Blur-robust symbol recognition using transform-based feature extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine-readable symbol readers face difficulties in accurately decoding symbols from highly blurred or distorted images due to noise and image quality issues, making it challenging to read and process machine-readable symbols effectively.
Innovation Solution
A processor-based system that transforms the image signal representation of machine-readable symbols using blur-robust orthonormal bases such as the Discrete Sine Transform, Discrete Cosine Transform, or Chebyshev Transform, followed by classification using trained classifiers like random forest or Mahalanobis classifiers, to enhance the recognition of blurred symbols.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing techniques are used to read machine-readable symbols, then the system is simple and easy to operate, but the recognition accuracy deteriorates when symbols are highly blurred or distorted
Solution Approach 1:
The patent segments the symbol recognition process into multiple specialized stages: image acquisition, preprocessing (noise reduction, distortion correction), feature extraction (identifying symbol elements, transitions, and patterns), and decoding. This segmentation allows each stage to be optimized independently, improving overall accuracy while maintaining manageable system complexity through modular design.
Solution Approach 2:
The patent applies preliminary actions by performing preprocessing operations before main recognition. This includes correcting geometric distortions, reducing noise, and enhancing contrast before the actual symbol decoding occurs. These preliminary steps prepare the degraded image data, making subsequent recognition more accurate even when symbols are highly blurred or distorted.
2Measurement precision
If advanced transform methods and trained classifiers are used to recognize blurred symbols, then recognition accuracy improves, but computational complexity and processing time increase
Solution Approach 1:
The patent applies partial action by selecting and applying only the necessary preprocessing and transform operations based on the specific image quality and symbol type. Rather than always applying the full suite of advanced techniques, the system adapts the level of processing, using more aggressive methods only when needed for highly degraded images, thus balancing accuracy with processing time.
Solution Approach 2:
The patent utilizes parameter changes by adjusting transform parameters and classifier thresholds based on image quality metrics. For example, the choice of transform method (DCT, DWT, Fourier) and the number of transform coefficients used are dynamically adjusted based on the measured degradation level, allowing the system to optimize between accuracy and processing speed for different input conditions.
3Productivity
If flood illumination is used to illuminate the entire machine-readable symbol, then the entire symbol is captured simultaneously, but image quality deteriorates due to noise and blur
Solution Approach 1:
The patent substitutes mechanical/optical improvements with digital signal processing. Rather than attempting to prevent blur and noise through perfect optical mechanics (illumination and focusing), the system captures the degraded image and uses digital transforms and classifiers to reconstruct the symbol information, replacing the need for perfect physical capture conditions.
Solution Approach 2:
The patent converts the harmful effects of noise and blur into beneficial processing opportunities. By intentionally modeling various degradation patterns during training, the classifier learns to recognize symbols even when distorted. The harmful noise and blur become part of the training data, enabling the system to robustly handle real-world degraded conditions.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for robust recognition of machine-readable symbols from highly blurred or distorted images. An image signal representation of a machine-readable symbol element is transformed into a different space using one or more transform operations, which moves an n-dimensional vector of measured light intensities into another n-dimensional space. The types of transform operations may include blur robust orthonormal bases, such as the Discrete Sine Transform, the Discrete Cosine Transform, the Chebyshev Transform, and the Lagrange Transform. A trained classifier (e.g., an artificial intelligence machine learning algorithm) may be used to classify the transformed signal in the transformed space. The types of trainable classifiers that may be used include random forest classifiers, Mahalanobis classifiers, support vector machines, and classification or regression trees.