Text Image Correction Using Row-Cumulant Frequency Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Skewed text in photographed document images due to human or non-human causes affects reading and subsequent processing, including printing and automatic verification.
Innovation Solution
A method involving preprocessing to binary images, rotating them in predetermined steps, calculating row cumulants, and using the maximum frequency to determine a compensation angle for correction, along with a text image correction apparatus for accurate skew angle estimation and correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional skew angle estimation methods are used, then the correction process is simple, but the estimation accuracy is low leading to poor correction效果
Solution Approach 1:
The patent segments the image processing into distinct stages: binary image generation, row cumulant calculation, frequency spectrum analysis, and peak detection. Each stage processes specific features independently, improving measurement precision through systematic analysis while keeping each individual processing step relatively simple and manageable.
Solution Approach 2:
The patent transforms the skew angle estimation problem from direct spatial analysis to frequency domain analysis by calculating row cumulants and performing spectral analysis. This dimensionality change from spatial domain to frequency domain enables more accurate angle detection through frequency peak identification, resolving the contradiction between accuracy and complexity.
2Reliability
If frequency spectrum analysis with sliding window smoothing is applied, then the skew angle estimation robustness is enhanced, but the computational complexity increases
Solution Approach 1:
The patent applies sliding window smoothing filtering to the row cumulant sequence before performing frequency spectrum analysis. This preliminary action removes noise and enhances the reliability of subsequent peak detection, ensuring more robust skew angle estimation while preparing the data in advance to avoid repeated processing.
Solution Approach 2:
The patent replaces direct geometric analysis methods with frequency domain analysis using spectral methods. This substitution transforms the mechanical image processing approach into a mathematical frequency analysis approach, enhancing robustness against noise and variations while providing a systematic framework for angle estimation.
3Measurement precision
If multiple processing steps including binary conversion and cumulant calculation are used, then the correction accuracy is improved, but the processing time increases
Solution Approach 1:
The patent divides the correction process into segmented, independent modules: binary image generation, row cumulant calculation, spectral analysis, and peak detection. Each module processes specific aspects of the image, allowing for optimized computation in each segment while maintaining overall high accuracy through the combination of all segments.
Data Source
AI summary
A text image correction method and a corresponding text image correction apparatus. Frequency information of a row-direction cumulative curve used by the method is sensitive to an error between a compensation angle for a tilt angle and a real tilt angle, and the method thus has good robustness. The method can accurately estimate the compensation angle for a tilt angle and correct a tilted text image. The method and apparatus can be applied to scenarios such as image pre-processing, automatic compensation for angles of scanned text images, automatic compensation for tilt angles of mobile phone photos.