雲端環境中壓縮感測於多媒體之應用

ASIAIR > College of Computer Science > Department of Computer Science and Information Engineering > Ministry of Science and Technology Research Project > Item 310904400/26243

Please use this identifier to cite or link to this item: http://asiair.asia.edu.tw/ir/handle/310904400/26243

Title:	雲端環境中壓縮感測於多媒體之應用
Authors:	林智揚
Contributors:	資訊學院;資訊工程學系
Keywords:	稀疏表示法;影像考貝偵測;影像相似度量測;Feature detection;image copy detection;image recognition;image retrieval;image similarity assessment;sparse representation
Date:	2011
Issue Date:	2013-07-18 07:52:30 (UTC+0)
Abstract:	壓縮感測是近年來最新的訊號擷取與壓縮技術，其最精彩之處為壓縮感測打破了Shannon的採樣定律⎯即sampling rate至少是訊號中maximum frequency的兩倍，壓縮感測可用更少的sampling rate即可還原原始訊號。壓縮感測的目的不是超越JPEG2000或是H.264，使壓縮空間得到更大改善；而是讓壓縮速度加快，降低壓縮複雜度，創造出相當低成本的攝像裝置，而不失其影像品質。壓縮感測的精神著重於sparse representation，具有抗雜訊的特性，且因其係數的重要性均等，有別於傳統的DCT、Wavelet技術，因此在抵抗失真與辨識的應用上更具有強韌性（robustness）。壓縮感測在國內仍是較新的技術，本計畫以壓縮感測為基礎，並以雲端環境為平台，提出三年期計畫，以期能深入瞭解壓縮感測之原理並提出新的應用與發展。在雲端環境的商業模式下，內容提供者需將內容先放在雲端伺服器中，以提供使用者購買與下載，使得安全性與隱密性受到極大的考驗。因此在第一年計畫中，我們針對持有不同瀏覽設備能力的client端，在雲端中提供安全的轉碼器（secure transcoder），使client端能依其設備能力得到相對應的影像品質，且雲端伺服器無法得知client所下載之內容。而計畫第二年，我們將以壓縮感測方法實現視訊監控，並將監控功能載入雲端，大量降低監控系統的使用複雜度。最後，在計畫第三年，我們將探討壓縮感測在雲端環境中可能的安全性漏洞，包括智慧財產權保護問題、隱私權保護問題、盜版等問題，並提供有效的處理方法來防堵或找出惡意的使用者。 Compressive sensing is a recently new technology of data sampling and compression. The most interesting part of compressive sensing is that it breaks the rule of Nyquist rate, one of the Shannon’s theorems, in which the sampling rate must be at least twice the maximum frequency of the original signal. In other words, the original signals can be reconstructed by using lower sampling rate based on compressing sensing. The goal of compressive sensing is not devoted to improving the compression rate of present JPEG2000 or H.264, but to reducing the compression complexity and creating low cost camera devices. Compressive sensing heavily relies on sparse representation, where each coefficient is with equal importance. Owning to this significant property, compressive sensing is more suitable for pattern recognition than traditional methods based on DCT or Wavelet. Since compressing sensing has distinguished characteristics and is still a new research topic in Taiwan, in this project, we will propose a three-year plan to draw on compressive sensing to other applications in a cloud computing environment. In the business model of cloud computing, the scenario that content providers need to send data to the remote server in order to satisfy buyers may cause the issues of security, privacy, and digital rights. To convince content providers to trust remote servers, privacy protection is an important factor in the promotion of cloud computing. Therefore, in the first year, we will propose a secure transcoder for the remote server such that the server needs to adapt sampling rates to meet the needs of different devices in the encryption domain. In this study, the server cannot comprehend the data and users can obtain the suitable image quality according to their devices. In the second year, we will propose a secure surveillance system based on compressive sensing in the remote server, which can perform general surveillance tasks in the encryption domain without knowing the real content. This study can not only greatly reduce the maintenance complexity of the surveillance system, but also highly preserve privacy for clients. Finally, in the third year, we will focus on the security issues of compressive sensing and the cloud environment, including malicious receivers and pirates. In addition, we will propose effective methods to prevent these leaks from damaging the whole system.
Appears in Collections:	[Department of Computer Science and Information Engineering] Ministry of Science and Technology Research Project

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	299	View/Open

Loading...