joeyee / VFSVM

Small-floating Target Detection in Sea Clutter by Classifying Visual Feature in Time Doppler Spectra of IPIX data sets

VFSVM

Small-floating Target Detection in Sea Clutter by Classifying Visual Feature in Time Doppler Spectra of IPIX data sets

This is the python implementation of the - Small-floating Target Detection in Sea Clutter by Classifying Visual Feature in Time Doppler Spectra.

Requirements

• python - 3.6.5
• opencv-python
• sklearn
• netCDF - 1.5.3

How to use the code

Step 1

Download the complex-sequential returns from the IPIX 1993 and IPIX 1998

We rewrite the Python code to load the IPIX raw data in 'Load_IPIX_xxxx.py'.

Convert the sequential returns to Time Doppler Spectra (TDS) images.

Step 2

Compute the Local Binary Patterns (LBP) histogram for each TDS images.

Step 3

Train the v-SVM with impure samples.

Step 4

Sort the distances to the learned center. Select the sample with maximal distance as the target.

Introduction

This algorithm is introduced in paper, which is under review. Once the paper is allowed to be published, we will release all the codes soon. Now we have published the data loading code, which is a Python re-implementation according to the Matlab version on IPIX website.

Bibtex

@misc{zhou2020smallfloating, title={Small-floating Target Detection in Sea Clutter via Visual Feature Classifying in the Time-Doppler Spectra}, author={Yi Zhou and Yin Cui and Xiaoke Xu and Jidong Suo and Xiaoming Liu}, year={2020}, eprint={2009.04185}, archivePrefix={arXiv}, primaryClass={eess.SP} }