彩色序列车牌图像超分辨率复原技术研究  

Research on the Image Restoration Technology with Super-resolution for Color Sequential Vehicle License Plate

作　　者：宁志刚[1] 杨恒[1] 杨玲[1] 程辉[1] 赖华超 

机构地区：[1]南华大学电气工程学院 [2]南山大学船山学院,湖南衡阳421001

出　　处：《自动化仪表》2013年第11期6-8,12,共4页Process Automation Instrumentation

基　　金：湖南省自然科学基金资助项目(编号:11JJ6021);湖南省大学生研究性学习和创新性实验计划基金资助项目(编号:[2012]402);南华大学博士启动基金资助项目(编号:2007XQD23)

摘　　要：为恢复高分辨率的彩色车牌图像,对序列车牌图像的模糊函数辨识、运动估计和图像重建进行了研究。采用支持向量机对模糊车牌图像的模糊类型进行分类,对模糊车牌的模糊参数进行辨识。对比分析了在不同空间和不同块匹配准则条件下块匹配算法的性能,得出了彩色车牌图像最佳运动估计方法,并比较了不同超分辨率重建算法的性能。试验数据表明,模糊函数辨识方法、运动估计方法和图像重建方法有利于图像超分辨率复原。The fuzzy function recognition,motion estimation and image reconstruction of the sequential vehicle license plate image are researched for restoring the image of color vehicle license plate with super-resolution.Blur types of blurred vehicle license plate images are classified by using support vector machine （SVM）,and recognition of the blur parameters of the blured license plate is conducted.The performances of the block matching algorithm under different space and different block matching criteria are compared,thus the optimal motion estimation method of the color license plate image is obtained,and the performance of the reconstruction algorithm with different resolution are analyzed and compared.The experimental data indicate that the fuzzy function recognition method,motion estimation method and image reconstruction method are beneficial to image restoration with super-resolution.

关 键 词：超分辨率复原 模糊函数辨识 运动估计 图像重建 图像传感器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]