<address id="xrvvb"><listing id="xrvvb"></listing></address>

          <form id="xrvvb"></form>

          <address id="xrvvb"></address>

          華中科技大學學報(自然科學版) 2020, Vol. 48 Issue (9): 38-43 DOI10.13245/j.hust.200907

          欄目:計算機與控制工程
          基于多特征融合的自適應權重目標分類方法
          王立鵬 , 張 智 , 蘇 麗 , 聶文昌
          哈爾濱工程大學自動化學院,黑龍江 哈爾濱 150001
          摘要 提出一種自適應權重的融合卷積特征和方向梯度直方圖(HOG)特征的目標分類方法,實現快速精準分類目的.首先,利用ResNet網絡框架提取圖像卷積特征,增加OpenCV接口以提取圖像HOG特征,對HOG特征圖擴維處理至與卷積特征同維;然后,在ResNet網絡框架中嵌入SENet模塊,計算卷積特征和HOG特征的權重向量,將卷積特征、HOG特征及相應的權重向量加以變權值疊加,實現多特征的自適應同步融合,以此構建二分類網絡模塊;其次,將二分類網絡模塊嵌入Faster Rcnn網絡,構成Faster Rcnn-HOG新型網絡,通過基于變閾值的粗檢測策略和先驗知識的區域關注策略得到圖像中目標預處理檢測框,利用二分類網絡模塊精確判定,實現目標分類.將Faster Rcnn-HOG與傳統Faster Rcnn網絡及另一特征融合網絡Net-BB-HOG進行對比試驗,三種方法在目標大類識別方面性能基本相當,但是Faster Rcnn-HOG在目標小類識別方面效果更佳,證明了提出的多特征融合自適應目標分類方法的有效性和正確性.
          關鍵詞 目標分類 ;深度學習 ;卷積神經網絡 ;多特征融合 ;自適應權重
          Target classification with adaptive weights based on multi-feature fusion
          WANG Lipeng , ZHANG Zhi , SU Li , NIE Wenchang
          College of Automation,Harbin Engineering University,Harbin 150001,China
          Abstract To solve precision forming problem of the antenna panels with large non-developable double,a target classification method with adaptive weights was proposed on the basis of fusion of convolution feature and histogram of oriented gradient (HOG) feature,which was utilized to classify the targets quickly and precisely.First of all,convolution feature was extracted through the ResNet framework,in which the OpenCV interface was increased to acquire the HOG feature of the images.The dimensions of HOG feature were enlarged to maintain the same dimensions as the convolution feature.Second,SENet module was imbedded into the ResNet framework so that the weight vectors of the convolution feature and HOG feature were calculated.The features of the images were adaptively and synchronously fused based on the convolution feature,HOG feature,and the weight vectors.An innovation binary network was established based on the multi-feature fusion.Third,the binary network was imbedded into the Faster Rcnn network to establish Faster Rcnn-HOG,in which the pre-processing detection frames of the image was acquired through the strategies of coarse detection of variable threshold and focus area of prior knowledge.Then the pre-processing detection frames was precisely judged by the proposed binary network to realize the target classification.The comparative experiments among faster Rcnn-HOG,the traditional Faster Rcnn,and another feature fusion network Net-BB-HOG were conducted.The results verify that the effect of the three methods is similar in the target classification of large categories.However,faster Rcnn-HOG is more effective in identifying small categories of the targets.The validity and correctness of the proposed method is proved.
          Keywords target classification ; deep learning ; convolutional neural network ; multi-feature fusion ; adaptive weights
          基金資助國家自然科學基金資助項目(61803116);中央高?;究蒲袠I務費專項資金資助項目(3072020CF0410)

          中圖分類號TP391.4
          文獻標志碼A
          文章編號1671-4512(2020)09-0038-06
          參考文獻
          [1] 張華.基于空間顏色特征的行人重識別方法[J].華中科技大學學報(自然科學版),2013,41(S2):209-212.
          [2] 王慧斌,高國偉,徐立中,等.基于紋理特征的多區域水平集圖像分割方法[J].電子學報,2018,11(11):2588-2596.
          [3] 薛夢霞,劉士榮,王堅.基于機器視覺的動態多目標識別[J].上海交通大學學報(自然科學版),2017,51(6):727-733.
          [4] PANG S C,COZ J J,YU Z Z,et al.Deep learning to frame objects for visual target tracking[J].Engineering Applications of Artificial Intelligence,2017,65:406-420.
          [5] REHMAN S U,TU S S,HUANG Y F,et al.CSFL:a novel unsupervised convolution neural network approach for visual pattern classification[J].AI Communica-ions,2017,30(5):311-324.
          [6] GIRSHICK R.Fast R-CNN[C]// Proceedings of the IEEE International Conference on Computer Vision.Santiago:Institute of Electrical and Electronics Engineers Inc,2015:1440-1448.
          [7] 黃祖微.基于深度學習的多特征融合的目標檢測[D].杭州:浙江大學計算機科學與技術學院,2018.
          [8] LU T W,WANG D D,ZHANG Y D.Fast object detection algorithm based on HOG and CNN[C]// Proc of 9th International Conference on Graphic and Image Processing (ICGIP).Qingdao:International Society for Optical Engineering,2017:1-6.
          [9] YE F,ZHANG X G.An automatic annotation algorithm for deep learning image datasets based on HOG features [C]// Proc of 2nd International Conference on Applied Mathematics,Modeling and Simulation (AMMS). Beijing:Destech Publications Inc,2018:128-135.
          [10] 顧江鵬,袁和金.基于HOG與卷積神經網絡的人臉識別研究[J].軟件導刊,2018,17(11):1-5.
          [11] CAI L,ZHU J Q,ZENG H Q,et al.HOG-assisted deep feature learning for pedestrian gender recognition[J].Journal of the Franklin Institute,2018,355(4):1991-2008.
          [12] 劉成云,常發亮,陳振學.基于感興趣區域和HOG- MBLBP特征的交通標識檢測[J].電子與信息學報,2016,38(5):1092-1098.
          [13] 郭保青,王寧.基于改進深度卷積網絡的鐵路入侵行人分類算法[J].光學精密工程.2018,26(12):3040-3050.
          [14] LI B Q,HE Y Y.An improved ResNet based on the adjustable shortcut connections[J].IEEE Access,2018,6:18967-18974.
          [15] HU J,SHEN L,ALBANIE S,et al.Squeeze-and- excitation networks[C]// Proc of 2018 IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition.Salt Lake City:IEEE Computer Society,2018:7132-7141.
          文獻來源
          王立鵬, 張 智, 蘇 麗, 聶文昌. 基于多特征融合的自適應權重目標分類方法[J]. 華中科技大學學報(自然科學版), 2020, 48(9): 38-43
          DOI:10.13245/j.hust.200907
          澳门盘口