppliers.
Edge Deployment with k3s
Lightweight Kubernetes for workshop environments
The entire system was deployed on local edge computing infrastructure using k3s (lightweight Kubernetes). This approach provided several advantages:
Performance Benefits:
- • No cloud latency - all processing happens locally
- • Real-time results for customers
- • No dependency on internet connectivity for core functionality
- • Reduced operational costs (no cloud compute charges)
Scalability & Management:
- • Container orchestration for easy updates
- • Microservices architecture for modularity
- • Automated recovery and health monitoring
- • Easy deployment across multiple workshop locations
Complete Technology Stack
Computer Vision
Point Cloud Libraries
3D Reconstruction
Hardware
Deep Learning
Infrastructure
Languages
Integration
Project Impact
Successfully delivered a complete automated tyre inspection system that combines 3D photogrammetry for tread depth measurement, advanced image processing for sidewall text recognition, and e-commerce integration for price comparison. The solution was deployed on edge infrastructure using k3s, enabling German workshops to quickly assess tyre condition, comply with legal requirements, and provide customers with the best purchasing options—all within seconds.