Abstract
SAR images contain a large amount of noise, and related algorithms will cause high complexity when increasing the accuracy. To overcome this problem, a neural network model based on the attention mechanism was proposed in this paper. The model extracted information in two stages. It gradually extracts high-level features by reducing noise first and then adding hybrid attention. First, use dual-channel one-dimensional convolution to reconstruct the residual shrinkage network to construct a lightweight and efficient feature module, which improved the information extraction of the module with the consumption of a small amount of computing resources. Then, it was used as the backbone for model construction. Subsequently, mixed adaptive pooling was adopted to improve the maximum pooling. After that, dimensionality was reduced by pooling and linear interpolation was used to increase dimensionality, so as to generate feature weights of mixed dimension. Tests were performed on MSTAR dataset. The results showed that compared with the advanced algorithms, the proposed model in this paper can greatly reduce the amount of parameters and complexity while ensuring accuracy. The robustness test demonstrated that the model can effectively identify images with noise being added.
Highlights
As mentioned in [1,2,3,4], synthetic aperture radar (SAR) is a kind of active microwave imaging radar, and it has been extensively applied to military and civil fields for the advantages of full-time and all-weather work
The demand for military reconnaissance has stimulated SAR image automatic target recognition (ATR) technology, that is generally divided into three stages: image preprocessing, feature extraction, and target classification and recognition
The soft thresholding module is the core of the model, and its structure is shown in Figure 1, and the soft thresholding module can be divided into two steps here.Threshold generation and threshold screening, the threshold generation is accomplished by global average pooling (GAP) and two consecutive fully connected layers
Summary
As mentioned in [1,2,3,4], synthetic aperture radar (SAR) is a kind of active microwave imaging radar, and it has been extensively applied to military and civil fields for the advantages of full-time and all-weather work. A two channel adaptive onedimensional convolution method is proposed to avoid dimensionality reduction and only conduct an appropriate amount of channel interaction, and the improved module is named S-DRSN This method improves the information transmission efficiency of the module while only consuming a small amount of parameters. 2) Aiming at the problem that the weights generated in the hybrid attention mechanism proposed by Wang et al are not accurate enough, an adaptive hybrid pooling method is proposed to improve the feature representation ability of down-sampling. This method takes into account both background information and texture information, and improves the accuracy of mask branches. The model has strong resistance to random noise and salt and pepper noise
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
