Image and Video Processing

Image & Video Processing Research Group:

Variable Gaussian Mixture for Image Modeling, Recognition, and Compression:

Gaussian mixtures have gained popularity in statistics and statistical signal processing applications for a variety of reasons, including their ability to well approximate a large class of interesting densities and the availability of algorithms such as Expectation-Maximization (EM) algorithm for constructing the models based on observed data. Modeling using Gaussian mixtures have been used in image segmentation to model and classify textures. They were also employed as soft vector quantizers to compress both audio and image data. Methods will be applied to face recognition, biomedical imaging, and financial data analysis.

Electrocardiogram (ECG) Reduction and Modeling:

ECG data compression is gaining importance in biomedical engineering due to the potential applications in telemedicine. The Holter monitoring system, for example, consists of obtaining cardiac signals over 24 h, using from 5 to 7 electrodes. Consequently, the storage of such signals is an important subject on the implementation of ECG acquisition systems. Also the transmission of the obtained signals is becoming more crucial with the growing advance of the telemedicine. Techniques borrowed from nature, such as neural nets and swarm intelligence can be exploited to provide tools for better compression and representation of physiological signals in general and ECG in particular.

Digital Video Coding:

Digital video coding has become an essential component of broadcast and entertainment media. Continuous improvements in digital video coding are required to help narrow the gap between the users’ demands and the limited capabilities of transmission networks and storage devices. This opened the door for the rise of many emerging topics such as: reconfigurable video coding, scalable video coding, multi-view video coding, high-performance video coding, next-generation video coding, as well as efficient hardware/software implementations for high-definition real-time video coding on reconfigurable and multi-core platforms.