首页 | 官方网站   微博 | 高级检索  
     

非均匀介质退化图像快速仿真模型的建立
引用本文:崔光茫,赵巨峰,冯华君,徐之海,李奇,陈跃庭.非均匀介质退化图像快速仿真模型的建立[J].浙江大学学报(自然科学版 ),2014,48(2):303-311.
作者姓名:崔光茫  赵巨峰  冯华君  徐之海  李奇  陈跃庭
作者单位:浙江大学 现代光学仪器重点实验室,浙江 杭州 310027
基金项目:国家自然科学基金资助项目(61275021、61178064).
摘    要:针对非均匀介质成像仿真速度较慢的问题,提出一种获取近似退化图像的快速仿真模型.按视场将图像分为若干子块,计算得到各子块中心位置的点扩散函数矩阵,建立与像素大小对应的成像退化矩阵.近似认为各子块内成像满足等晕条件,分块卷积快速得到各子块退化图像,并采用少量重叠的方法实现整个视场的无缝拼接,加入畸变信息得到最终的退化图像.建立仿真成像系统,做了多组对比实验,并对光追法得到的精确退化图像和该模型得到的近似退化图像进行相似度的比较.结果表明,在一定相似度的要求下,对于512×512像素的图像实现毫秒级的处理速度.


Construction of fast simulation model for degraded image by inhomogeneous medium
CUI Guang-mang,ZHAO Ju-feng,FENG Hua-jun,XU Zhi-hai,LI Qi,CHEN Yue-ting.Construction of fast simulation model for degraded image by inhomogeneous medium[J].Journal of Zhejiang University(Engineering Science),2014,48(2):303-311.
Authors:CUI Guang-mang  ZHAO Ju-feng  FENG Hua-jun  XU Zhi-hai  LI Qi  CHEN Yue-ting
Abstract:A fast simulation model for getting approximate degraded images was proposed to raise the low speed of the simulation of images degraded by inhomogeneous medium. The image was divided into several sub-blocks according to the field view. Point spread function matrix of the center in each sub-block was computed to construct the degraded matrix, which was corresponding to the pixel size. Each sub-blocks were approximately regarded as isoplanatic regions and the sub-block degrade images were obtained by convolution calculation. Seamless splicing of the entire field of view images were achieved by the way of a little overlap of the adjacent blocks. The distortion was also added to obtain the final degraded image. With an imaging simulation system built, multiple experiments were done to compare the similarity between the degraded images obtaining by proposed model and ray tracing method. Experiment results show that with the requirement of a certain similarity, the time of simulating an image with the size of 512×512 is about several milliseconds.
Keywords:
点击此处可从《浙江大学学报(自然科学版 )》浏览原始摘要信息
点击此处可从《浙江大学学报(自然科学版 )》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号