摘要
针对复杂环境因素影响下的视觉表格图像,提出了一种有效的图像分割方法。首先利用图像的闭运算,估计出图像的背景变化,进而利用定义的图像对比度实现了图像的背景补偿。然后利用背景估计图像的均值作为补偿图像的全局分割阈值,成功实现了表格图像分割。实验仿真表明,本文方法的分割结果体现了较好的视觉效果,而且与基于补偿图像的OSTU方法和基于局部阈值的Niblack方法相比,它在图像单元格区域的过分割与欠分割之间实现了较好的均衡。
For the visual form images under complex environments,an effective image segmentation method was pres-ented.Firstly,the changes of image background were estimated by closed operation of images,and then the image background compensation was realized by defined image contrast.Then taking the mean value of the image background estimation as a global segmentation threshold of the compensation image,the form image segmentation was successfully achieved.Experimental results show that the segmentation method can obtain a good visual effect,and compared with the OSTU method based on compensation image and Niblack method based on local threshold,it can achieve a good balance between over-segmentation and under-segmentation of the image cell regions.
出处
《激光与红外》
CAS
CSCD
北大核心
2017年第2期234-238,共5页
Laser & Infrared
基金
国家自然科学基金项目(No.61463052
61463049)
云南省科技创新强省计划(No.2014AB016)资助
关键词
图像分割
背景估计
图像补偿
阈值估计
image segmentation
background estimation
image compensation
threshold estimation
