一种基于方向白游程的文本图像倾斜校正的改进算法

An Improved Algorithm of Skew Correction of Text Imagine Based on the White Run-Length of the Direction

摘要: 文本图像在扫描阶段的倾斜现象会对后续的图像分割以及光学字符识别(OCR)等处理造成极大的影响。本文提出了一种方向白游程算法的改进算法，通过对文本图像旋转从而对不同倾角的白游程进行扫描，然后统计每一个倾角上的白游程长度，最后找出最大白游程长度的倾角来调整文本图像。实验结果表明该算法能够准确并快速的校正文本图像。

Abstract: During the scanning phase, the tilt of text has a great impact on the following steps (image seg-mentation, OCR). This paper presents an improved arithmetic based on white run-length of direc-tion. By rotating the image, it can scan the white run-length of different angles. Then it calculates the white run-length’s length of every angle, and it finally finds out the angel of the longest white run-length. The experimental results show this algorithm can correct the text image accurately and quickly.