Specific ship detection for electronic reconnaissance data based on clustering and NNs. Issue 8 (1st August 2018)

• Record Type: Journal Article
• Authors: Chen, Wenbin; Xu, Guangluan; Ren, Wenjuan; Huang, Tinglei
• Is Part Of: IET radar, sonar & navigation. Volume 12:Issue 8(2018)
• Page Start: 868
• Page End: 872
• Publication Date: 2018-08-01
• Subjects: object detection -- neural nets -- feature extraction -- time series -- learning (artificial intelligence); ship detection -- electronic reconnaissance data -- clustering methods -- machine‐learning methods -- long short‐term memory models -- time‐series features -- one‐dimensional convolution neural networks; Signal processing -- Periodicals; Radar -- Periodicals; Sonar -- Periodicals; Electronics in navigation -- Periodicals; Navigation -- Periodicals; 621.3848
• DOI: 10.1049/iet-rsn.2018.0003
• Languages: English
• ISSNs: 1751-8784

Abstract

Abstract : Ship detection is an important issue in both civil and military fields. There are many researches about ship detection in wide sea areas and inshore areas. Most of these researches use the image dataset. In this research, a novel framework for specific ship detection using the electronic reconnaissance dataset is proposed. In the new framework, the clustering methods based on time and trajectory are used to process the dataset. Then, several machine‐learning methods are used to build detection models. The long short‐term memory models are used to extract time‐series features and the carefully designed one‐dimensional (1D) convolution neural networks (CNNs) are introduced to extract local features. Experimental results on a large electronic reconnaissance dataset collected from real scenario show the model based on 1D CNN gets better performance than classic detection models and the authors' system achieves a good detection accuracy of 92.5%. Above all, this research is a very valuable exploration for the detection of specific ship based on electronic reconnaissance dataset.