Applications of intelligent image processing

该项目涉及模式识别和机器学习技术在图像检索、图像场景分析和生物识别问题中的应用研究。除了人脸识别应用外，机器学习技术在计算机视觉应用中的使用是相对较新的。我们建议研究一种从大型图像数据库中进行有效分类和检索的方法，并将其应用到皮肤病图像中，为皮肤科医生和肿瘤学家提供一种诊断工具，以解决涉及虹膜识别的生物识别问题。我们还研究了移动计算中涉及基于生物特征的身份识别和验证的问题。我们提出了一种通过将图像分类为不同的语义类别并使用概率相似性度量来进行图像检索的方法。随着医学领域信息技术和数字成像革命的日益普及，诊所经常产生大量的数字皮肤病图像用于图像引导诊断。针对皮肤科领域的数字医学图像，提出了一种基于内容的智能图像检索方法。它可以作为一种有效的培训工具，供医学生、住院医生和研究人员根据其视觉属性高效地搜索与疾病相关的大量图集。作为机器学习在生物特征识别中的应用，我们提出了一种基于支持向量机的生物特征识别技术的虹膜识别系统。该项目在一定程度上是跨学科的，将提供与该行业和医生的互动。该项目将培养三名研究生和一名博士后研究员。

The project involves investigations on the applications of pattern recognition and machine learning techniques to problems in image retrieval, image scene analysis and biometrics. The use of machine learning techniques in computer vision applications is relatively new except for applications to face recognition. We propose to investigate an efficient classification and retrieval from large image databases, and also its applications to dermatological images to provide a diagnostic tool to dermatologists and oncologists, problems in biometrics involving iris recognition. We also investigate problems in mobile computing involving biometrics-based identification and verification. We propose an approach to image retrieval by classifying images into different semantic categories and using probabilistic similarity measures. With the growing availability of information technology and digital imaging revolution in the medical domain, large collections of digital dermatological images are regularly being generated by clinics for image-guided diagnosis. We propose an intelligent content-based image retrieval approach of digital medical images in the domain of dermatology. It may be used as an efficient training tool for medical students, residents and researchers to search efficiently large collection of disease-related illustrations by their visual attributes. As an application of machine learning to biometrics, we propose an iris recognition system as a biometrically based technology for person identification using support vector machines. The project is partly interdisciplinary and would provide interactions with the industry and medical doctors. The project would train three  graduate students and a Postdoctoral Fellow.