研究方向

• 模式识别与机器学习；
• 图像与视频处理；
• 高分辨率遥感图像处理与应用；
• 高光谱图像处理与分析；
• 目标检测与识别；
• 图像解混；
• 信号处理及图像处理中的数学理论研究；
• 图像处理应用系统研究。

科学基金列表 (20余项，部分列表)

As Project Leader or Principal Investigator, Dr. Zhenwei Shi has been in charge of the National Natural Science Foundation of China, sub-project of Key Funding Projects of National Natural Science Foundation of China, sub-project of National Basic Research Program of China, the National 863 Project of China, the Beijing Natural Science Foundation, the Research Fund for the Doctoral Program of Higher Education of China and practical engineering tasks, etc.

• Name:Hyperspectral image fusion and target detection algorithms based on blind, signal processing, PI, National Natural Science Foundation of China, 2013.1-2016.12
  Description: We treat the hyperspectral image fusion and target detection as a series of special blind signal processing problems. Based on the thoughts, the project will study a series of challenging problems. These include features analysis of typical hyperspectral images, hyperspectral image fusion based on blind source separation, target detection in fused hyperspectral image based on blind signal extraction, etc.

• Name: Traffic sign graphic detection, recognition and understanding under complex conditions, Co-PI, Key Funding Projects of National Natural Science Foundation of China, 2012.1-2015.12
  Description: Image perception and understanding is an important research topic. Automated detection, identification and understanding of traffic signs make great sense when applied to the driverless vehicles with a natural environment perception and intelligent behavior of decision-making capacity. The project is mainly focused on the problem of automated detection, identification and understanding of traffic signs in complex traffic conditions.

• Name: Blind separation of moved and superimposed images, PI, National Natural Science Foundation of China, 2010.1-2012.12
  Description: Traditional Blind Source Separation (BSS) approaches often fail in separating practical superimposed images, because they cannot handle the layer motions which usually consist of translations, rotations, scaling and other effects. To change this situation, we propose a new research project -- Blind Separation of Moved and Superimposed Images for separating superimposed images and recovering component layers.

• Name: Algorithms for time dependent component analysis, PI, National Natural Science Foundation of China, 2007.1-2009.12
  Description: Contract to independent component analysis (ICA), temporally dependent component analysis mainly uses the time information which the data possibly contain and other data information (for example, higher order statistics information) to propose the novel models, theories and algorithms. It can be applied to the fields such as biomedical signal processing, speech signal processing, image processing, etc.

• Name: Multi-modal high-dimensional heterogeneous data feature extraction and description method, PI, sub-project of National Basic Research Program of China, 2010.1-2014.8
  Description: This research will present some novel high-dimensional data feature extraction and description methods for image processing.

• Name:Space target recognition method based on hyperspectral imaging technology, PI, Program for New Century Excellent Talents in University of Ministry of Education of China, 2012.1-2014.12
  Description: Hyperspectral remote sensing in space provides information related to surface material characteristics of spacecraft or planets that can be exploited to perform automated detection of targets of interest. This research will present some novel space target recognition methods for hyperspectral remote images.

• Name: Hyperspectral image unmixing based on non-negative component analysis , PI, Beijing Natural Science Foundation, 2011.1-2013.12
  Description: Hyperspectral unmixing is a process aiming at identifying the constituent materials and estimating the corresponding fractions from hyperspectral imagery of a scene. This research will present novel hyperspectral image unmixing methods based on non-negative component analysis.