- Sparse and Compressive Sensing Techniques
- Image and Signal Denoising Methods
- Biometric Identification and Security
- Advanced Image Fusion Techniques
- Advanced Vision and Imaging
- Advanced Measurement and Detection Methods
- Optical measurement and interference techniques
- Chaos-based Image/Signal Encryption
- Advanced Optical Sensing Technologies
- Infrared Target Detection Methodologies
- Computer Graphics and Visualization Techniques
- 3D Surveying and Cultural Heritage
- Fire Detection and Safety Systems
- Video Surveillance and Tracking Methods
- Image Processing Techniques and Applications
- Advanced Sensor and Control Systems
- Advanced materials and composites
- Optical Systems and Laser Technology
- Advanced ceramic materials synthesis
- User Authentication and Security Systems
- Color Science and Applications
- Non-Destructive Testing Techniques
- Handwritten Text Recognition Techniques
- Face recognition and analysis
- Ferroelectric and Negative Capacitance Devices
Liaoning University
2024
Zhengzhou University of Aeronautics
2008-2024
University of Chinese Academy of Sciences
2012-2020
Shenzhen Institutes of Advanced Technology
2018-2020
Chinese Academy of Sciences
2011-2020
Institute of Automation
2014
Institute of Coal Chemistry
2012
Naval University of Engineering
2010
Tsinghua University
2002
Compared with existing depth cameras, such as RGB-D, RealSense and Kinect, stripe-based structured light (SL) has the potential for micrometer-level 3D measurement; this can be attributed to its higher coding capacity. While surface texture, high-reflective region, occlusion remain some of main sources leading degraded reconstruction quality in complex objects, methods that are only based on SL cannot completely solve problems object reconstruction. In paper, we developed an advanced fusion...
Large language models (LLMs) with hundreds of billions parameters have sparked a new wave exciting AI applications. However, they are computationally expensive at inference time. Sparsity is natural approach to reduce this cost, but existing methods either require costly retraining, forgo LLM's in-context learning ability, or do not yield wall-clock time speedup on modern hardware. We hypothesize that contextual sparsity, which small, input-dependent sets attention heads and MLP...
GPUs are high performance co-processors of CPU for scientific computing including CFD. We present an optimistic shared memory allocation strategy to solve 2D CFD problems using Red-Black SOR method on GPU with CUDA (Compute Unified Device Architecture). Lid-driven results compared the benchmark data. The speed up ratio same problem size by NVDIA GTX480 and Intel Core-Dual 3.0GHz processor is discussed, 120 times faster than sequential code 756756. Based this work, we conclude that hierarchy...
This paper presents a GPU computing algorithm, used to accelerate the Continuous-based Discrete Element Method (CDEM). Using NVIDIA GTX VGA card, speed achieved an average 650 times speedup ratio vs. Intel Core-Dual 2.66 GHz CPU. To parallelize CDEM clone node force refreshing process is separated from elemental calculation, and replaced by “Node Group” assignment process, which ensures data independence in parallel execution.
In this paper, we propose a method for joint 2D segmentation and 2D-3D pose tracking. First, define novel energy functional which considers the discrimination between statistical appearance models coherence among neighboring pixels simultaneously. And then, particle filter-like stochastic optimization technique is adopted to solve functional, so that preferable initial value can be provided subsequent damped Newton method. Furthermore, an occlusion-aware updating strategy utilized models,...
Timely and accurate monitoring of typical coastal targets using remote sensing technology is crucial for maintaining marine ecological stability. Hyperspectral target detection proves to be an effective tool in extracting various materials along the coastline. Traditional methods spectral domain information can effectively retain intrinsic properties material. However, it difficult recognize homogeneous regions by only information, which may lead insufficient utilization spatial information....
In recent years, the chaos based cryptographic algorithms have suggested some new and efficient way to develop secure fingerprint image encryption techniques. this paper, a novel chaotic images scheme is proposed combining with shuttle operation nonlinear dynamic system. Firstly, total shuffling employed shuffle positions of pixels in spatial-domain. Then discrete output signal system preprocessed confuse relationship between shuttled cipher-image. The results several experimental,...
In this paper, a novel image denosing scheme is proposed by applying 2D dual-tree complex wavelet transform (DTCWT) to the second bandelet transform. Compared with traditional discrete (DWT), DTCWT has nearly shift invariant and directionally selective in two higher dimensions important properties, which are suitable for denoising. The offer an asymptotically optimal representation geometric images, but it lacks translation invariant. Firstly, employed obtain subands, then bandeletization...
Background There is a consensus that both nonsuicidal self-injury (NSSI) and suicidal ideation as risk factors for behavior have strong connection. However, lack of longitudinal information has limited the clarification concrete relationship between them. Aims This study aimed to examine specific mechanism NSSI over time, during adolescence. Method A was conducted with 193 Chinese adolescents. ideations were examined course 1year followed-up, three waves data collected. Results The at time...
In this work, lossless grayscale image compression methods are compared on a public palmprint database. The total number of test images was about 7752, with 8 bit rates. performance using different algorithms the ratios when processing sample data is investigated, in particular, we relate application CALIC, JPEG-LS,RAR, JPEG2000 (lossless version), PNG (Portable Network Graphics) LJPEG JPEG), JPEG (Binary DCT) and S+P transform. Based testing results an open database, CALIC gives high...
In this paper, we propose a novel palmprint verification scheme named as Orthogonal Code based on multiple 2D Gabor Phase feature. A image contains various features with orientations, so only single orientation in the can not distinguish subject well. The filters different orientations are employed to extract texture information and phase-coding is used represent palmprint. At matching stage, Hamming distance similarity metric. experimental results of Hong Kong Polytechnic University...
A particle filter object video tracking algorithm based on dynamic feature fusion is proposed in this paper. This uses the complementary features, which are gray histogram and gradient histogram, to represent model. In process of tracking, confidence for each adjusted according discrimination between background, model dynamically established updated online. The presented method can improve accuracy modeling furthermore algorithm. Experimental results have demonstrated effectiveness our...
Broadband wireless access (BWA) is a novel technology in broadband communications. The structure, high rates, and flexibility make it appropriate for telephone, videoconference, Ethernet, VoD multimedia services. And promising among its competitors. key technologies MAC of BWA system are discussed this article. We also introduce the implementation 3.5 GHz system. apply TDMA/TDM/FDD mode control.
The acquisition of the geometry general scenes is related to interplay surface geometry, material properties and illumination characteristics. Surface texture non-Lambertian reflectance degrade reconstruction results by structured light technique. Existing techniques focus on different coding strategy sources improve accuracy. hybrid system consisting a technique photometric stereo combines depth value with normal information refine results. In this paper, we propose novel stripe-based...
This paper presents a chaos-based renewable and privacy preserving binary palmprint phase template for cancellable recognition. In the proposed approach, novel 2D orthogonal Log-Gabor filter is firstly constructed by two filters with orientations to extract biometric template. And then templates are encrypted chaotic stream ciphers encrypt run in integer domain generate templates. During matching stage, Hamming distance between employed measure similarity of different palmprints. The...