摘要
针对电信机房中继器上的光纤跳线与端子进行匹配的问题,提出了一种基于深度学习的目标定位与数字识别的系统。该系统优化了深度学习中单点多盒探测器(SSD)算法与快速基于区域的卷积神经网络(Faster R-CNN)算法的网络结构,结合SSD算法提取有效区域速度快的特点,对自然场景下拍摄的图片进行读数区域的有效分割,然后使用Faster R-CNN算法进行读数区域识别。该系统在实验中测试成功率达到99.9%,能够确保端子号和光纤跳线做到一一对应。
In order to solve the problem of mismatch between optical fiber jumper and terminal on relay in telecommunication room, a target location and digital recognition system based on deep learning is proposed. The system optimizes the network structure of single shot multibox detector(SSD) algorithm and Faster region-based convolutional neural networks(Faster R-CNN) algorithm in deep learning. Combining with the fast feature of SSD algorithm to extract the effective area, the system effectively segmentes the reading area of the natural scene images, and then uses Faster R-CNN algorithm to recognize the reading area. The test success rate of the system reaches 99.9% in the experiment, which can ensure that the terminal number and the optical fiber jumper are one-to-one correspondence.
引文
[1]UIJLINGS J R R,SANDE K E A V D,GEVERS T,et al.Selective Search for Object Recognition[J].International Journal of Computer Vision,2013,104(2):154-171.
[2]SHI H,WANG L,CHU T G.Swarming behavior of multi-agent systems[J].Journal of Control Theory&Applications,2004(4):313-318.
[3]LIU F,GLEICHER M.Automatic image retargeting with fisheye-view warping[C]//Proceedings of the 18th annual ACM symposium on User interface software and technology,October 23-26,2005,Seattle,USA.New York:ACM,2005.
[4]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:Towards Real-Time Object Detection with Region Proposal Networks[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,39(6):1137-1149.
[5]吴川,朱明,杨冬.模糊理论与BP网络在目标识别中的应用[J].测试技术学报,2005,19(3):287-293.
[6]吴梦麟.图像式水表读数识别方法研究[D].南京:南京理工大学,2006.
[7]基于数字图像处理的水表读数识别系统应用研究[D].成都:电子科技大学,2017.
[8]周成伟.基于卷积神经网络的自然场景中数字的识别[D].南京:南京邮电大学,2017.
[9]TENENBAUM J B.A Global Geometric Framework for Nonlinear Dimensionality Reduction[J].Science,2000,290(5500):2319-2323.
[10]YANG J,ZHANG D,FRANGI A F,et al.Two-dimensional PCA:a new approach to appearance-based face representation and recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2004,26(1):131-137.
[11]CHANG Chih-Chung,LIN Chih-Jen.A Library for Support Vector Machines[Z/OL].2011-05-20[2018-12-06].https://wenku.baidu.com/view/b50dec6cb84ae45c3b358c18.html.
[12]王敏,黄心汉.基于模板匹配和神经网络的车牌字符识别方法[J].模式识别与人工智能,2001,14(1):96-98.