HCNN-PSI: A hybrid CNN with partial semantic information for space target recognition

作者：

Highlights：

• We use data augmentation techniques and image processing to represent the wide image of space target in the deep space background.

• We propose a new localization method improved from minimum bounding rectangle (MBR) model.

• We propose a hybrid network that can integrate multi-source information which can get a better recognition performance than that who only utilize single information.

摘要

•We use data augmentation techniques and image processing to represent the wide image of space target in the deep space background.•We propose a new localization method improved from minimum bounding rectangle (MBR) model.•We propose a hybrid network that can integrate multi-source information which can get a better recognition performance than that who only utilize single information.

论文关键词：Space target recognition,Hybrid convolutional neural network with partial semantic information,Data augmentation,Components segment

论文评审过程：Received 23 October 2019, Revised 29 April 2020, Accepted 1 July 2020, Available online 6 July 2020, Version of Record 11 July 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107531