Novel image analysis and computer vision techniques for multimedia, communication, and other applications

信息科学和电子学是21世纪最重要的科学技术领域之一。关键的网络化和数字化生活技术将对我们的日常生活和各个工业部门产生重大影响。如何利用视听信息进行交互式多媒体传播是近年来学术界和产业界都十分关注的问题。本研究将涉及广泛的主题，涉及开发用于安全、通信和其他应用的新型图像处理和计算机视觉算法和系统。拟议研究的目标是：(a)。为了提高视频编码的压缩效率，提出了一种改进的、高效的帧内预测算法。(b)。进一步发展刚性和非刚性场景下图像序列的摄像机标定和三维重建的新理论和新方法。(c)。提出一种新的生物特征水印方案，使生物特征信息在充分保护下进行通信和传输。由于这些（指纹、DNA或虹膜）是个人的唯一标识符和永久特征，生物识别技术在公共部门和私营部门正获得越来越多的兴趣。这些特性使得生物识别技术不仅是一个非常强大的工具，可以在多个数据库中匹配不同的个人信息，而且还可以尝试提高安全性。

Information science and electronics are among the most important areas of science and technology in the 21st century. The key networked and digital living technologies will have a major impact to our daily life and various industrial sectors. How to employ both audio and visual information for interactive multimedia, communication has recently received much attention in both academia and industry. This research will deal with a broad range of topics concerning the development of novel image processing and computer vision algorithms and systems for security, communication, and other applications.The objectives of the proposed research are: (a). To develop an improved and efficient intra prediction algorithm for video coding with high compression efficiency. (b). To further develop new theory and new methods on camera calibration and three dimensional reconstruction from image sequences of rigid and nonrigid scenarios. (c). To develop novel biometric watermarking scheme to communicate and transmit biometrics with full protection. Biometrics is gaining more and more interest in public as well as in private sector since these (fingerprints, DNA, or irises) are the unique identifiers and permanent characteristic of the individuals. These properties make biometrics not only a very powerful tool for matching different pieces of information of individuals across multiple databases but also attempt to increase the security.