<progress id="blzj5"></progress>
<menuitem id="blzj5"><del id="blzj5"><address id="blzj5"></address></del></menuitem><thead id="blzj5"></thead>
<menuitem id="blzj5"><dl id="blzj5"></dl></menuitem>
<thead id="blzj5"><dl id="blzj5"></dl></thead>
<var id="blzj5"><dl id="blzj5"></dl></var>
<listing id="blzj5"><dl id="blzj5"><noframes id="blzj5">
<thead id="blzj5"><dl id="blzj5"></dl></thead><listing id="blzj5"></listing><listing id="blzj5"></listing>
<menuitem id="blzj5"></menuitem>
<ins id="blzj5"></ins>
<thead id="blzj5"><del id="blzj5"><th id="blzj5"></th></del></thead><listing id="blzj5"></listing>
<thead id="blzj5"></thead>
熊承義,李世宇,高志榮,金鑫.級聯模型展開與殘差學習的壓縮感知重構[J].中南民族大學學報自然科學版,2019,(2):265-272
級聯模型展開與殘差學習的壓縮感知重構
Compressive sensing reconstruction via stacked unfolding model and residual learning
  
DOI:10.12130/znmdzk.20190221
中文關鍵詞: 壓縮感知  深度學習  模型展開  殘差學習
英文關鍵詞: compressive sensing  deep learning  model unfolding  residual learning
基金項目:國家自然科學基金資助項目(61471400),中央高?;究蒲袠I務經費專項資金項目(CZY19016)
作者單位
熊承義1,李世宇1,高志榮2,金鑫1 1 中南民族大學 電子信息工程學院,武漢430074; 2中南民族大學 計算機科學學院,武漢 430074 
摘要點擊次數: 196
全文下載次數: 185
中文摘要:
      基于傳統優化模型展開的深度網絡由于集成了深度學習與傳統優化方法的優點,具有良好的可解釋性,在當前圖像處理與計算機視覺領域得到廣泛關注。提出了一種級聯模型展開與殘差學習的圖像壓縮感知重構深度網絡框架,以實現重構圖像質量的進一步改善。第一級的基于模型展開的深度網絡根據輸入的壓縮測量值得到初始的重構圖像,第二級的深度殘差網絡對初始重構圖像進行去噪處理,最終得到高質量的重構結果。該兩級級聯網絡的訓練分別獨立完成,訓練過程簡單易實現,將ADMM-Net與ResNet級聯實現對磁共振圖像重構,將ISTA-Net+與ResNet級聯實現對自然圖像重構。大量實驗結果比較驗證了所提出方法的有效性。
英文摘要:
      Deep networks based on unfolding conventional optimization model have been paid widely attention in many fields including image processing, computer vision and so on, because they not only combine the advantages of current deep leaning and conventional optimization-based approach, but also character well interpretability. A novel deep network architecture for compressive sensing image reconstruction is proposed by cascading model unfolding and residual learning, which aims to further improving the reconstructed image quality. The first stage of deep network is designed based on model unfolding to transform the compressed measurements of input into the initial reconstruction, and the second stage is a deep residual network to remove the noise in the initial reconstruction, consequently producing higher quality of reconstruction image. The training of the two-stage network is completed independently, which is simple and easy to conduct. Specifically, stacking the ADMM-Net and ResNet to reconstruct magnetic resonance imaging, and stacking the ISTA-Net+ and ResNet to reconstruct natural images. Extensive experimental results comparison demonstrates the effectiveness of the proposed method.
查看全文   查看/發表評論  下載PDF閱讀器
關閉
秒速赛车开奖查询结果