基于组的非局部稀疏表示图像去噪算法

Group-based Nonlocal Sparse Representation Image Denoising Algorithm

提出一种基于Android智能手机平台的定位系统。该系统利用智能手机自带惯性测量单元(陀螺仪、加速度计以及磁力计)的数据经过行人航迹推算(PDR)算法处理,在封装算法时采用交叉编译环境Cygwin和Android NDK来编译算法c程序,以得到可被安卓系统调用的so库文件,来获取室内行人的实时步长与方向角数据,并通过手机WIFI模块将实时数据传送到服务器,实现室内行人实时定位监控。测试软件结果表明:所设计手机软件安全性能得到提高,并且运行良好,能够很好地达到实时监测室内行人的目的。

Focus on the phenomenon that artifacts are made by the existing denoising algorithms, combining nonloeal sparse repre- sentation with group -based sparse representation, a new denoising model is proposed. In the new model, a group is a basic unit for sparse representation, firstly the new model trains the dictionary groups from clear natural images, and then, obtains the nonlo- cal estimation by low rank approach, finally, the iterative shrinkage algorithm is used to solve the proposed model. Experiments show that not only the PSNR and SSIM of the proposed model is higher than others＇, but also the artifacts are reduced. This mod- el can protect important features and improve the visual effects of the image.