- Image and Object Detection Techniques
- Traffic control and management
- Advanced Image and Video Retrieval Techniques
- Advanced Image Fusion Techniques
- Robotic Path Planning Algorithms
- Consumer Market Behavior and Pricing
- Traffic Prediction and Management Techniques
- Transportation Planning and Optimization
- Medical Image Segmentation Techniques
- Image Enhancement Techniques
- Advanced Differential Equations and Dynamical Systems
- Text and Document Classification Technologies
- Neural Networks and Applications
- Decision Support System Applications
- Aerospace Engineering and Control Systems
- Consumer Retail Behavior Studies
- Fractal and DNA sequence analysis
- Robotic Mechanisms and Dynamics
- Business Process Modeling and Analysis
- Optimization and Search Problems
- Elevator Systems and Control
- Modular Robots and Swarm Intelligence
- Digital Image Processing Techniques
- Remote Sensing and Land Use
- Vehicle emissions and performance
Tongji University
2024-2025
Dalian Maritime University
2021-2022
North China University of Technology
2002-2010
Hunan University
2007-2009
Institute of Automation
2002
Peking University
1991
University of Pittsburgh
1972
Treating each intersection as basic agent, multi-agent reinforcement learning (MARL) methods have emerged the predominant approach for distributed adaptive traffic signal control (ATSC) in multi-intersection scenarios, such arterial coordination. MARL-based ATSC currently faces two challenges: disturbances from policies of other intersections may impair and stability agents; heterogeneous features across complicate coordination efforts. To address these challenges, this study proposes a...
A novel region-based color mapping method is proposed to render fused image of visible and infrared (IR). The based on segmentation, region recognition, fusion transfer. Firstly IR are the curvelet transform. At same time, a database formed by grouping set natural images according their scene contents. false obtained assigning source RGB channels segmented with multiscale normalized cuts method. Then each recognized support vector machine (SVM) automatically associated image. Finally colors...
A novel color fusion method for visible and infrared (IR) images is presented in this paper. The image fused with the original reverse polarity of (IIR) by a grayscale scheme based on curvelet transformation, respectively. These two results are red blue space final image. And linear combination result IR assigned to green channel result. rules consistent human's vision associations warm cool blue. experimental show that will maintain high resolution improve efficiently discrimination between...
The PnP problem is a widely used method in pose determination since it single-view based approach and does not involve point correspondences across image, difficult task computer vision. multiplicity of its solutions an important issue, sometimes indispensible to know exactly how many possible exist real applications. In this work, issue will be investigated terms P3P problem, the results could provide some insight on general, reference people field.
The increasing number of e-petition services requires accurate calculation methods to perform rapid and automated delivery. Automated text classification significantly reduces the burden manual sorting, improving service efficiency. Moreover, existing focus on sole models with an insufficient exploration hybrid models. research lacks combinatorial model selection schemes that yield satisfactory performance for petition applications. To address these issues, we propose a deep-learning can...
Unmanned traffic signal control is regarded as a sustainable intelligent management methodology. However, it faces the challenge of unpredictable flow due to stochastic arrivals. The digital twin (DT) has emerged promising approach address challenges time-varying demand in urban transportation. Previous studies DT-based adaptive (ATSC) methods all assume ideal synchronization conditions between DT and physical (PT). It means that can immediately figure out next neglecting limitation...
Some key factors influencing the performance of a speaker recognition system using neural network approach are addressed. The selection input pattern affects complexity and network. In experiment, seven kinds patterns were selected. distribution every kind is different, which training time rate. with most suitable gives best requires time. optimal scale affected by problem. Applying subnetwork may keep acceptable or reasonable as number speakers increases. A method for text-independent also...
Curve matching is one of key issues in computer vision, image analysis and pattern recognition. Based on discrete V-transform, the distance calculated between curves using descriptor V-system to find curves, then parameters are evaluated this article. The new approach can efficiently rough location a short extracted curve long reference curve. Different from existing approaches, it needn't extract feature points. Extensive tests show that efficient.
The study of human chromosomes with respect to the health monitoring, clinical karyotyping, abnormality detection, and genetic counseling, has urged development an efficient automatic system improve speed, reliability capacity for chromosome analysis classification 1-5. Prior 1970, in blood cells were analyzed terms their size gross morphology. In normal cells, there are forty-six which form twenty-two homologous pairs plus two sex chromosomes, XX or XY. These divided into seven groups: A,...
The main purpose of quasar recognition is to determine the observed spectrum's redshift value. In past, template rest frame was mainly constructed based on astronomers' inference. Due inaccuracy such a template, it hard value by matching spectrum with directly. This paper's contributions are two-fold. Firstly, in our paper principal component analysis (PCA) method from some selected spectra known readshift values, hence resulting more realistic. Secondly, 2D Hough transform, rather than 1D...
This paper proposes a novel approach of in-store route data obtained by RFID for customer segmentation based on process mining (PM). The PM technique can dig out the hidden information customers' shopping deeply. Firstly, we use to customize and decompose consumers' movement paths. Then, divide fish area into patterns, define these patterns as passing area, staying entering respectively. Finally, customer's path several sub-processes according moving sequences technology. In experiment,...
Consumer behavior analysis is of great significance to retail merchandise marketing. This article aims establish a customer classification model that includes the complexity shopping paths. First, we select target area capture customer's movement path data. The data contains series points with (x, y) coordinates. We plot into map via transforming pixels from In this stage, are connected by lines according time sequence. Secondly, box-counting method used calculate fractal dimension each map....
We propose a new motion dynamics approach to solve the PnP problem, where dynamic simulation system is constituted by springs and balls. The equivalence between minimizing energy of solving problem proved. With assumption existence resistances, solution original can be solved through process movement