English  |  正體中文  |  简体中文  |  Items with full text/Total items : 58323/91867 (63%)
Visitors : 14035660      Online Users : 87
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    Please use this identifier to cite or link to this item: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/52499


    Title: A study on discrete wavelet transform for video compression and intelligent video surveillance system
    Other Titles: 以離散小波轉換應用於視訊壓縮與智慧型視訊監控系統之探討
    Authors: 夏至賢;Hsia, Chih-hsien
    Contributors: 淡江大學電機工程學系博士班
    江正雄;Chiang, Jen-shiun
    Keywords: 智慧型監控系統;提升式離散小波轉換;直接低低頻遮罩法;高解析度;intelligent video surveillance system;lifting-based discrete wavelet transform;direct LL-mask based scheme;high resolution
    Date: 2010
    Issue Date: 2010-09-23 17:51:41 (UTC+8)
    Abstract: 本論文將提供一個具有低成本且有效的二維離散小波轉換應用於智慧型視訊監控系統。其目的是將此技術應用於視訊壓縮以及電腦視覺中,利用離散小波轉換的特色發展出低轉置記憶體的超大型積體電路架構以及快速物件偵測演算法,以達到智慧型視訊監控系統的規格,並且兼具有簡單性、即時性、以及安全性的功能。在本論文中,我們將研究提升式離散小波轉換所產生出的缺點進而改良其演算法,再應用於智慧型視訊監控系統與視訊資料壓縮技術中並且達到即時處理。
    小波轉換已經在影像壓縮與處理中日漸重要,它允許在空間與頻率同時分析以及可調性視訊處理。基於離散小波轉換的特性,在本文中提出兩個新的方法:首先,我們提出新的二維雙模組提升式離散小波轉換硬體架構;在一般的二維提升式離散小波轉換中會有大量使用其轉置記憶體的缺點,其提出交錯讀取掃瞄演算法,以支援高解析度視訊以及減低轉置記憶體的成本,達到謹2N與4N(5/3或9/7模式)的使用量;另外,本方法以平行以及管線架構來增加運算時間,使其適用於Motion-JPEG2000。第二,我們提出對稱式遮罩法並利用於智慧型視訊監控系統;此方法改善二維提升式離散小波轉換的問題,以達到低延遲、低複雜度、以及低轉置記憶體;它可以達到低複雜度離散小波轉換運算以提高準確度與即時性的移動物件偵測系統,在測試的16個場景中,平均物件偵測的準確率與處理速度各別為89.59%以及每秒可執行47.1張影格數,並且應用於連續偵測之室內、外、以及日、夜間場景。
    在上述的兩種方法中均是以提升式離散小波轉換的改良為基礎。在實驗結果中得知,我們所提出的方法可以提供低轉置記憶體以及即時處理之效果應用於高解析度智慧型視訊監控系統。
    This thesis attempts to develop a low cost, practical application of a two-dimensional (2-D) discrete wavelet transform (DWT) to an intelligent video surveillance system. The goal is to combine the video compression and computer vision processing used in a wavelet-based system and to adopt DWT to develop a low transpose memory VLSI architecture and a fast object detection algorithm that can meet the specifications of intelligent video surveillance and can also process information easily, in real-time, and with safe functioning. This thesis investigates the fundamental concept behind the wavelet transform and provides an overview of some improved algorithms for lifting-based DWT (LDWT). The video surveillance system is able to detect moving object events, and video compression information is captured by surveillance cameras in real-time.
    Wavelet transformation has become increasingly important in image compression and processing since wavelets allow both simultaneous spatial and frequency analysis and scalable video processing. This thesis presents two new approaches. First, we propose new hardware architectures to address critical issues in 2-D dual-mode LDWT (supporting 5/3 lossless and 9/7 lossy coding modes). A considerably large transpose memory is the most critical requirement for LDWT. The proposed architecture can support high-resolution videos and reduce the internal memory requirement significantly. In our LDWT approach, the signal flow is revised from row-wise only to mixed row- and column-wise, and a new architecture, called interlaced read scan architecture (IRSA), is used to reduce the transpose memory. With the IRSA approach, the transpose memory size is only 2N or 4N (5/3 or 9/7 mode) for an N´N DWT. In addition, the proposed 2-D LDWT operates with parallel and pipelined schemes that increase its operation speed. It can be applied to real-time video operations for Motion-JPEG2000. Second, we propose the called symmetric mask-based DWT (SMDWT) for an intelligent video surveillance system. SMDWT improves the critical issue of the 2-D LDWT, and then obtains the benefit of low latency, reduced complexity, and low transpose memory. A highly precise and real-time moving object detection algorithm based on a low complexity SMDWT offers a mechanism for the sequential detection of both indoor (all day) and outdoor (all day) scenes. Computer simulations verified that the present algorithm performs well. It has high accuracy rate of more than 89.59% on average and the average frame rate can reach 47.1 frame per second (FPS).
    The abovementioned two algorithms for the LDWT were improved. The experimental results indicate that the proposed methods can provide low transpose memory and real-time processing for a high resolution intelligent video surveillance system.
    Appears in Collections:[Graduate Institute & Department of Electrical Engineering] Thesis

    Files in This Item:

    File SizeFormat
    index.html0KbHTML411View/Open

    All items in 機構典藏 are protected by copyright, with all rights reserved.


    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - Feedback