摘要
均匀分块压缩感知对图像信号进行压缩采样,无法有效地分离出重要区域和背景区域.为此,给出一种基于图像内容的自适应分块算法,以图像内相邻像素间的灰度差值作为块大小分割的依据,利用四叉树算法进行图像自适应块大小的划分;并将分块结果根据相邻像素DCT系数的差值大小分成快速变化块、缓慢变化块和过渡块3类,适时分配相应采样率.实验结果表明:给出的算法对仿真实验选取的图像重构质量高于均匀分块方法1~3dB,且重构时间减少20~40ms;在重构质量近似的情况下,重构时间比基于图像块像素方差的块分类方法减少40~60ms.
The uniform block compressed sensing is used to compress and sample image signals,but it cannot separate the important region from background for image signals effectively.This paper presents an adaptive blocking method based on image content,which utilizes the gray difference in image adjacent pixels as a criteria for block size division.The quad-tree algorithm is used to divide the image into adaptive block size.According to the difference of DCT coefficients of neighboring pixels,the result of partitioning is divided into three categories：rapid change blocks,slowly change blocks and transition blocks,and appropriate sampling rates are distributed to them.Simulation results show that the image reconstruction quality of the proposed method is higher 1-3 dB than the uniform blocking method,and the reconstruction time is reduced 20-40 ms;with the approximate reconstruction quality,the reconstruction time is less about 40-60 ms than that block classification method based on the mean square difference of pixels grey.
作者
李如春
程云霄
李林
常丽萍
LI Ruchun, CHENG Yunxiao, LI Lin, CHANG Liping(College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, China)
出处
《浙江工业大学学报》
CAS
北大核心
2018年第4期392-395,406,共5页
Journal of Zhejiang University of Technology
基金
国家自然科学基金资助项目(61304124)
关键词
压缩感知
图像压缩
自适应分块
采样率设定
仿真测试
compressive sensing
image compressing
adaptive blocking
sampling rate set
simulation test
