《压缩传感理论的研究与实现》由会员分享，可在线阅读，更多相关《压缩传感理论的研究与实现（41页珍藏版）》请在金锄头文库上搜索。
1、北京邮电大学本科毕业设计（论文）传感压缩理论的研究与实现摘要当今信息社会需要存储压缩的数据量不断增长，而信息存储的硬件空间增长速度有限，数据压缩图像压缩也就日益重要。根据奈奎斯特采样定理，传统的采样频率不得低于信号最高频率的 2 倍。但是对于数字图像等信息，传统的采样会导致海量采样，意味着还需要对海量的数据进行后续处理，大大增加了存储和传输的代价。本文对传感压缩理论进行研究，传感压缩采用较低的采样频率对数字信号进行采样并压缩，将采样和压缩合并为一步而进行，节省了数字信号采样开销、存储开销和传输开销。传感压缩采用非自适应线性投影来保持信号的原始结果，能通过数值最优化问题确定重构原始信号，而且重构
2、信号能保证较高精度。传感压缩不仅可以用在一般数据信号的压缩上，也可以用在图像信号的压缩上。本文利用 matlab 仿真了传感压缩对图像压缩的存储与重构，将传感压缩算法与 JPEG 压缩算法做了实验对比，进一步分析传感压缩的应用前景。在传感压缩理论和多种编码压缩算法的研究基础上，本文对二维图像压缩的新方法进行探索，利用 matlab 进行实验探索，进一步分析新算法改进方案。文内称此图像压缩的新方法为相似块删除算法。关键词：传感压缩、图像压缩、采样、重构信号北京邮电大学本科毕业设计（论文）ABSTRACT Todays information society need to store the g
3、rowing amount of data compression, and information storage growth rate of the hardware space is limited, data compression, image compression will become increasingly important. According to Nyquist sampling theorem, the traditional sampling frequency not less than 2 times the maximum frequency of th
4、e signal. But for digital images and other information, the traditional sampling will lead to mass sampling, means that vast amounts of data also need to follow-up treatment, greatly increasing the storage and transmission costs.In this paper, the theory of compressed sensing, sensing a lower sampli
5、ng frequency of compressed digital signal sampling and compression, and compression into the sample carried a step to save the digital signal sampling overhead, storage overheadand transmission overhead. Non-adaptive compressed sensing to maintain a linear projection of the original signal results d
6、etermined through numerical optimization problem to reconstruct the original signal and reconstructed signal can guaranteehigh accuracy. Compressed sensing data not only can be used in general signal compression, but also can be used in the image signal compression. In this paper,MATLAB simulation o
7、f the sensor storage compression for image compression and reconstruction, the sensor with the JPEG compression algorithm compression algorithm to do the experimental comparison, further analysis of the compressed sensing applications.In the sensing theory and a variety of compression coding algorit
8、hm based on this new two-dimensional image compression methods to explore, experiment using MATLAB to explore further analysis of new algorithms improve the program. Article stating that this new method of image compression is similar to the block erase algorithm. KEY WORDS: Sensor compression, imag
9、e compression, sampling, signal reconstruction 北京邮电大学本科毕业设计（论文）I目录第一章 绪论 .11.1 课题背景 .11.2 传感压缩的提出 .11.3 本文作者的主要工作和成果 .21.4 论文安排 .3第二章 传感压缩理论 .42.1 稀疏表示 .42.2 测量矩阵 .52.3 重构 .62.4 衡量压缩算法的指标 .72.4.1 采样次数 .72.4.2 压缩比 .72.4.3 重构误差 .82.4.4 时间开销 .82.4.5 空间开销 .82.5 本章小结 .8第三章 传感压缩算法实现 .103.1 传感压缩实现的主要步骤 .10
10、3.1.1 小波矩阵生成 .103.1.2 测量矩阵的生成 .123.1.3 OMP 算法重构稀疏信号 .123.2 参数分析 .143.3 本章小结 .16第四章 传感压缩与 JPEG 对比 .184.1 JPEG .184.1.1 霍夫曼编码 .184.1.2 行程编码 .194.1.3 变换编码 .204.1.4 预测编码 .204.1.5 JPEG 算法 .214.2 JPEG 压缩实验 .224.3 传感压缩的进一步编码 .234.3.1 直接霍夫曼编码 .244.3.2 保留两位有效小数位的霍夫曼编码 .244.4 对比实验 .264.5 本章小结 .28第五章 相似块删除算法 .305.1 图像块判定相似方法 .305.2 相似则删除保存索引 .31北京邮电大学本科毕业
