A5045283922- Robotics and Sensor-Based Localization
- Indoor and Outdoor Localization Technologies
- Traffic Prediction and Management Techniques
- Target Tracking and Data Fusion in Sensor Networks
- Evacuation and Crowd Dynamics
- Hand Gesture Recognition Systems
- Advanced Vision and Imaging
- Video Surveillance and Tracking Methods
- Advanced Image and Video Retrieval Techniques
- Video Coding and Compression Technologies
- Robot Manipulation and Learning
- Image and Video Quality Assessment
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- Vascular Malformations and Hemangiomas
- UAV Applications and Optimization
- Vehicle License Plate Recognition
- Energy Efficient Wireless Sensor Networks
- Advanced Data Compression Techniques
- Safety and Risk Management
- 3D Surveying and Cultural Heritage
- Maritime Navigation and Safety
- Reinforcement Learning in Robotics
- Advanced Image Processing Techniques
- Image Enhancement Techniques
Jimei University
2017-2024
Xiamen University
2007-2011
There exists a class of complex tasks that can be distinctly separated into multiple concatenated simpler [1-2]. This and continuous motion processes then considered as sub-processes, switch brings these sub-processes together in accordance to an input function. paper validates the potential value this approach by proposing method train task through series sub-tasks, concatenating them switch. Through classic simulation physical OpenAI Gym[3]: cartpole, demonstrates viability approach. An...
Objective To investigate whether N6-methyladenine DNA(6-mA DNA) modification is related to the occurrence of infantile hemangiomas (IH) at epigenetic level. Methods The genomic 6-mA DNA data were obtained by MeDIP and high-throughput sequencing. The methylation levels in 3 proliferative hemangioma specimens adjacent skin tissues compared u-test. functional differences modified genes analyzed GO analysis. Results The level IH tissue was higher. coverage Peaks genome 0.037% tumor...
In traditional video surveillance systems, supervisors need to watch multiple images at the same time. Surveillance resources are scattered and huge, making it difficult browse them all extract useful information from in a timely manner. The splicing system can be used merge splice fuse different perspectives form whole image, which is convenient for personnel supervision. This article introduces multi-view system: consists of several cameras. Through GStreamer framework, streaming performed...
Indoor growing is a trend in agriculture the future. Using drones to replace manual monitoring and picking of fruits can save costs free up manpower. Achieving precise by requires strong indoor positioning capabilities, but complex environment orchards, using single sensor may produce data fluctuations. This project proposes method fuse depth camera UWB solve problem insufficient performance sensor: independently obtain UWB, use Kalman filter process noise, calculate fluctuations degree,...
This paper proposes an edge-directed interpolation algorithm for infrared images. At present, the resolution of focus planar array (IFPA) is relatively low. Conventional linear schemes such as pixels replacement, bilinear and bicubic result in blurred edges zigzag pictures. The correlation different edge direction was calculated at each pixel to be interpolated detect direction. There are 13 directions two quadrants we have chosen. Most can detected this range. Pixels along edge. non-edge...
There are many researches on pedestrian detection.These mainly used in ADAS systems based the front view of pedestrian.However some situation, for example bus or public access, passenger flow calculation needs to be done top images.This paper presents an efficient FPGA-based a Xilinx Zynq SoC.The algorithms presented include hair model and head detection.The hardware software collaborative design realized accurate detection real-time.
There are many researches on pedestrian detection. These mainly used in ADAS systems based the front view of pedestrian. However some situation, for example bus or public access, passenger flow calculation needs to be done top images. This paper presents an efficient FPGA-based a Xilinx Zynq SoC. The algorithms presented include hair model and head hardware software collaborative design realized accurate detection real-time.
Both direct and optical flow visual odometers are based on a strong assumption that the gray scale is invariant. Because of this assumption, system sensitive to luminosity change image. To solve problem, paper proposes monocular visual-inertial odometry exploiting both multilevel Oriented Fast Rotated Brief (ORB) feature tightly-coupled fusion strategy. The purpose method improve speed robustness matching. Furthermore, it also can build high-precision initialization map ensure successful...
This paper proposes a low-latency multi-layer pipelined hardware implementation scheme of context based adaptive binary arithmetic coding (CABAC) for H.265/HEVC, which improves the performance and reduces encoding latency with less overhead. It is verified, debugged optimized on FPGA platform. From synthesis results Xilinx K7 series FPGA, it can be seen that at 200MHz clock, CABAC this only uses 12K LUT resources, supports 1080P60Hz video encoding, as low 6.7us. Experiments calculations show...
Video latency plays a critical role in applications that require real time delivery of video content. Key areas include teleconferencing, remote control, cloud gaming, and live news gathering. Given the value into these markets, measurement remains an important topic. With advent artificial intelligence deep learning neural networks, reliable low transmission becomes for automated control. Human judgements one-off manual procedures are no longer reliable. Frame freezes, frame jumps, slow...
Dynamic real-time control through a remote network is becoming reality, given the drive of latest generation wireless networks. As cloud computing takes on an even stronger role in industry, desire to offload time critical processes into becomes possibility. With combination edge processing centers, and UUL standardization for sub-millisecond delay over 5G networks, applications requiring dynamic closed loop now option. support processors, advanced complex analysis algorithms, such deep...