Handwritten Text Segmentation via Down-Sampling and Path Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Handwritten text line and word segmentation presents challenges due to variances in handwriting, with existing methods often balancing accuracy and efficiency inadequately.
Innovation Solution
A method involving horizontal down-sampling of input images using specific ratios to detect connected regions, forming line and word segments, with a path-score comparison approach to optimize word recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing line and word segmentation methods are used for handwritten text, then segmentation can be performed, but accuracy and efficiency cannot be adequately balanced due to handwriting variances
Solution Approach 1:
The patent divides the segmentation process into two distinct stages: line segmentation followed by word segmentation. In line segmentation, connected regions are grouped into line lists based on vertical position overlap. In word segmentation, connected regions within each line are further divided into word segments based on horizontal gaps. This multi-stage segmentation approach improves accuracy by handling different segmentation tasks separately while maintaining efficiency through systematic processing.
Solution Approach 2:
The patent uses down-sampling ratios as adjustable parameters to optimize the segmentation process. By converting images to lower resolution using specific down-sampling ratios, the computational complexity is reduced while preserving the essential structural information needed for accurate segmentation. This parameter change enables efficient processing without sacrificing segmentation quality.
2Productivity
If down-sampling is used to improve processing efficiency, then computational speed increases, but image detail may be lost affecting segmentation accuracy
Solution Approach 1:
The patent applies down-sampling with specific ratios to convert high-resolution handwritten text images into lower-resolution representations. This parameter change reduces the computational burden and increases processing speed while the method is designed to preserve the essential structural information needed for accurate line and word segmentation.
Solution Approach 2:
By performing segmentation on down-sampled images rather than original high-resolution images, the patent achieves efficient processing. The segmentation algorithms work on the simplified structures in down-sampled images to identify line and word boundaries, then apply these segmentation results to the original image structure.
Data Source
AI summary
A method for segmenting an image containing handwritten text into line segments and word segments. The image is horizontally down sampled at a first ratio. Connected regions in the down-sampled image are detected; horizontal neighboring ones are merged to form lines, to segment the original image into line images. Each line image is horizontally down sampled at a second ratio which is smaller than the first ratio. Connected regions in the down-sampled line image are detected to obtain potential word segmentation positions. A path is a way of dividing the line at some or all of the potential word segmentation positions into multiple path segments; for each of all possible paths, word recognition is applied to each path segment to calculate a word recognition score, and an average word recognition score for the path is calculated; the path with the highest score gives the final word segmentation.