Homepage People Research Publications Demos News Internal Login


Address:

Engineering Research Center
(ERC) 817,
College of Engineering and
Applied Science,
University of Cincinnati


Phone:

(513)556-4773


Email:
weewg@ucmail.uc.edu

 

Research
Current Projects:
1. External project # 1: NDE 2D and 3D Data Processing with GE Aviation, 2011-2015
2. External project # 2: Feasibility Study of an Automatic Parameter Calibration Method for Multiple sets of Pre-calibrated Emitters on an IS-900 Controller with GE Aviation, 2011

3. University of Cincinnati Summer Graduate Student Research Fellowship Project: Tao Ma, A 3D Virtual Learning System for STEM Education, 2011

Research Activities:

Jin Quan, Email: quanjn@mail.uc.edu

* A Wavelet Based Image Denoising Method

A new wavelet based image denoising method is developed by using a set of elementary denoising functions in the form of derivatives of Gaussian. The wavelet coefficients used are derived from an improved context modeling procedure in terms of MSE estimation combining inter- and intra-subband data. The denoising method results in a two-step denoising effort which outperforms the state-of-the-art non-redundant methods. This denoising method is also extended to the overcomplete expansion by applying cycle spinning, which provides additional denoising performance and yields better results than the orthogonal transform.

* The Reduction of Non-Gaussian Distribution Noise on Image Based on Wavelet Transform

In many applications of image denoising, generic models for quantitative analysis are less promising due to the various physical acquisition process. Thus, a non-Gaussian model is likely to yield better performance in terms of PSNR.Currently, a wavelet based image denoising algorithm for non-Gaussian Noise is under development


Zhen Jia, Email: jiazn@mail.uc.edu

* Major topic of image registration, with focus on the Non-Destructive Evaluation (NDE) application.

Image registration is to find the transformation between two images so that they can be spatially aligned. It involves several sub topics and some of them can be expanded and studied as Ph.D. thesis themselves which makes image registration a very comprehensive topic.

Papers on this topic started to emerge from two decades ago, and based on my literature review on the papers, it is an application driven topic with major applications on medical area. For example, image guided surgery and medical images information fusion. Each paper proposes different algorithms dealing with different image modalities, such as PET/MRI, CT/MRI, or different imaging objects, such as brain, skull, and vessel.

The NDE application I am working on comes from the contract project of GE Aviation. Their inspection team uses the most sophisticated inspection technologies such as digital X-ray, phased array ultrasonic, and infrared thermography to evaluate the integrity of engine components and materials. I am cooperating with them studying the registration between 2D Infrared image and 3D CT image. It is a double-challenge problem: first, images to be registered are of different dimensions; second, infrared/CT is a brand new modalities combination that has never been studied in registration area before. Therefore, algorithms proposed before are not guaranteed to work on this new application.

Until now, work I have done can be classified into three blocks: imaging principle of infrared and CT images, literature review of previous work on image registration algorithm, and experimental test. Using information theory measure, I have had some preliminary results. From there, I will improve the algorithm to meet the speed and accuracy requirement.


Tao Ma, Email: mata@mail.uc.edu
Webpage:   http://homepages.uc.edu/~mata

* A 3D Virtual Learning System for STEM Education

With the fast development of technology and the benefits it brings to modern society, the application of technology in education attracts great attention. Digital virtual worlds have been used in education for a number of years, as the common use of which, the issues of providing effective support for teaching and learning arouse continuing discussions. Many current educational tools instead of enhancing the notions of interaction and student-centered learning only strengthen the traditional teaching methods using a new platform without changing the nature of how to teach and learn. The functions of computers in online learning environments and the existing interactive systems are far from what is desirable. Some classroom technologies do not only maximize the authority and the influence of teacher, but also help teacher directly transfer their understanding and experiences to students. On the other hand, game, especially TV and computer game shows a great appeal to young people. The question remains to educators and computer scientists is, what we can learn from game systems? 

A recent boom has been seen in 3D virtual worlds for entertainment, and this in turn has led to a surge of interest in their educational applications. Will 3D bring much differences to education compared with previous virtual worlds systems? In this paper we will briefly review the use of virtual worlds in education, critically analyze the existing virtual learning systems and consider what is required for a virtual learning environment. In this we will focus on the development of a 3D virtual learning system that merges real experiments in virtual laboratory and brings entertainment to knowledge learning. The design of the system aims at improving teaching and learning efficiency and interest in the STEM field by introducing advanced human machine interface and game-based teaching software into classroom and self-study. Learners using this system are able to operate virtual objects by hand gestures in virtual laboratories, practice experiments for better learning, acquire immediate feedback of performance and learn knowledge as playing games, etc. A PC system with additional hardware and software is designed to provide a 3D virtual space with a virtual toolbox such that science experiments can be interactively designed and implemented. The operation of this system will be demonstrated at the conference and possible future applications will be discussed by the author.


Yi Sun, Email: suny9@mail.uc.edu

* Sensor Selection, Arrangement and Fusion on Multiple Circumstances

Nowadays, numerous technologies and approaches have been applied on motion tracking systems. The physical principles they based on, the performance they exhibiting, and the purposes they designed for are all different. So how to select these approaches to apply to different circumstances is a hard problem for people to work on.
Our work is based on analysis of different aspects of working environment, like the volume, shape, material, etc; and different properties of trackers, like size, range, accuracy, and cost effectiveness, to select appropriate type of sensors and emitters to make accurate measurement of physical values, and provide data to motion tracking systems. Now we are focusing on the combination of inertial sensing with ultrasound technology, to provide accurate tracking data for motion tracking system, which can be used on civilian and industrial applications.

Xinhua Xiao, Email: xiaoxa@mail.uc.edu

* Image Model Optimizer for Anomaly Detection in Nondestructive Testing

X-ray inspection is widely used for identifying potential defects in manufacturing parts. The main inspection task is to locate abnormalities which may be located inside the manufacturing parts via a computer-aided analysis of Digital X-ray images (DXR). Many assisted defect recognition systems have been developed to enable manufacturing parts inspection using DXR images.

For the assisted defect recognition system, reference method is largely used in identifying defects, in which a test image is compared with the reference image model. If the difference is significant, the test piece is considered as defective. Therefore, the reference model image selection is of great importance. In the present time, human expert is doing the selection task. It is meaningful to develop an automatic method to classify the DXR images into acceptable or unacceptable for model images, and select the representative from the acceptable images.