多层曝光图像序列轮廓波域平滑算法  

Multilayer Exposure Image Sequence Contourlet Domain Smoothing Algorithm

作　　者：夏辉丽[1] 高静[1] 

机构地区：[1]中原工学院信息商务学院计算机科学系,郑州450007

出　　处：《科技通报》2015年第2期158-160,共3页Bulletin of Science and Technology

摘　　要：提出一种基于伪信息去除的多层曝光图像序列轮廓波域平滑算法,首先设计图像序列块曝光生成和轮廓边缘检测算法,基于非局部均值滤波进行图像特征提取,对图像进行连续性小波变换,得到图像的伪信息中心矩特征,求得多层曝光图像的最大后验估计算子,最终生成灰度直方图二进制均衡系数,进行特征匹配后得到了去除伪信息的输出图像,实现对轮廓波域平滑算法改进。仿真实验表明,采用该算法得到图像轮廓波域平滑处理后的PSNR最高,保留了图像大部分的细节信息,展示了较好的伪信息去除性能,图像纹理清晰可见,图像的视觉效果更好,实现准确的轮廓波域平滑处理。The multilayer exposure image sequence contourlet domain smoothing algorithm is proposed based on pseudo information removal, first exposure generating and profile edge detection algorithm, non local mean filter based on image feature extraction, continuous wavelet transform of image, get a pseudo Information Center moment feature of image, multi exposure images obtained maximum a posteriori estimation operator, finally generate the histogram equalization coefficient obtained binary image feature matching, output after removal of false information, thus improving the contourlet domain smoothing algorithm. Simulation results show that, the algorithm of image contours has the highest PSNR, it retains the detail information of the image most, it shows pseudo information removal performance good, texture is clear, it has better visual effect of image, the contourlet domain quasi smooth processing is improved.

关 键 词：伪信息 图像 轮廓波域 平滑 

分 类 号：TN317.4[电子电信—物理电子学]