A5089287034- Advanced Vision and Imaging
- Advanced Algorithms and Applications
- Advanced Fiber Optic Sensors
- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Anomaly Detection Techniques and Applications
- Advanced Measurement and Detection Methods
- Video Surveillance and Tracking Methods
- Image and Video Quality Assessment
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Advanced Image Fusion Techniques
- Human Pose and Action Recognition
- Color Science and Applications
- Image and Object Detection Techniques
- Advanced Fiber Laser Technologies
- Advanced Sensor and Control Systems
- Soil Carbon and Nitrogen Dynamics
- Water Systems and Optimization
- Optical measurement and interference techniques
- Grey System Theory Applications
- Simulation and Modeling Applications
- Image and Video Stabilization
- Optical Coherence Tomography Applications
- Advanced Chemical Sensor Technologies
Zhengzhou University
2024
Northeast Petroleum University
2023
Sun Yat-sen University
2022
Tampere University
2017-2022
Southwest University
2022
Shanghai University of Engineering Science
2020-2021
Tampere University of Applied Sciences
2020
NARI Group (China)
2019
Beijing University of Posts and Telecommunications
2019
Hangzhou Dianzi University
2019
Live video streaming delivery over Dynamic Adaptive Video Streaming (DASH) is challenging as it requires low end-to-end latency, more prone to stall, and the receiver has decide online which representation at bitrate download whether adjust playback speed control latency. To encourage research community come together address this challenge, we organize Grand Challenge ACM Multimedia 2019. This grand challenge provides a simulation platform onto participants can implement their adaptive (ABR)...
Motion deblurring in dynamic scenes is a challenging task when the blurring caused by one or combination of various reasons such as moving objects, camera movement, etc. Since event cameras can detect changes intensity with low latency, necessary motion information inherently captured data, which could be quite useful for standard images. The degradation does not show homogeneity across an image due to factors like object depth, speed, We propose two-branch network structure, Aware Double...
Recent works on no-reference image quality assessment (NR-IQA) have reported good performance for various datasets. However, they suffer from significant drops in cross-dataset evaluations which indicates poor generalization power. We propose a Siamese architecture and training procedures deep NR-IQA that achieves clearly better performance. Moreover, we show the can be further boosted by i) pre-training with large aesthetics dataset ii) adding low-level cues, sharpness, tone colourfulness,...
We propose a reverse camera imaging pipeline to convert arbitrary images raw RGB responses of specific camera. The requires only that the camera's are characterized. reversed helps developers generate and use them train learning-based algorithms. In our experiments, three recent deep color constancy architectures achieve superior results in cross-dataset setting using generated images.
The optical fiber pre-warning system (OFPS) has been gradually considered as one of the important means for pipeline safety monitoring. Intrusion signal types are correctly identified which could reduce cost troubleshooting and maintenance pipeline. Most previous feature extraction methods in OFPS usually quested from view time domain. However, some cases, there is no distinguishing In paper, firstly, intrusion features running, digging, pick mattock extracted frequency domain by multi-level...
We propose a novel hierarchical sliding slice regression which in coarse-to-fine manner represents global circular target space with number of ordinally localized and overlapping subspaces. Our method is particularly suitable for visual problems where the (e.g., car viewing angle) similarity inconsistent over repetitive appearance). A good application example camera-based angle estimation problem, different views highly inconsistent-front back left right side are pair-wise similar, but...
Recent semantic segmentation networks mainly focus on how to fuse multi-level features from classification improve accuracy. Some researches evenly emphasize the correlation of pixels in a global region, such as conditional random field (CRF). However, strong feature limited region is less considered previous and remedy ability local severely ignored. To deal with this problem, we introduce hierarchical adjacency dependent network (HadNet), which an dependency module (ADM) constructed by...
As the medium of human-computer interaction, it is crucial to correctly and quickly interpret motion information surface electromyography (sEMG). Deep learning can recognize a variety sEMG actions by end-to-end training. However, most existing deep approaches have complex structures numerous parameters, which make network optimization problem difficult realize. In this paper, novel PSO-based optimized lightweight convolution neural (PLCNN) designed improve accuracy optimize model with...
Smartphones are widely used in daily life to access services and various functions require continuous communication, which leads increased energy consumption. However, the development of battery related saving technology can not meet demand for Much current research work focuses on models caring much about consumption every single application. In this paper, we propose a data forecasting-based strategy increasing efficiency smartphones based predictability be accessed. To achieve this,...
With the continuous development of optical fiber sensing technology, Optical Fiber Pre-Warning System (OFPS) has been widely used in various fields. The OFPS identifies type intrusion based on detected vibration signal to monitor surrounding environment. Aiming at real-time requirements OFPS, this paper presents a fast algorithm accelerate detection and recognition processing signals. is implemented an embedded system that composed digital processor (DSP). flow divided into two parts. First,...
针对Harris检测出的角点位置会发生偏移和易产生伪角点，以及在角点匹配过程中计算复杂，容易产生误匹配等缺点，该文提出了基于良分布的亚像素定位角点的图像配准方法。该方法首先使用多尺度Harris算子检测图像的角点作为初始兴趣点，并采用自适应非极大值抑制对兴趣点的数量进行限制，以减少后续过程的计算复杂度，提高算法效率，同时使得角点在图像中处于良分布状态。然后利用亚像素定位技术进行精确定位，排除伪角点和不稳定的角点。最后使用随机抽样一致性算法对初始匹配进行鲁棒的模型参数估计。实验结果表明算法配准效率改进明显，并具有良好的精确性和鲁棒性。
The descriptive capability of color images for a scene is better than that grayscale images. Traditional registration methods to register convert them ones, in which the precision decreased due loss information. By extending some properties discrete Fourier transform quaternion domain, discrete-quaternion-Fourier-transform-based method proposed this paper. Color are represented form, and parameters rotation-scale-translation model recovered using rotation scale DQFT phase correlation....
Remote sensing images are often affected by severe stripe noise, which degrades image quality. In this paper, a destriping method is proposed with the multi-scale variation model on basis of noise analysis. First, established. Then, energy function constructed unidirectional characteristics noise. Last, optimized split Bregman approach, and destriped obtained accumulation component in each scale. Experimental results MODIS level 1B data show that quality quantitative evaluation index can be...
A new adaptive variant of Artificial Bee Colony algorithm, PAABC, is proposed to improve optimization performance and enhance the robustness ABC algorithm by incorporating into searching strategy updating control parameter equation adaptively . The incorporation information helpful accelerate convergence while avoiding prematurity especially multimodal problem. According characteristic problem, will be update adaptively. better value associated with mean p% best individual survive next...
Abstract Optical fiber pre-warning system (OFPS) is often used to monitor the occurrence of disasters such as leakage oil and natural gas pipeline. It analyzes collected vibration signals judge whether there any harmful intrusion (HI) events. At present, research in this field mainly focused on constant false alarm rate (CFAR) methods derivative algorithms detect signals. However, performance CFAR limited actual distribution. found that background noise usually obeys non-independent...
A number of spatially-localised semantic parts vehicles sensitive to pose changes are encoded their visible probabilities into a mid-level feature vector. Car estimation is then formulated regression on concatenated low- and mid- level features continuously changing viewing angles. Each dimension our visibility-aware part codes separates all the training samples two groups according its visual existence in images, which provides additional part-specific range constraint Moreover, proposed...
It is vital to evaluate the effects of China’s farmland certification programme on farmers’ economic welfare and clarify impact mechanism. This paper uses China Labour Dynamics Survey (2014, 2016), 1 building endogenous switching regression probit models test net agricultural income poverty status after correcting for endogeneity. A robustness analysis conducted applying instrumental variable propensity score matching methods. The results indicate that can significantly increase farmer...