My research interest is the 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.
yes, Jia.
ReplyDelete