Automatic target detection and recognition of military vehicles in synthetic aperture radar images is fostered by optimizing VGG-googLeNet with the giraffe kicking optimization algorithm

• Published in: Signal, image and video processing Vol. 18; no. 8-9; pp. 6491 - 6502
• Main Authors: Shakin Banu, A., Shahul Hameed, K. A.
• Format: Journal Article
• Language: English
• Published: London Springer London 01.09.2024; Springer Nature B.V
• Subjects: Algorithms; Automatic target recognition; Automatic vehicle identification systems; Clustering; Computer Imaging; Computer Science; Decision trees; Flight planning; Fuzzy sets; Gaussian elimination; Gaussian process; Image acquisition; Image filters; Image Processing and Computer Vision; Military vehicles; Multimedia Information Systems; Optimization; Optimization algorithms; Optimization techniques; Original Paper; Parameters; Pattern Recognition and Graphics; Performance measurement; Radar detection; Radar imaging; Signal,Image and Speech Processing; Synthetic aperture radar; Target acquisition; Target detection; Vision; Giraffe kicking optimization algorithm; Military and civilian vehicles classification dataset; Hesitant fuzzy linguistic bi-objective clustering; Adaptive multiscale gaussian co-occurrence filtering; Synthetic aperture radar
• ISSN: 1863-1703; 1863-1711
• DOI: 10.1007/s11760-024-03332-9

Artificial Aperture Radar (SAR) Multiview pictures have the ability for providing far more detailed information than single-view images for automatic target recognition. Acquiring a series of SAR images from suitable viewpoints and determining the best SAR platform flight routes are required to perform Multiview SAR ATR precisely and effectively. In this paper, Automatic Target Detection and Recognition of Military Vehicles in Synthetic Aperture Radar Images is fostered by optimizing VGG-GoogLeNet with the Giraffe Kicking Optimization Algorithm (VGG-GNet-TDMV-SAR) is proposed. Initially, SAR images are taken from the Military and Civilian Vehicles Classification dataset. Then the input SAR image is pre-processing using Adaptive Multi scale Gaussian Co-Occurrence Filtering for eliminating speckle noise present in the SAR imageries. The Hesitant Fuzzy Linguistic Bi-objective Clustering technique is utilized to extract Region of Interest after pre-processing. The extracted SAR image is given to Target Detection using Integrated Fuzzy Decision Tree. After that, VGG-Google Net (VGG-GNet) classifier classifies the military and non-military vehicle. In general, VGG-GNet does not adapt any optimization techniques to define the optimal parameter and to ensure the precise detection. So, Giraffe Kicking Optimization Algorithm is proposed to optimize VGG-GNet classifier’s weight parameters. The proposed technique is implemented using MATLAB and analyzed with performance metrics. The proposed technique achieves greater accuracy, lower computation time, greater ROC compared with existing techniques.