摘要
在光学字符识别(OCR)图像扫描输入过程中,图像或多或少地会出现某种程度的倾斜,这往往给字符分割和识别造成很大的影响和难度。因此,文档图像的倾斜矫正是很重要的环节。Hough变换可以用来将图像边缘连接起来得到边界线,该变换受噪声和边界线间断的影响较小,而广泛用于各种类型文档图像的倾斜矫正。但Hough变换计算复杂度较高,本文对此作了一些改进,从而大大减少了计算量。
In the process of the Optical Character Recognition scanning and inputting, The image meets to skew in some degree more or less, sometimes this makes the character division and recognition prodigiously influent and difficult. So the document image skew correction is very important. Hough transformation may use to connect the image edge in order to gain the borderline. The transformation has the virtue that the impaction of the voice and the borderline breaking is less, so it is widely used of a kind of document image skew correction. But the computing complexity of Hough transformation is higher, the article is to improve the computing complexity, greatly reduces the computing.
出处
《科技信息》
2007年第19期73-74,共2页
Science & Technology Information
