A5038759055- Remote-Sensing Image Classification
- Advanced Image Fusion Techniques
- Remote Sensing and Land Use
- Speech and Audio Processing
- Image Enhancement Techniques
- Fuzzy Logic and Control Systems
- Face recognition and analysis
- Advanced Measurement and Detection Methods
- Advanced Steganography and Watermarking Techniques
- Music and Audio Processing
- Diverse Musicological Studies
- Infrared Target Detection Methodologies
- Video Surveillance and Tracking Methods
- Industrial Vision Systems and Defect Detection
- Biometric Identification and Security
- Advanced Algorithms and Applications
- Industrial Technology and Control Systems
Xi’an University of Posts and Telecommunications
2021-2024
Wuhan Institute of Technology
2021
University of Science and Technology Liaoning
2012
Hyperspectral image (HSI) can provide rich spectral information which be helpful for accurate classification in many applications. Yet, incorporating spatial the process improve accuracy even further. Existing convolutional neural network (CNN) usually only focuses on local features hyperspectral cubes, whereas burgeoning vision transformer (ViT) is interested global HSIs. In this letter, we propose a deep aggregated framework HSI called convolution mixer (CTMixer) to combine advantages of...
Convolutional neural networks (CNNs) have attained remarkable performance in hyperspectral image (HSI) classification owing to excellent locally modeling ability. However, the existing CNNs cannot capture global context information from HSI. Recently, vision transformer (ViT) has been proven be effective field. its retrieval of local space HSI is not satisfactory, and input mode always leads loss spatial location information. In this letter, we propose a novel convolution fusion splicing...
Recently, with the extensive application of deep learning techniques in hyperspectral image (HSI) field, particularly convolutional neural network (CNN), research HSI classification has stepped into a new stage. To avoid problem that receptive field naive convolution is small, dilated introduced classification. However, usually generates blind spots resulting discontinuous spatial information obtained. In order to solve above problem, densely connected pyramidal (PDCNet) proposed this paper....
Recently, convolutional neural networks (CNNs) show excellent performance on the hyperspectral image (HSI) classification tasks. However, traditional CNNs usually have insufficient feature discrimination and a large number of network parameters. In response to above problems, residual dense asymmetric (RDACN) for HSI is proposed in this paper. Firstly, we de-sign novel block effectively leverage information previous layers. Moreover, adopts two fusion methods addition channel stacking...
Deep learning algorithms have shown significant advantages in hyperspectral image (HSI) classification. However, these usually require a large number of labeled samples and the annotation consumes massive time resource costs. To achieve effective classification results situations with small samples, semi-supervised co-training model using convolution transformer (SCM-CT) is proposed this letter. Firstly, two different networks, namely multi-scale parallel CNN (MPCNN) global local fusion...
Aiming at the poor recognition performance and low security of single-modal biometric template protection method, a cancelable method based on fingerprint iris feature fusion is proposed. This performs PCA concatenation processing features, then obtains fused vector through random scrambling, wavelet transform DFT operations. Finally, multi-modal generated partial Hadamard transform. The contains two vectors fingerprints finger veins, which has better distinguishability security. Theoretical...
Work in the field of industrial production equipment machine is a mechanical noise used to diagnosis important feature working state dimension. Early when abnormal running condition equipment, often accompanied by audio signal. But signal with normal generated class spacing small, which leads failure features difficult locate. In order effectively solve this problem, paper puts forward kind engineering and convolutional neural network model based on measure analysis method. First all,...