Adaptive signal analysis: novel methods and applications

信号处理在消费电子、医疗设备、地球物理应用、金融分析等领域具有显著的先进技术。拟议的研究将开发新的先进的信号处理方法，适应真实的世界的信号通常遇到的生物识别和多媒体应用。与这些系统相关的信号的基本特征是，它们与时变行为相关联，它们具有时间局部化信息，并且还表现出相对于信号频谱特性的许多可变性。为了从信号中提取信息并描述事件的特征，并了解潜在的环境，需要设计复杂的信号分析算法（和分析工具）。拟议的研究将调查新兴的信号处理方法，使用信号分解框架和表示的显着信号特征的鲁棒提取。特征提取与分析有着广泛的实际应用，是多媒体、生物医学、生物识别等智能系统中信号处理系统的核心。在拟议的研究中，提取的功能将用于智能嵌入信息（水印和指纹）在多媒体（音频）文件的版权保护，广播监控和内容识别。我们还将研究无处不在的生物识别系统，使用生物标记和语音来识别用户。这些系统的人机交互的成功设计的关键也将取决于强大的特征提取模块，这是本研究的主要研究课题。这些算法将应用于我们实验室和其他开源数据库中收集的真实的世界数据集。据设想，特征提取算法将揭示潜在的信号生成/修改机制，这可能有助于识别隐藏的签名，以便在智能系统的其他领域中更好地识别和分类应用。算法和数据库也将与世界各地其他感兴趣的研究小组共享。

Signal processing has significantly advanced technologies in consumer electronics, medical devices, geophysical applications, financial analysis and many more. The proposed research will develop novel advanced signal processing methods adapted to real world signals typically encountered in biometrics and multimedia applications. The underlying characteristics of signals involved with these systems is that they are associated with time-varying behavior, they have time localized information, and also exhibit lot of variability with respect to the signal spectral properties. In order to extract information from the signal and characterize the event, and to understand the underlying environment, sophisticated signal analysis algorithms (and analytical tools) need to be designed. The proposed research will investigate emerging signal processing methodology that uses signal decomposition framework and representation for robust extraction of salient signal features. Feature extraction and analysis has lots of practical applications, and is the core of signal processing systems encountered in multimedia, biomedical, biometrics and many other intelligent systems. In the proposed research, the extracted features will be used for intelligently embedding information (watermarking and fingerprinting) in multimedia (audio) files for copyright protection, broadcast monitoring, and content identification. We will also investigate ubiquitous biometric systems that uses keystrokes and speech to identify the users. The key to successful design of these systems for human-machine interaction will also depend on the robust feature extraction module which is the main topic of investigation in this research. The algorithms will be applied to real world datasets collected in our lab and other open source databases. It is envisioned the feature extraction algorithms will reveal underlying signal generation/modification mechanisms that could be useful in recognizing hidden signatures for better recognition and classification applications in other areas of intelligent systems. The algorithms and the databases will also be shared with other interested research groups around the world.