國立高雄大學圖資館 |

語系: 繁體中文

說明(常見問題)

圖資館首頁

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

基於相對梯度邊界成分分析之盲訊號分離演算法開發 = Developmen...

劉韋慶

基於相對梯度邊界成分分析之盲訊號分離演算法開發 = Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach

紀錄類型:	書目-語言資料,印刷品 : 單行本
並列題名:	Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach
作者:	劉韋慶,
其他團體作者:	國立高雄大學
出版地:	[高雄市]
出版者:	國立高雄大學;
出版年:	2015[民104]
面頁冊數:	[12]，94葉圖，表 : 30公分;
標題:	盲訊號分離
標題:	blind source separation (BSS)
電子資源:	https://hdl.handle.net/11296/ucq4pe
附註:	109年11月18日公開
摘要註:	本論文所提出的相對梯度邊界成分分析演算法使用相對梯度法做為邊界成分分析演算法的最佳化方法,可以化簡矩陣求逆和求共變異數矩陣的運算。我們發現演算法不用經過去均值和白化等預處理過程,即可進行疊代運算。分析複雜度後發現,時間複雜度和空間複雜度比原本的邊界成分分析演算法低。論文中使用獨立和具有相關度的訊號和圖像進行盲訊號分離的模擬,並比較演算法效能。在數據長度為200筆到10000筆的子高斯訊號分離實驗中,使用相對梯度邊界成分分析演算法分離的效能參數數值大部分都小於0.1,分離結果比原本的邊界成分分析演算法穩定。在具有相關度的圖像分離中,SSIM數值都大於0.99。最後,我們使用45張圖像,每次取2張圖做為原始圖像,配對出990組盲圖像,分離結果顯示相對梯度邊界成分分析演算法的效能參數數值有79.29 %都小於0.01。 In this thesis, we proposed the relative-gradient bounded component analysis (RGBCA) algorithm using the relative gradient to find the maximum value of the objective function. The algorithm does not need to compute the inverse matrix and the covariance matrix, and it can separate the mixed signals without whitening. The time complexity and the space complexity of this algorithm are both lower than those of the original BCA algorithm. Both of the independent sources and the dependent sources were used in the experiments for comparing the separation performances of the algorithms. For the sub-Gaussian signals with the data lengths ranging from 200 to 10000, we found that the performance indices of RGBCA algorithm are almost less than 0.1. Also, the stability of RGBCA algorithm is higher than that of BCA algorithm. For the dependent image separation, the qualities of the output images are good with the SSIM indices greater than 0.99. Finally, we analyzed the 990 results of the blind image separation of RGBCA algorithm. The simulation results indicated that 79.29 % of the 990 outcomes with the performance indices less than 0.01.

基於相對梯度邊界成分分析之盲訊號分離演算法開發 = Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach
劉, 韋慶

基於相對梯度邊界成分分析之盲訊號分離演算法開發 = Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach / 劉韋慶撰 - [高雄市] : 國立高雄大學, 2015[民104]. - [12]，94葉 ; 圖，表 ; 30公分.
109年11月18日公開.
參考書目：葉80-81.
盲訊號分離blind source separation (BSS)

基於相對梯度邊界成分分析之盲訊號分離演算法開發 = Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach
LDR:03486nam0a2200277 450 001 458458
005 20220310133912.0
010 0 $b 精裝
010 0 $b 平裝
100 $a 20151007d2015 k y0chiy50 e
101 0 $a chi $d chi $d eng
102 $a tw
105 $a ak am 000yy
200 1 $a 基於相對梯度邊界成分分析之盲訊號分離演算法開發 $d Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach $z eng $f 劉韋慶撰
210 $a [高雄市] $c 國立高雄大學 $d 2015[民104]
215 0 $a [12]，94葉 $c 圖，表 $d 30公分
300 $a 109年11月18日公開
314 $a 指導教授：陳春僥博士
320 $a 參考書目：葉80-81
328 $a 碩士論文--國立高雄大學電機工程學系碩士班
330 $a 本論文所提出的相對梯度邊界成分分析演算法使用相對梯度法做為邊界成分分析演算法的最佳化方法,可以化簡矩陣求逆和求共變異數矩陣的運算。我們發現演算法不用經過去均值和白化等預處理過程,即可進行疊代運算。分析複雜度後發現,時間複雜度和空間複雜度比原本的邊界成分分析演算法低。論文中使用獨立和具有相關度的訊號和圖像進行盲訊號分離的模擬,並比較演算法效能。在數據長度為200筆到10000筆的子高斯訊號分離實驗中,使用相對梯度邊界成分分析演算法分離的效能參數數值大部分都小於0.1,分離結果比原本的邊界成分分析演算法穩定。在具有相關度的圖像分離中,SSIM數值都大於0.99。最後,我們使用45張圖像,每次取2張圖做為原始圖像,配對出990組盲圖像,分離結果顯示相對梯度邊界成分分析演算法的效能參數數值有79.29 %都小於0.01。 In this thesis, we proposed the relative-gradient bounded component analysis (RGBCA) algorithm using the relative gradient to find the maximum value of the objective function. The algorithm does not need to compute the inverse matrix and the covariance matrix, and it can separate the mixed signals without whitening. The time complexity and the space complexity of this algorithm are both lower than those of the original BCA algorithm. Both of the independent sources and the dependent sources were used in the experiments for comparing the separation performances of the algorithms. For the sub-Gaussian signals with the data lengths ranging from 200 to 10000, we found that the performance indices of RGBCA algorithm are almost less than 0.1. Also, the stability of RGBCA algorithm is higher than that of BCA algorithm. For the dependent image separation, the qualities of the output images are good with the SSIM indices greater than 0.99. Finally, we analyzed the 990 results of the blind image separation of RGBCA algorithm. The simulation results indicated that 79.29 % of the 990 outcomes with the performance indices less than 0.01.
510 1 $a Development of Blind-Source-Separation Algorithm based on Bounded-Component Analysis Associated with Relative-Gradient Approach $z eng
610 # 0 $a 盲訊號分離 $a 邊界成分分析 $a 相對梯度 $a 獨立成分分析 $a 盲圖像分離
610 # 1 $a blind source separation (BSS) $a bounded component analysis (BCA) $a relative gradient $a independent component analysis (ICA) $a blind image separation
681 $a 008M/0019 $b 542201 7240 $v 2007年版
700 1 $a 劉 $b 韋慶 $4 撰 $3 709697
712 0 2 $a 國立高雄大學 $b 電機工程學系碩士班 $3 166118
801 0 $a tw $b NUK $c 20201112 $g CCR
856 7 # $u https://hdl.handle.net/11296/ucq4pe $z 電子資源 $2 http