The 11th Lunch Seminar
Date and time: Tuesday, April 21st, from 12h20.
Place: 理学部会議室 (Science building 3, rooms 207-209)
Speaker: Md. Khayrul Bashar (Computer Sciences)
Title: Efficient Vision Methods for Extracting Image Information
With the rapid progress of the imaging technology and the recent explosion of biomedical imaging data, there has been an increasing demand of developing new methods for automated extraction of image information, especially from high dimensional images. In response to above demand, researchers have recently developed a few methods, especially in the field of bio-image informatics. However, they are mostly manual and their outcomes include subjective errors. Sometimes, no results are possible to obtain due to low contrast, high noise, and the complex natures of the biological objects that are present in the imaging data.
In this talk, two automated methods will be discussed. First is a method that employs wavelet theory for pattern segmentation and the supervised classification of informative frames from a long medical video. On the other hand, a Gaussian multiscale based appearance and morphological features are used in the second method for automatic detection and tracking of cell nuclei from 4D mouse embryonic images. Developed methods can play useful roles towards various biomedical applications.