基于L_0正则化的车牌图像去模糊  

License plate image deblurring based on L_0 regularization

作　　者：王超[1] 陈亚运[1] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《电子设计工程》2016年第5期55-57,61,共4页Electronic Design Engineering

摘　　要：随着视频监控技术的发展、监控摄像头的普及和人们对安全信息需求的提升,监控中图像去模糊技术得到迫切需求。基于图像强度和梯度的L0正则化去模糊算法能有效恢复模糊的文本图像,但是由于车牌图像的灰度值为零的点较少,导致该方法对模糊的车牌图像复原结果有瑕疵。本文根据的灰度直方图的特性,提出一种改进方法,并与两种方法对比,本文方法取得较好的结果。With the development of video surveillance technology, surveillance cameras has become more and more popular. Thus the requirement for protecting individual information and monitoring the image deblurring technology is increasingly demanded. We can restore blurred text image by L0-Regularized Intensity and Gradient Prior. However, this method do not work efficiently when it comes to the restoration of license plate images, since it has less pixel witch the intensity is zero. According to the characteristics of histogram of pixel intensities, the paper will put forward an improved method, compare it with two methods and get better results.

关 键 词：图像去模糊 L0正则化 灰度直方图 梯度直方图 

分 类 号：TN911.73[电子电信—通信与信息系统]