Synthetic Aperture Radar SAR Image Target Recognition Algorithm Based on Attention Mechanism

Shi, Baodai; Zhang, Qin; Dayan, Wang; Yao, Li

doi:10.1109/access.2021.3118034

Cited by 11 publications

(8 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Then mixed attention is added to gradually extract advanced features. On the MSTAR dataset, compared to advanced algorithms, it can significantly reduce the parameters and complexity [16]. Du C et al proposed a practical deceptive interference method based on fragile position perception for adversarial attacks.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Target Recognition of Multi Source Machine Vision Pan Tilt Integrated Inspection Robot for Power Inspection

Zhu,

Zhang,

Zhou

et al. 2024

IEEE Access

View full text Add to dashboard Cite

With the continuous changes in socio-economic needs, traditional methods of power facility inspection can no longer meet practical needs due to their low efficiency and lack of scalability. In response to this challenge, this study delves into the integrated motion control technology of inspection robots equipped with gimbal mechanisms, aiming to improve the convenience and efficiency of dynamic data collection. A customized multi-source heterogeneous visual detection and recognition model based on the YOLOv3 framework has been proposed, and simultaneously using path aggregation networks to enhance information processing capacity by fusing multi-scale features. Experimental analysis shows that as the robot's movement speed increases, the error rate correspondingly increases, indicating the direction of optimization. In the target recognition experiment, the proposed model achieved an average accuracy of 94.26% in visible light images and 68.05% in infrared images. In addition, Sub_ The YOLO algorithm demonstrates a fast detection speed of 30 frames per second, with an average accuracy of over 80%, marking an important progress in real-time object detection applications. In the linear motion test, the relative error of the robot's motion accuracy was 0.33% at a speed of 500 millimeters per second. However, when the speed was increased to 1200 millimeters per second, the error increased to 2.45%, indicating a significant increase in slip. This indicates that the linear motion accuracy of the robot is acceptable at low to medium speeds, but the accuracy decreases significantly at high speeds.Overall, the research results confirm the synergistic effect of integrated motion control between inspection robots and gimbals, as well as Sub_ The superiority of YOLO in target recognition has improved the ability to use wheeled robots for electrical inspections, bringing substantial technological progress to the field of autonomous inspection.

show abstract

Section: Related Workmentioning

confidence: 99%

“…In equation ( 15), a and v introduce aspect ratios to optimize the difference in width and height between the predicted box and the actual box. The corresponding loss function is shown in equation (16).…”

Section: B Multi Source Heterogeneous Machine Vision Detection and Ob...mentioning

confidence: 99%

Target Recognition of Multi Source Machine Vision Pan Tilt Integrated Inspection Robot for Power Inspection

Zhu,

Zhang,

Zhou

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Figure 7a demonstrates that the backbone of the encoder part is based on a deep residual shrinkage network (DRSN), which has been demonstrated to be applicable to the process of highly noised images by several studies [56][57][58][59]. The DRSN in this study was rebuilt from the ResNet-18.…”

Section: Proposed Drs-unetmentioning

confidence: 99%

DRs-UNet: A Deep Semantic Segmentation Network for the Recognition of Active Landslides from InSAR Imagery in the Three Rivers Region of the Qinghai–Tibet Plateau

et al. 2022

View full text Add to dashboard Cite

At present, Synthetic Aperture Radar Interferometry (InSAR) has been an important technique for active landslides recognition in the geological survey field. However, the traditional interpretation method through human–computer interaction highly relies on expert experience, which is time-consuming and subjective. To solve the problem, this study designed an end-to-end semantic segmentation network, called deep residual shrinkage U-Net (DRs-UNet), to automatically extract potential active landslides in InSAR imagery. The proposed model was inspired by the structure of U-Net and adopted a residual shrinkage building unit (RSBU) as the feature extraction block in its encoder part. The method of this study has three main advantages: (1) The RSBU in the encoder part incorporated with soft thresholding can reduce the influence of noise from InSAR images. (2) The residual connection of the RSBU makes the training of the network easier and accelerates the convergency process. (3) The feature fusion of the corresponding layers between the encoder and decoder effectively improves the classification accuracy. Two widely used networks, U-Net and SegNet, were trained under the same experiment environment to compare with the proposed method. The experiment results in the test set show that our method achieved the best performance; specifically, the F1 score is 1.48% and 4.1% higher than U-Net and SegNet, which indicates a better balance between precision and recall. Additionally, our method has the best IoU score of over 90%. Furthermore, we applied our network to a test area located in Zhongxinrong County along Jinsha River where landslides are highly evolved. The quantitative evaluation results prove that our method is effective for the automatic recognition of potential active landslide hazards from InSAR imagery.

show abstract

“…Due to the powerful working ability under various weather conditions, including rain, snow, cloud, and frog, synthetic aperture radar (SAR) (Chen et al., 2020; Wang et al., 2020; Zhou et al., 2019) has been widely used in both of military and civilian fields. As one of the most popular application fields of SAR, SAR target recognition (Shi et al., 2021; Wang et al., 2021; Wen et al., 2019) has attracted increasing popularity nowadays and demonstrated great vitality in environmental monitoring, disaster relief, surveillance and reconnaissance areas. Plenty of new techniques have been adopted to realize SAR target recognition (Chen et al., 2020; Wang et al., 2020, 2021).…”

Section: Introductionmentioning

confidence: 99%

Multi‐Manifold Based Sparse Representation With Application to SAR Target Configuration Recognition

Liu

Chen

2022

Radio Science

View full text Add to dashboard Cite

A multi‐manifold based sparse representation (MMSR) algorithm which can preserve the local structure of the datasets in the reconstruction space with multiple descriptions is proposed for synthetic aperture radar (SAR) target configuration recognition. Different from the traditional manifold learning‐based algorithms that preserve the local structure of the datasets in the low‐dimensional space, the proposed MMSR guarantees local structure preserving of the datasets after feature extraction and sparse representation. The proposed MMSR can avoid corrupted relationships of the samples to ease the obstacle of target aspect angle sensitivity, which can realize more accurate SAR target configuration recognition. Satisfying results are obtained on the moving and stationary target acquisition and recognition database.

show abstract

Synthetic Aperture Radar SAR Image Target Recognition Algorithm Based on Attention Mechanism

Cited by 11 publications

References 37 publications

Target Recognition of Multi Source Machine Vision Pan Tilt Integrated Inspection Robot for Power Inspection

Target Recognition of Multi Source Machine Vision Pan Tilt Integrated Inspection Robot for Power Inspection

DRs-UNet: A Deep Semantic Segmentation Network for the Recognition of Active Landslides from InSAR Imagery in the Three Rivers Region of the Qinghai–Tibet Plateau

Multi‐Manifold Based Sparse Representation With Application to SAR Target Configuration Recognition

Contact Info

Product

Resources

About