Classification of Geometrical Targets Using Natural Resonances and Principal Components Analysis

Garzón-Guerrero, José A.; Ruíz, Diego P.; Carrión, M.C.

doi:10.1109/tap.2013.2266091

Cited by 18 publications

(13 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Sens and Spec are higher than 0.99 for all classes. For 10dB SNR, results start to decrease but are still high and similar to those obtained in [20] for the PEC sphere. However, accuracy respectively as they miss-classify all the samples of some classes as in fig 7.…”

supporting

confidence: 81%

Generalization Ability of Deep Learning Algorithms Trained using SEM Data for Objects Classification

Zaky¹,

fortino²,

Miramond³

et al. 2021

Preprint

View full text Add to dashboard Cite

This paper proposes a workflow to efficiently determine the material of spherical objects and the location of the receiving antenna relative to their position in bi-static measurements using supervised learning techniques. From a single observation, we compare classification performances resulting from the application of several classifiers on different data types: the Ultra-Wide Band scattered field in time and frequency domains and pre-processed data from the Singularity Expansion Method (SEM). Indeed, the resonances extracted using the SEM are aspect independent and therefore, are used to discriminate the objects. As for the residues, they depend upon the aspect angle and can hence be exploited to determine the observation angle. We construct 3 datasets to assess which one yields the highest accuracy while using the simplest and fastest classifiers. Hence, 80% of each dataset is used for training and the remaining 20% are used for testing. In a further step, we test with sphere sizes and data with several noisy levels that were not in the training datasets. Although SEM is noise sensitive, associating a robust feature extraction technique with a simple but reliable classifier is promising, particularly when generalizing to data not included in the training set.

show abstract

supporting

confidence: 81%

Generalization Ability of Deep Learning Algorithms Trained using SEM Data for Objects Classification

Zaky¹,

fortino²,

Miramond³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Time-shift invariance features, such as bispectra [ 15 ] and higher-order spectra [ 16 ], were applied to target classification. Since the high dimensionality of UWB echo data causes a big burden to the storage system and the classifier, some dimensional reduction techniques were proposed, such as linear discriminant analysis (LDA) [ 17 ] and principal component analysis (PCA) [ 18 ]. However, such methods based on one-dimensional signals can only make use of limited information, which decreases the identification performance.…”

Section: Introductionmentioning

confidence: 99%

Target Identification with Improved 2D-VMD for Carrier-Free UWB Radar

Zhu

Zhang

Zhao

et al. 2021

Sensors

View full text Add to dashboard Cite

In recent years, the interest in radar automatic target recognition (RATR) based on the carrier-free ultra-wideband (UWB) radar has been increasing. Compared with narrow-band and other bandwidth radars, the echo signal of the carrier-free UWB radar includes more comprehensive and detailed information with respect to the targeted object. In this paper, we first utilized 3ds Max to acquire accurate geometric models and applied a time-domain integral equation (TDIE) for echo signal acquisition under the condition that the transmitted signals had an extremely short duration period. By comparing the simulated waveform with the actual one, the accuracy of the electromagnetic modeling is verified. Furthermore, given that the actual environment is full of noise and clutter, we propose an improved two-dimensional variational mode decomposition (2D-IVMD), and an algorithm is proposed to eliminate noise and extract edge features preliminarily, which lays a foundation for further in-depth feature extraction. Then, the deep conventional neural network (DCNN) is introduced for the final recognition. The results show that the proposed methods achieve promising classification performance under the condition of low signal-to-noise ratio (SNR) values.

show abstract

“…The HRRP spectrum features are time‐shift invariant, but they have high dimensions and bring the heavy computational load to the real RATR systems. Feature dimension reduction methods, such as principal component analysis [13], linear discriminant analysis [7], maximum scatter difference [14], adaptive neighbourhood‐preserving discriminant projection [8] and so on, are used to project the high‐dimensional HRRP features into low‐dimensional subspace. The projection features have no physical meanings and only represent the global property of the target.…”

Section: Introductionmentioning

confidence: 99%

Radar HRRP target recognition via statistics‐based scattering centre set registration

Wang

Liu

et al. 2019

IET Radar, Sonar & Navigation

View full text Add to dashboard Cite

Classification of Geometrical Targets Using Natural Resonances and Principal Components Analysis

Cited by 18 publications

References 8 publications

Generalization Ability of Deep Learning Algorithms Trained using SEM Data for Objects Classification

Generalization Ability of Deep Learning Algorithms Trained using SEM Data for Objects Classification

Target Identification with Improved 2D-VMD for Carrier-Free UWB Radar

Radar HRRP target recognition via statistics‐based scattering centre set registration

Contact Info

Product

Resources

About