基于Bandelet稀疏的移动机器人环境视觉纹理图像的压缩传感与重构  

Bandelet Sparsity Based Compressive Sensing and Reconstruction of Texture Images for Mobile Robot Environmental Visions

作　　者：马如远[1,2] 刘继忠[3] 金明亮[3] 柴国钟[1] 王光辉[3] 

机构地区：[1]浙江工业大学,杭州310000 [2]嘉兴学院,嘉兴314001 [3]南昌大学,南昌330031

出　　处：《中国机械工程》2014年第11期1535-1540,共6页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(61273282);江西省教育厅自然科学基金资助项目(GJJ12005)

摘　　要：压缩传感技术为移动机器人环境视觉的实时高效处理与传输提供了一种新的解决方法。结合Bandelet变换自适应跟踪图像正则方向的特点,进行了基于Bandelet稀疏和正交匹配追踪(OMP)算法的环境纹理图像压缩传感重构分析研究。结果表明:在较大观测值下,Bandelet稀疏重构与传统小波稀疏重构效果差别不大;在较小观测值下,传统sym8小波稀疏重构出现不稳定状态,出现块状信息缺失,不能有效重构,而Bandelet稀疏重构效果相对稳定;在给定观测值下,Bandelet稀疏重构的边缘细节表达能力优于sym8小波,说明Bandelet变换在压缩传感采样高压缩比下恢复重构具有有效性和稳定性。The occurrence of compressed sampling provided a novel efficient, real-time processing methodology for robot environmental vision mass informations. Combined with the Bandelet trans- form＇s advantage of adaptive tracking regularization direction of image, the research of the corn pressed sampling and reconstruction of environmental texture images was carried out herein, which was based on the Bandelet transform and orthogonal matching pursuit （OMP） algorithm. The experimental results show： the reconstruction results with Bandelet sparsity （BS） and wavelet sparsity （WS） are similar with a bigger measurement value; however, with a smaller measurement value, the reconstruction based on WS takes on an unstable state and sometimes the image can not be reconstructed effectively; with the same given measurement value, the reconstruction ability of edge details in the image based on BSis better than that of WS. The results illustrate that under higher compressive sensing compression ratio, the BS based environmental texture image reconstruction has a better ef- fectiveness and stability than that of WS based reconstruction.

关 键 词：机器人环境视觉 图像重构 压缩传感 Bandelet稀疏 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置] TP242.6[自动化与计算机技术—控制科学与工程]