Document Skew Detection Using Connected Components Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for detecting document skew in scanned images are inefficient and lack the capability to effectively straighten documents, which is crucial for improving compression rates, visualization, and Optical Character Recognition (OCR) accuracy.

Innovation Solution

The method employs connected components analysis to detect skew by grouping pixels based on connectivity, identifying blobs, sorting them by location, and calculating the skew angle using a computer system with a blob identifying module and a skew detection module, enabling fast and efficient skew detection with a small memory footprint.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If prior art methods (edge detection or line segmentation) are used for skew detection, then skew angle can be computed, but the processing speed is insufficient for high-volume scanning operations (20-30 pages per second)

Engineering Contradiction:

Improveskew detection speedVSAvoidtime spent on skew detection

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent segments the scanned document into connected components (blobs) based on pixel connectivity and intensity values. By dividing the image into discrete components rather than analyzing every pixel or scan line, the system achieves faster processing suitable for high-volume scanning operations at 20-30 pages per second

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent replaces traditional mechanical edge detection algorithms with connected components analysis. Instead of comparing scan lines with background pixels or detecting edges through neighborhood analysis, the system uses blob identification based on pixel connectivity, which is computationally more efficient and enables faster skew detection

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If traditional skew detection methods are used, then skew angle can be determined, but memory consumption is excessive

Engineering Contradiction:

Improveskew angle detection accuracyVSAvoidmemory footprint

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential information needed for skew detection by identifying connected components and their boundary points. Instead of storing and processing entire scan lines or neighborhood pixel data, the system extracts minimal sufficient data (blob locations and boundaries) to compute the skew angle, significantly reducing memory footprint while maintaining detection accuracy

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS8290302B2Method and system for skew detection of a scanned document using connected components analysis

Publication Date: 2012.10.16 XEROX CORP

AI summary

A method and system for skew detection is provided using connected components analysis. The methodology includes extracting connected components corresponding to the image and analyzing the image based on said connected components for determining skew of the image.