English  |  正體中文  |  简体中文  |  Items with full text/Total items : 26988/38789
Visitors : 2348957      Online Users : 37
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/28702

Title: 水下步進相移陣列超音波影像系統之研發(I)-子計畫五:動態物件追蹤暨影像定位之技術開發
Authors: 賴俊廷;曾俊舜;鄭雅芸;王榮華
Contributors: NTOU:Department of Electrical Engineering
國立臺灣海洋大學:電機工程學系
Keywords: image segmentation;neural networks;noise filtering;feature extraction;recognition;underwater ultrasonic
Date: 2009
Issue Date: 2011-10-21T02:39:24Z
Publisher: 國科會專題研究計畫成果報告
Abstract: Abstract:This final report describes the results of fulfilling a sub-project of a joint research granted by NSC, namely, Research and development of an underwater stepping phased- array ultrasonic imaging system. This report presents an underwater ultrasonic image processing system, wherein the accuracy of image segmentation is essential to the entire project. A novel approach for segmentation, Self-organizing Fusion Neural Network (SOFNN), with robust noise tolerance, accurate segmentation results, and high computational efficiency is developed. SOFNN is applicable to real-time tracking system. We also adopt the Scale Shrinking High-order Filter (SSHF), which employs a high order network for removing multiplicative noise present in underwater ultrasonic images. We incorporate Kovesi Detector and Laplacian of Gaussians (LOG) to extract rotational invariant features. Finally, in order to enhance object recognition capability, we use Normalized Cross-Correlation (NCC) to find the corresponding features.
摘要:本論文係「水下步進相移陣列超音波影像系統之研發(I)」整合型計劃中子計畫六「動態物件追蹤暨影像定位之技術開發」之成果報告。本計畫擬開發一整合性之水下超音波影像處理系統,其中影響此一系統最甚之成敗關鍵即為影像分割技術之精準性,吾人成功研發一適用於水下影像之影像分割演算法自我組織式融合神經網路(Self-organizing Fusion Neural Network, SOFNN) [2],其具有強健的雜訊容忍度、精準的分割正確性,以及快速的運算效率,對於未來建構即時性的影像追蹤將有相當之助益。而為完整建構一影像處理系統,吾人引用先前研發之刻度收縮高階網路濾波器(Scale Shrinking High-order Filter, SSHF) [9],以其高階神經網路於函數逼近之功能有效地將影像中之乘法性雜訊濾除;並利用先前所提出之Kovesi Detector與Laplacian of Gaussians (LOG)之接合進行特徵值之擷取,係藉由LOG克服了習知超音波影像辨識在物件旋轉後辨識效果不佳之缺點,與運用Normalized Cross-Correlation (NCC)作為對應點匹配之方法,進行準確之影像辨識作業。
URI: http://ntour.ntou.edu.tw/handle/987654321/28702
Appears in Collections:[電機工程學系] 研究計畫

Files in This Item:

There are no files associated with this item.



All items in NTOUR are protected by copyright, with all rights reserved.

 


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback