摘要
针对传统二维层面的倾斜校正算法难以处理扭曲文本的倾斜,提出一种基于点云数据的扭曲文本图像倾斜校正方法。首先,通过三维扫描仪采集文本的数据信息,重建文本的三维点云模型,接着通过检测文本在空间中的位置,定义为空间倾斜与平面倾斜,最后实现文本的自动倾斜校正。实验结果表明,基于点云数据的倾斜校正算法在处理扭曲文本的倾斜校正正确率达到95%以上。
In view of that traditional two-dimensional tilt correction algorithm is difficult to deal with the tilts of distorted text, we put forward a point cloud data-based tilt correction method for distortion text and image. First, it collects data information of text by a 3D scanner,and reconstructs the 3D point cloud model of text. Then according to the examined text position in space, it defines the tilt as either the space tilt or plane tilt. Finally it realises automatic tilt correction of the text. Experimental results show that the correction rate of tilt correction algorithm based on point cloud data reaches over 9 5 % in dealing with the distorted text.
作者
童立靖
郑俊朝
Tong Lijing;Zheng Junchao(College of Computer, North China University of Technology, Beijing 100144 , China)
出处
《计算机应用与软件》
CSCD
2016年第6期176-179,184,共5页
Computer Applications and Software
基金
国家自然科学基金项目(61371142)
北京市教委专项(PXM2013_014212_000026)
作者简介
童立靖,副教授，主研领域：图像处理，智能控制，系统辨识等。;郑俊朝，硕士生。
