Ground Based SAR Data Obtained With Different Polarizations
Description
The dataset consists of 150 pairs of Ground Based SAR data recorded using horizontal and vertical polarization. All data is recorded using GBSAR-Pi. GBSAR-Pi works in stop-and-go mode. GBSAR sensor is FMCW (Frequency Modulated Continuous Wave) radar with central frequency at 24 GHz. Total GBSAR aperture is 60 cm with 1 cm step size (60 steps). FMCW radar periodically transmits sawtooth signal every 155 ms. Scene GBSARP-Pi recorded scenes which contained three equally spaced objects. The objects were set approximately 30 cm away from the radar. Resulting matrix obtained after one recording has dimensions of 1024x60. 1024 represents the number of frequency points, while 60 is the number of GBSAR steps. That matrix is split into three segments each of which covered one object. The dimensions of one segment are 1024x20. Given dataset includes 300 examples of such segments. Each scene is recorded with horizontal (HH) and vertical (VV) polarization. Therefore, the dataset contains 150 pairs of the same segment observed with different polarizations. Classes The observed objects were approximately of same size but made of different materials (aluminum, glass, plastic) and had different shapes (cylinder and cuboid). Each of the six combinations of materials and shapes are represented by one object and each object stands for one class. Six classes and their marks are: PR - plastic cylinder PS - plastic cuboid GR - glass cylinder GS - glass cuboid AR - aluminum cylinder AS - aluminum cuboid Naming of the files Class mark is based on the material of the object and its shape. The rest of the filename is following: {class mark}_{position in the scene (1, 2 or 3)}_{polarization}_{recording id}.npy Therefore, 'AR_1_hh_21' and 'AR_1_vv_21' represent the same object recorded in the same scene conditions (recording id) but with different polarizations (horizontal and vertical, respectively). Data format GBSAR matrices are given in .npy format. The dimensions of the matrices are 1024x20. This work was supported in part by Croatian Science Foundation (HRZZ) under the project number IP-2019-04-1064. [1] Kačan M, Turčinović F, Bojanjac D, Bosiljevac M. Deep Learning Approach for Object Classification on Raw and Reconstructed GBSAR Data. Remote Sensing. 2022; 14(22):5673. https://doi.org/10.3390/rs14225673
Files
Steps to reproduce
Developed GBSAR is based on Raspberry Pi 4B (RPi) microcomputer. It controls VCO in FMCW module Innosent IVS-362 (https://www.innosent.de/radarsensoren/ivs-series/ivs-362/) which has integrated transmitting and receiving antenna, and mixer. RPi and FMCW module are set on the platform which is moved along the rail track for 1 cm in each step with 5V stepper motor also controlled by RPi.