基于深度神经网络的多模态计算机图像识别  

Multimodal Computer Image Recognition Based on Deep Neural Network

作　　者：越缙 周晓成 YUE Jin;ZHOU Xiaocheng(School of Computer Engineering,Anhui Wenda University of Information Engineering,Hefei 231201,China)

机构地区：[1]安徽文达信息工程学院计算机工程学院,安徽合肥231201

出　　处：《安阳师范学院学报》2023年第5期31-35,共5页Journal of Anyang Normal University

基　　金：安徽省教育厅重点科研项目(项目编号:2022AH052847)。

摘　　要：为消除锐化与噪声,提高多种不同类型多模态图像的识别精度,基于深度神经网络构建多模态图像识别模型。所提出的图像识别模型通过缩减原则降低相邻顶点间的相似性,采用相似度计算规则估算相邻两个顶点的相似性,有效提升计算机图像识别效率。为节约图像识别运算的空间,在Spark中引入了GraphX的GXDSGC。将提出的方法应用于实际的多模态图像识别中,结果表明所提出的识别算法无须占用大量的硬盘I/O资源,所耗费时间明显缩短,且GXDSGC算法比Hadoop中基于MapReduce框架的算法快30倍以上,显著提高了大数据分析中计算机图像识别的效率。To eliminate sharpening and noise,and improve the recognition accuracy of various types of multimodal images,a multimodal image recognition model is constructed based on deep neural networks.The proposed image recognition model reduces the similarity between adjacent vertices through the reduction principle,and uses similarity calculation rules to estimate the similarity between adjacent two vertices,effectively improving the efficiency of computer image recognition.To save space for image recognition operations,GraphX's GXDSGC was introduced in Spark.The proposed method is applied to the actual multimodal image recognition,and the results show that the proposed recognition algorithm does not need to occupy a lot of hard disk I/O resources,and the time consumed is significantly shortened,and the GXDSGC algorithm is more than 30 times faster than the algorithm based on the MapReduce framework in Hadoop,which significantly improves the efficiency of computer image recognition in Big data analysis.

关 键 词：深度神经网络 多模态图像 图像识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]