Integrating deep learning-based data driven and model-based approaches for inverse synthetic aperture radar target recognition

Theagarajan, Rajkumar; Bhanu, Bir; Erpek, Tugba; Hue, Yik-Kiong; Schwieterman, Robert; Davaslıoğlu, Kemal; Shi, Yi; Sagduyu, Yalin E.

doi:10.1117/1.oe.59.5.051407

Cited by 10 publications

(8 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The scale function can be replaced by other template operators, which may generate more accurate weight maps. Besides, we took some inspiration from [41], which demonstrates a deep learning approach and a model based approach (using scattering centers) for classification, and also provides various fusion techniques. In [41], Theagarajan et al use segmentation and local maxima to extract the scattering centers.…”

Section: Discussionmentioning

confidence: 99%

“…Besides, we took some inspiration from [41], which demonstrates a deep learning approach and a model based approach (using scattering centers) for classification, and also provides various fusion techniques. In [41], Theagarajan et al use segmentation and local maxima to extract the scattering centers. Due to its simple and practical operation, it is a worthwhile attempt to use this algorithm in deep learning.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

A Lightweight Detection Model for SAR Aircraft in a Complex Environment

Wen

Huang

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Recently, deep learning has been widely used in synthetic aperture radar (SAR) aircraft detection. However, the complex environment of the airport—consider the boarding bridges, for instance—greatly interferes with aircraft detection. Besides, the detection speed is also an important indicator in practical applications. To alleviate these problems, we propose a lightweight detection model (LDM), mainly including a reuse block (RB) and an information correction block (ICB) based on the Yolov3 framework. The RB module helps the neural network extract rich aircraft features by aggregating multi-layer information. While the RB module brings more effective information, there is also redundant and useless information aggregated by the reuse block, which is harmful to detection precision. Therefore, to accurately extract more aircraft features, we propose an ICB module combining scattering mechanism characteristics by extracting the gray features and enhancing spatial information, which helps suppress interference in a complex environment and redundant information. Finally, we conducted a series of experiments on the SAR aircraft detection dataset (SAR-ADD). The average precision was 0.6954, which is superior to the precision values achieved by other methods. In addition, the average detection time of LDM was only 6.38 ms, making it much faster than other methods.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

A Lightweight Detection Model for SAR Aircraft in a Complex Environment

Wen

Huang

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…subverts the development rules of traditional artificial intelligence systems, enabling computers to simulate the operation mode of the brain and learn and recognize abstract patterns through multilayer convolution neural networks, so as to solve some general pattern recognition problems [25][26][27], which means that any task involving a large amount of data may benefit from deep learning. Based on the idea of deep learning, this paper puts forward a technical idea of target recognition of college students' psychological stress indicators [28,29].…”

Section: Cnn Model Construction Deep Learning Technologymentioning

confidence: 99%

Identification and Modeling of College Students’ Psychological Stress Indicators for Deep Learning

Tian

2022

Scientific Programming

View full text Add to dashboard Cite

Aiming at the problems of low accuracy of recognition results, long recognition time, and easy interference in traditional methods, a deep learning-oriented recognition modeling method of college students' psychological stress indicators is proposed. First, the ECG signal is collected by the ECG signal acquisition system, and the wavelet transform method is used to denoise the collected ECG signal. Then, the sequential backward selection algorithm is used to select the features of psychological stress indicators to reduce the feature dimension. Finally, based on the convolutional neural network in deep learning technology, a mental pressure indicator recognition model is established and the model parameters are optimized to realize the recognition of college students’ mental pressure indicators. Experimental results show that the method in this paper has high recognition accuracy, has high recognition efficiency, is not susceptible to interference, and has certain feasibility and effectiveness.

show abstract

“…According to the existing literature, the fusion methods can be divided into three main strategies: early feature fusion (EFF), late feature fusion (LFF), and decision level fusion (DLF) [43]. Eff refers to fusion using low-level features calculated based on original data and LFF refers to fusion using higher-level features.…”

Section: A Related Workmentioning

confidence: 99%

“…After we get two one-dimensional probability distribution vectors from the local feature extraction branch and global feature extraction branch respectively, we need to fuse the two vectors to get the final recognition result. There are many fusion strategies such as early feature fusion (EFF), late feature fusion (LFF), and decision level fusion (DLF) [43] as mentioned in the previous introduction. The difference between these three strategies is that the fusion operation is located at different levels of the algorithm.…”

Section: Target Recognition Based On Dlfmentioning

confidence: 99%