A5090982546- Advanced Data Compression Techniques
- Domain Adaptation and Few-Shot Learning
- Sparse and Compressive Sensing Techniques
- Wireless Signal Modulation Classification
- Advanced Data Storage Technologies
- Advanced Steganography and Watermarking Techniques
- Caching and Content Delivery
- Advanced Vision and Imaging
- Machine Learning and ELM
- Energy Efficient Wireless Sensor Networks
- Vehicle License Plate Recognition
- Advanced MIMO Systems Optimization
- Context-Aware Activity Recognition Systems
- Image and Video Stabilization
- Advanced Neural Network Applications
- Chaos-based Image/Signal Encryption
- Advanced Measurement and Detection Methods
- Error Correcting Code Techniques
- Animal Vocal Communication and Behavior
- Advanced Image Processing Techniques
- Service-Oriented Architecture and Web Services
- Image Enhancement Techniques
- Video Coding and Compression Technologies
- Image Retrieval and Classification Techniques
- Network Security and Intrusion Detection
Xidian University
2014-2025
Wireless communication has achieved great success in the past several decades. The challenge is of improving bandwidth with limited spectrum and power consumption, which however gradually become a bottleneck evolution going on. intrinsic problem that modeled as message transportation from sender to receiver pursues for an exact replication Shannon's information theory, certainly leads large requirements data explosion. However, goal among intelligent agents, entities intelligence including...
Loop filters are used in video coding to remove artifacts or improve performance. Recent advances deploying convolutional neural network (CNN) replace traditional loop show large gains but with problems for practical application. First, different model is frames encoded quantization parameter (QP), respectively. It expensive hardware. Second, float points operation CNN leads inconsistency between encoding and decoding across platforms. Third, redundancy within consumes precious computational...
Large pre-trained vision-language models (VLMs) like CLIP have shown great potential for solving the unsupervised domain adaptation (UDA) problem. Existing prompt learning UDA based on unsupervised-trained VLMs requires distribution alignment between source and target domains in common space both vision language branches. However, it is difficult rough cross-domain to maintain discriminative semantic structure of domains. Besides, coarse features with non-informative noises due ignoring...
We consider efficient image transmission via time-varying channels. To improve the performance, we propose a new distributed compressive sensing (CS) scheme that can leverage similar images in cloud. It is featured by channel SNR and bandwidth scalability, high efficiency, low encoding complexity. For each image, compressed thumbnail first transmitted after forward error correction (FEC) modulation to retrieve generate side information (SI) The residual subtracting decompressed then coded CS...
With multimedia flourishing on the Web, it is easy to find similar images for a query, especially landmark images. Traditional image coding, such as JPEG, cannot exploit correlations with external Existing vision-based approaches are able by reconstructing from local descriptors but ensure pixel-level fidelity of reconstruction. In this paper, cloud-based distributed coding (Cloud-DIC) scheme proposed mobile photo uploading. For each input image, thumbnail transmitted retrieve correlated and...
In recent years, semantic communication based on deep learning for source-channel joint encoding has garnered significant attention. It utilizes network models trained end-to-end to represent signals as embedding vectors and demonstrated superior performance compared traditional methods. However, due the disparity between human language, it can be challenging succinctly capture abstract semantics such scenes. this paper, we introduce Scene Graph-based Generative Semantic Communication...
This paper proposes a cloud-based distributed image coding scheme (Cloud-DIC) to exploit the strong correlations with external partial-duplicate images in cloud. It features both high efficiency and low encoder complexity, which makes it suitable for photo sharing on mobile devices. To get side information cloud, thumbnail of current is transmitted retrieve highly correlated reconstruct through geometrical registration adaptive patched-based stitching. The then compressed by transform-domain...
Redundancy is necessary for a storage system to recover from errors. The frequent errors in large-scale systems, e.g. cloud, make it desired reduce the recovery cost. Among all kinds of data stored video takes large portion due its volume. other characteristic that certain distortion can be tolerated. This paper investigates using scalable representation and unequal error protection scheme costs cloud. By introducing more base layer less on enhancement layers, achieve better tradeoff between...
This paper proposes a distributed compressive sensing (CS) scheme for robust image transmission over unknown or time-varying channels with highly correlated images at the decoder. A compressed thumbnail is first transmitted after digital forward error correction (FEC) and modulation to retrieve generate side information (SI) The current residual subtracting decompressed then coded by CS through very dense constellation without FEC. linear representation of signal measurements rateless...
Loop filters are used in video coding to remove artifacts or improve performance. Recent advances deploying convolutional neural network (CNN) replace traditional loop show large gains but with problems for practical application. First, different model is frames encoded quantization parameter (QP), respectively. It expensive hardware. Second, float points operation CNN leads inconsistency between encoding and decoding across platforms. Third, redundancy within consumes precious computational...
Redundancy is necessary for a storage system to achieve reliability. Frequent errors in large-scale systems, example, cloud, make it desirable reduce the cost of recovery. Among all types data cloud storage, videos generally occupy significant amounts space due high volumes and rapid development video sharing video-on-demand services. Unlike general data, can tolerate certain level quality degradation. This paper investigates multilayer representations, such as scalable simulcast streaming,...
Monitoring multimodal signals provides a more comprehensive understanding of health conditions compared to singlemode monitoring. In the face significant volumes signals, existing IoT monitoring systems primarily focus on high-fidelity signal transmission by encoding separately. However, due lack consideration for downstream applications and correlation between portion bandwidth resources is wasted task-irrelevant information intermodal redundancy. To address this issue, we propose...
Abstract Since residual learning was proposed, identity mapping has been widely utilized in various neural networks. The method enables information transfer without any attenuation, which plays a significant role training deeper However, interference with unhindered transmission also affects the network’s performance. Accordingly, we propose generalized architecture called reverse attention (RA), applies high-level semantic features to supervise low-level branch. It means that higher...
In order to achieve multiple video images fusion processing, domestic and foreign technology commonly used DSP image algorithm, limited serial instruction execution speed, it was difficult meet the requirements of modern high-speed digital signal processing.An based on decoder chip ADV7180 FPGA hardware platform is presented, with analog input sampled by ADV7180, utilization parallel processing capabilities, can be implemented.The test results show that this method simple flexible, able...