基于压缩感知的图像特征提取算法研究.doc

Foshan University 本科生毕业设计（论文） 基于压缩感知的图像特征提取算法研究 学 院： 专 业： 学 号： 学生姓名： 指导教师： （职称） 二〇 年 月 摘 要 在过去的几十年里，人们获取数据的能力不断提高，需要处理的数据量也越来越大，因此信号的带宽也越来越大，所以对信号处理的速度和采样速率的要求也随之提高。众所周知，奈奎斯特采样定理要求采样率不得低于信号带宽的两倍，这对目前的信号处理能力提出了巨大的挑战。 压缩感知理论近年来已经成为了处理稀疏信号或可压缩信号的重要手段。传统的奈奎斯特采样定理在当前的很多实际应用中（例如：超宽带通信，超宽带信号处理，核磁共振等）要求有非常高的采样率，伴随着高采样率，随之而来的就是高成本，甚至一些过高的采样率无法从技术上实现。而压缩感知的出现给解决这些问题带来了希望，在压缩感知理论中，信号的采样速率不取决于信号的带宽，而是决定于信息在信号中的位置和内容，在采样的同时即对信号数据进行适当压缩。 压缩感知(Compressed Sensing, CS)理论是一个充分利用信号的稀疏性(或可压缩性)的全新信号采集、编解码理论。该理论指出，只要信号是稀疏的或可压缩的(即在某个变换域上是稀疏的)，那么就可以用一个与变换基不相关的采样矩阵将变换所得的高维信号投影到一个低维空间上，然后通过求解一个优化问题，从这些少量的投影中以高概率重构出原信号。 关键词：图像技术、压缩感知、图像特征、GUI、matlab Image Feature Extraction Algorithm Based on Compressed Sensing XINGMING Abstract In the past few decades, people the ability to obtain data continues to improve, the amount of data to be processed is also growing, so the signal bandwidth is also growing, so the speed and sampling rate signal processing requirements are over of the increase. It is well known Nyquist sampling theorem requires the sampling rate of not less than twice the signal bandwidth, which is currently the signal processing capability presented a huge challenge. Compressed sensing theory in recent years has become an important means of dealing with sparse signal or compressible signal. The traditional Nyquist sampling theorem in the current many practical applications (for example: ultra-wideband communications, ultra-wideband signal processing, nuclear magnetic resonance, etc.) require a very high sampling rate, along with the high sampling rate, followed by It is costly, and even some high sampling rate can not be achieved technically. The compressed sensing appears to solve these pro