面向水下多源数据特征级融合方法  

Feature-Level Fusion Method for Underwater Multisource Data

作　　者：宋奎勇 周连科[1] 王红滨[1] SONG Kuiyong;ZHOU Lianke;WANG Hongbin(College of Computer Science and Technology,Harbin Engineering University,Harbin 150000,China;Department of Information Engineering,Hulunbuir Vocational Technical College,HulunBuir 021000,China)

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,哈尔滨150000 [2]呼伦贝尔职业技术学院信息工程系,内蒙古呼伦贝尔021000

出　　处：《吉林大学学报（信息科学版）》2021年第3期331-338,共8页Journal of Jilin University（Information Science Edition）

基　　金：国家自然科学基金资助项目(61772152);国家重点研发计划基金资助项目(2018YFC0806800);技术基础科研基金资助项目(JSQB2017206C002);中国博士后科学基金资助项目(2019M651262);教育部人文社科研究青年基金资助项目(20YJCZH172);黑龙江省博士后基金会基金资助项目(LBH-Z19015)。

摘　　要：海洋环境复杂多变,单一水下传感器目标识别精度不能满足系统性能要求,并且水下数据噪声大、维度高,直接进行数据融合并不能得到较好的结果。为此,针对多场景水下多源试验数据,使用去噪自编码和多种降维方法进行多角度特征级融合。首先,使用去噪自编码器去除噪声、降低数据维度并且抽取出深层特征;然后,对深层特征使用数据层叠方法进行多源数据融合。融合方法包括主成分分析、独立分量分析和等度量映射。不同场景下对比试验表明该方法取得较好的分类结果,其中主成分分析取得较高目标识别率。The marine environment is complex and changeable,and the target recognition accuracy of a single underwater sensor can not meet the performance requirements of the system. Multi-source sensor fusion is an effective method that can improve the target recognition rate. It has received extensive attention and research.Underwater data is noisy and has high dimensions,and direct data fusion can not get better results. For multiscene underwater multi-source test data,denoising autoencoder and multiple dimensionality reduction methods is used for multi-angle feature-level fusion. First,the denoising autoencoder is used to remove noise and reduce the data dimension,and extract new features from the source data. Then,the data cascade method is used for multisource data fusion for new features. The fusion methods include principal component analysis,independent component analysis and isometric mapping. Comparative experiment results in different scenarios show that the proposed method gets better classification results,and principal component analysis can achieve a higher target recognition rate.

关 键 词：水下多源数据 特征级融合 去噪自编码 降维方法 数据层叠方法 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]