Open AccessDetection and classification of small-scale objects in images obtained by synthetic-aperture radar stations
Author(s) -
Е. А. Казачков,
С. Н. Матюгин,
И. В. Попов,
В. В. Шаронов
Publication year - 2018
Publication title -
vestnik koncerna vko «almaz - antej»
Language(s) - English
Resource type - Journals
ISSN - 2542-0542
DOI - 10.38013/2542-0542-2018-1-93-99
Subject(s) - artificial intelligence , computer science , convolutional neural network , synthetic aperture radar , pattern recognition (psychology) , radar , automatic target recognition , radar imaging , scale (ratio) , object detection , computer vision , contextual image classification , deep learning , remote sensing , image (mathematics) , geology , geography , telecommunications , cartography
The investigation deals with the problem of simultaneous detection and classification (that is, recognition) of several classes of objects in radar images by means of convolutional neural networks. We present a two-stage processing algorithm that detects and recognises objects. It also features an intermediate sub-stage that increases the resolution of those zones where objects have been detected. We show that a considerable increase in detection and recognition probabilities is possible if the recognition module is trained using high-resolution data. We implemented the detection and recognition stages using deep learning approaches for convolutional neural networks.