·1590· 工程科学学报，第39卷，第10期 grator for matching decision-making problems.J Univ Sci Technol 图像美感分类.自动化学报.2016,42(6)：904) Beng,2012,34(1):6 [16]Tang Y P,Han G D,Lu S H,et al.Flaw recognition method for (杨博，程振波，邓志东.面向匹配决策问题的漏整合神经 gun barrel panoramic images based on convolutional neural net- 元稀疏ESN网络.北京科技大学学报，2012,34(1)：6) work.Chin J Scientific Instrum,2016,37(4):871 [14]Wu A C.Neural Netork and Deep Learning.Beijing:Publish- (汤一平，韩国栋，鲁少辉，等.基于卷积神经网络的火炮身 ing House of Electronics Industry,2016 管全景图像疵病识别方法.仪器仪表学报，2016,37(4)： (吴岸城.神经网络与深度学习.北京：电子工业出版社， 871) 2016) [17]Zhang X C.Fu M Y.Selection method for scene matching area [15]Wang W N,Wang L,Zhao MQ,et al.Image aesthetic classifi- based on information entropy.Syst Eng Electr,2011,33 (5): cation using parallel deep convolutional neural networks.Acta 1089 Autom Sin,2016,42(6):904 (张晓晨，付梦印.基于信息熵的景象匹配区选取方法，系 (王伟凝，王励，赵明权，等.基于并行深度卷积神经网络的 统工程与电子技术，2011,33(5)：1089)工程科学学报,第 39 卷,第 10 期 grator for matching decision鄄making problems. J Univ Sci Technol Beijing, 2012, 34(1): 6 (杨博, 程振波, 邓志东. 面向匹配决策问题的漏整合神经 元稀疏 ESN 网络. 北京科技大学学报, 2012, 34(1): 6) [14] Wu A C. Neural Network and Deep Learning. Beijing: Publish鄄 ing House of Electronics Industry, 2016 (吴岸城. 神经网络与深度学习. 北京: 电子工业出版社, 2016) [15] Wang W N, Wang L, Zhao M Q, et al. Image aesthetic classifi鄄 cation using parallel deep convolutional neural networks. Acta Autom Sin, 2016, 42(6): 904 (王伟凝, 王励, 赵明权, 等. 基于并行深度卷积神经网络的 图像美感分类. 自动化学报, 2016, 42(6): 904) [16] Tang Y P, Han G D, Lu S H, et al. Flaw recognition method for gun barrel panoramic images based on convolutional neural net鄄 work. Chin J Scientific Instrum, 2016, 37(4): 871 (汤一平, 韩国栋, 鲁少辉, 等. 基于卷积神经网络的火炮身 管全景图像疵病识别方法. 仪器仪表学报, 2016, 37 (4): 871) [17] Zhang X C, Fu M Y. Selection method for scene matching area based on information entropy. Syst Eng Electr, 2011, 33 (5 ): 1089 (张晓晨, 付梦印. 基于信息熵的景象匹配区选取方法. 系 统工程与电子技术, 2011, 33(5): 1089) ·1590·