Document Text Line Segmentation for Handwriting and Skewed Printed Text
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing document image processing methods fail to effectively segment text lines containing both printed and handwritten text, especially when the text is skewed or has narrow line spacing, as they struggle to separate lines with intersecting or overlapping characters and significant distortion.
Innovation Solution
A method involving connected component analysis, cluster analysis using the k-means++ algorithm, and X-Y cut projection to categorize and separate text lines, removing extra spaces and handling large objects like handwriting by designating object boards for each text line, thereby isolating printed text from handwritten elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If X-Y cut projection method is used to segment document images into text lines, then the method is simple and widely applicable, but it fails to separate text lines when handwriting intersects printed characters or when text lines have significant skew and narrow line spacing
Solution Approach 1:
The patent segments the document image into multiple connected components based on spatial proximity and visual characteristics. Each connected component is treated as an independent unit for further processing, allowing the system to handle complex interactions between handwriting and printed text without requiring complete line separation in the original image.
Solution Approach 2:
The patent assigns different processing treatments to different types of connected components based on their local characteristics. By categorizing components as handwriting, printed text, or other elements and applying line spacing calculation only to printed text components, the system adapts its behavior to local conditions rather than applying a uniform approach throughout the document.
2Productivity
If traditional projection methods are used for text line segmentation, then the processing is fast and straightforward, but they cannot handle skewed text lines where start/end positions coincide with adjacent lines
Solution Approach 1:
The patent performs preliminary connected component analysis and categorization before attempting line segmentation. By pre-identifying which components are printed text and calculating their line spaces in advance, the system prepares the data structure needed for accurate skewed line detection without requiring complex real-time calculations during the segmentation process.
Solution Approach 2:
The patent transitions from traditional single-dimensional projection-based segmentation to a multi-dimensional approach that considers spatial coordinates, component types, and calculated line spaces simultaneously. This allows the system to detect skewed lines by analyzing the two-dimensional arrangement of connected components rather than relying solely on vertical projections.
Data Source
AI summary
A text line segmentation method for a document image containing printed text and handwriting, or document image containing skewed lines or printed text. Connected component (CC) are obtained for the document, and their bounding boxes and centroids are calculated. The CCs are categorized into three categories based on bounding box sizes: small objects, regular text objects, and large objects involving handwriting. The centroids of regular text objects are used in a cluster analysis to find the vertical centers of the N text lines. Then, each CC is classified into one of the N lines based on the vertical distance between its centroid and the vertical centers of text lines, and copied into to a corresponding object board. Extra spaces are removed from the object boards to obtain the line segments. The large object involving handwriting will be classified into one of the lines but absent from other lines.