雾天车辆超分辨率视频图像清晰度识别仿真  

Simulation of Resolution Recognition for Super Resolution Video Images in Foggy Vehicles

作　　者：汤嘉立[1] 杜卓明 TANG Jia - li DU Zhuo - ruing(Jiangsu University School of Mechanical Engineering,Jiangsu Zhenjiang 212013, China Nanjing Normal University School of Mathematical Seienees,Jiangsu Zhenjiang 210023 ,China)

机构地区：[1]江苏大学机械工程学院,江苏镇江212013 [2]南京师范大学数学科学学院,江苏南京210023

出　　处：《计算机仿真》2017年第10期155-158,共4页Computer Simulation

基　　金：江苏省高校自然科学研究面上项目(16KJB520010);常州市科技支撑计划(社会发展)(CE20165028)

摘　　要：对雾天车辆超分辨视频图像清晰度进行识别,能够更好的解决模糊车辆图像失真问题。对超分辨率视频图像的识别,需要先获取图像边缘细节,进而调整低通子带系数,完成图像的清晰度识别。传统方法先检测图像模糊边缘,获得各个不同区域中像素点权值,但忽略了调整低通子带系数,导致图像识别精度偏低。提出基于分数阶的雾天车辆超分辨率图像清晰度识别方法。该方法先对雾天车辆超分辨率图像中的各个像素点进行分数阶积分运算来处理噪声,获得不同强度纹理点相应的微分阶次,通过对图像的Contourlet分解处理,进行图像边缘细节识别,低通子带系数则根据灰度级变换和局部均值的增益函数进行调整,由此完成雾天车辆超分辨率图像清晰度识别。实验结果表明,所提方法能够有效提升雾天车辆超分辨率图像的视觉效果,且能保留较多的图像边缘细节信息。This research focuses on definition recognition method for video image with super resolution of vehicle in foggy weather. Our research carried out integral operation of fractional order for each pixel point in the image and obtained derivative order corresponding to texture point with different intensity. Then we carried out detail recognition of image edge via contourlet decomposition processing on the images. The coefficient of low - pass sub band according was adjusted to gain function of gray level conversion and local mean value. Experimental results show that the meth- od can improve visual effect of the image with super resolution and reserve more detail of image edges.

关 键 词：雾天车辆 超分辨率图像 清晰度识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]