Handwritten Text Recognition via Neural Network Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for recognizing handwritten text face challenges due to variations in shape, size, and consistency of handprinted characters, as well as ambiguity in individual character recognition and incorrect image segmentation.
Innovation Solution
A system and method using deep neural network models to recognize handwritten text by segmenting images into fragment and grapheme images, employing hypotheses and confidence scores to improve segmentation and recognition accuracy, and combining structural and neural network classifiers to enhance classification confidence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional algorithms are used to recognize handwritten text, then the system complexity is low, but the recognition accuracy deteriorates due to variations in shape, size, and consistency of handprinted characters
Solution Approach 1:
The patent segments the handwritten text recognition process into multiple components: image segmentation into character regions, generation of multiple segmentation hypotheses, individual character recognition, and hypothesis scoring. This segmentation allows the complex recognition task to be broken down into manageable steps, improving accuracy while organizing system complexity in a structured manner
Solution Approach 2:
The patent performs preliminary actions by generating multiple segmentation hypotheses before final recognition. The system creates several possible ways to segment the input image into characters, evaluates each hypothesis, and selects the best one. This preliminary hypothesis generation improves recognition accuracy by considering multiple possibilities before committing to a final interpretation
Solution Approach 3:
The patent implements feedback mechanisms through confidence scoring. Each segmentation hypothesis is evaluated and assigned a confidence score based on how well it explains the input image. This feedback loop allows the system to iteratively refine its segmentation and recognition decisions, improving overall accuracy
2Reliability
If multiple segmentation hypotheses are generated and evaluated, then the recognition accuracy improves, but the processing time increases
Solution Approach 1:
The patent applies partial action by generating a limited number of segmentation hypotheses rather than exhaustively exploring all possible segmentations. The system generates multiple hypotheses but processes only the most promising ones, evaluating them with confidence scores to select the best segmentation without exhaustive computation
Solution Approach 2:
The patent replaces traditional mechanical segmentation approaches with neural network-based hypothesis evaluation. Instead of using rigid rules to determine segmentation, the system uses trained neural networks to evaluate multiple hypotheses and select the best one, significantly improving processing efficiency while maintaining accuracy
3Measurement precision
If deep neural network models are used for character recognition, then the classification accuracy improves, but the computational resources required increase
Solution Approach 1:
The patent segments the computational workload by processing each character independently after segmentation. Instead of analyzing the entire text image as one complex pattern, the system divides it into individual character regions and processes each with a neural network, reducing the computational burden per character while maintaining high classification accuracy
Solution Approach 2:
The patent performs preliminary image processing and segmentation before neural network classification. By pre-processing the input image to extract and segment character regions, the system reduces the complexity of the classification task, allowing the neural network to focus on recognizing individual characters rather than handling the entire image, thus reducing computational resources
Data Source
AI summary
In one embodiment, a system receives an image depicting a line of text. The system segments the image into two or more fragment images. For each of the two or more fragment images, the system determines a first hypothesis to segment the fragment image into a first plurality of grapheme images and a first fragmentation confidence score. The system determines a second hypothesis to segment the fragment image into a second plurality of grapheme images and a second fragmentation confidence score. The system determines that the first fragmentation confidence score is greater than the second fragmentation confidence score. The system translates the first plurality of grapheme images defined by the first hypothesis to symbols. The system assembles the symbols of each fragment image to derive the line of text.