AI Radar Sensor: Creating Radar Depth Sounder Images Based on Generative Adversarial Network

Abstract

Significant resources have been spent in collecting and storing large and heterogeneous radar datasets during expensive Arctic and Antarctic fieldwork. The vast majority of data available is unlabeled, and the labeling process is both time-consuming and expensive. One possible alternative to the labeling process is the use of synthetically generated data with artificial intelligence. Instead of labeling real images, we can generate synthetic data based on arbitrary labels. In this way, training data can be quickly augmented with additional images. In this research, we evaluated the performance of synthetically generated radar images based on modified cycle-consistent adversarial networks. We conducted several experiments to test the quality of the generated radar imagery. We also tested the quality of a state-of-the-art contour detection algorithm on synthetic data and different combinations of real and synthetic data. Our experiments show that synthetic radar images generated by generative adversarial network (GAN) can be used in combination with real images for data augmentation and training of deep neural networks. However, the synthetic images generated by GANs cannot be used solely for training a neural network (training on synthetic and testing on real) as they cannot simulate all of the radar characteristics such as noise or Doppler effects. To the best of our knowledge, this is the first work in creating radar sounder imagery based on generative adversarial network.

Multi-Objective Heterogeneous Multi-Asset Collection Scheduling Optimization with High-Level Information Fusion

“…These methods were faced with the lack of suitable datasets to predict the quality for different categories of images and the lack of datasets containing as many attributes as desired. [76,77] highlight these problems and suggest solutions that can lead to datasets with synthetic images.…”

Automatic Detection of Basal Units beneath Antarctic Ice Sheet in Radargram Based on Deep Learning

“…Literature [26] used U-Net network and ASPP module to detect ice layer structure, classifying radargram into bedrock, echo free area, ice layer and basal ice, which is somewhat similar to basal units. Recently, deep learning has been applied to radar image recognition and achieved good results, mainly for: 1) ice detection [26][27]; 2) Simulate RS image using generative countermeasure network (GAN) [28]; 3) Target detection [29]; And 4) Radargram segmentation [24] [30] [31].…”

Material selection in sensor design for additive manufacturing

“…Accelerometer [24], Radiometer (Hyperspectral, Imaging etc.) [25][26][27][28], Sounder [29], Spectrometer [30] and Spectroradiometer [31], which do not require any external power source to produce signals, have the desired physical or chemical values. Sensors that can convert it into an output variable are Passive sensors.…”