Tyre Sidewall Imaging Using HOG-CNN for Low Contrast Code Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for reading embossed and engraved markings on vehicle tires, such as OCR technologies, face challenges with low contrast and color difference in outdoor conditions, leading to inefficiencies and inaccuracies, especially when trying to automate the process for moving vehicles or in varying weather conditions.
Innovation Solution
A method combining Histogram of Oriented Gradients (HOG) features with convolutional neural network (CNN) layers to improve the efficiency and accuracy of identifying regions of interest on low contrast images, reducing the need for complex deep CNN architectures and resource-intensive hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional OCR techniques are used on tyre sidewall images, then the system can read tyre codes, but the accuracy is low due to low contrast and color difference in outdoor conditions
Solution Approach 1:
The patent applies color space transformation (converting images to different color spaces like LAB, YCrCb, or HSV) to enhance the contrast between text and background. By transforming the color representation, the system can better distinguish the embossed text from the tyre sidewall surface, overcoming the low contrast problem in outdoor conditions.
Solution Approach 2:
The patent modifies image processing parameters by applying various filters (Gaussian blur, adaptive histogram equalization) and adjusting processing thresholds. These parameter changes enhance the visibility of text features and improve the robustness of character recognition under varying lighting and weather conditions.
2Measurement precision
If deep convolutional neural networks are used to improve text recognition accuracy, then the system can handle noisy environments, but the computational overhead and hardware requirements increase significantly
Solution Approach 1:
The patent divides the text recognition task into multiple stages: initial image preprocessing, candidate character detection, sequence assembly, and verification. This segmentation allows each stage to be optimized independently, reducing the overall computational burden while maintaining high accuracy through focused processing at each step.
Solution Approach 2:
The patent implements a multi-pass recognition approach where the system performs recognition on subsets of characters or regions of interest, rather than processing the entire image uniformly. This partial action strategy reduces computational overhead by focusing resources on the most informative parts of the image while still achieving accurate text recognition.
3Measurement precision
If 3D scanner based systems or handheld laser devices are used, then the system can read tyre codes accurately, but the manufacturing cost is high and the system is challenging to calibrate
Solution Approach 1:
The patent replaces complex mechanical 3D scanning systems with a simplified optical imaging system using standard cameras. By substituting mechanical depth-sensing hardware with computational image processing techniques, the system achieves comparable accuracy while dramatically reducing manufacturing costs and calibration requirements.
Solution Approach 2:
The patent creates a digital representation (copy) of the 3D surface information through 2D image capture and processing. Instead of directly measuring 3D geometry with complex scanners, the system captures optical images and extracts text information through image processing, effectively copying the necessary information in a simpler form.
Data Source
AI summary
A computer implemented method for generating a region of interest on a digital image of a sidewall of a tyre, the sidewall having one or more embossed and/or engraved markings, is provided. The method comprises generating a histogram of oriented gradients feature map of the digital image, inputting the histogram of oriented gradients feature map into a trained convolutional neural network, wherein said trained convolutional neural network is configured to output a first probability based on the input histogram of oriented gradients feature map that a region of pixels of the digital image contains the embossed and/or engraved markings, and if the first probability is at or above a first predetermined threshold, accepting said region of pixels as said region of interest.