Print
Univis
Search
 
FAU-Logo
Techn. Fakultät Website deprecated and outdated. Click here for the new site. FAU-Logo

Image Analysis

The Image Analysis Group is dedicated to extract information from images. Examples are the outlining of specific structures in 2D and 3D images, like extraction of pages in CT scans of books or the detection of lesions in mammographic images.

Place

Usually the colloquium takes place in room RZ 2.009 (conference room). If the room is not available, you can find us either in e-Studio (RZ 2.037) or in our own seminar room 01.134.

Next sessions

DateResponsible PersonSpeakerType & Title of Contribution

 

04.06.2019

 

Viktor Haase

 

Improving FISTA: Faster, Smarter and Greedier

28.05.2019

29.01.2019

Weilin Fu

Lennart Husvogt

MONet: Unsupervised Scene Decomposition and Representation

OCTA Image Generation

05.02.2019Viktor HaaseExploring the Space Between Smoothed and Non-Smooth Total Variation for 3-D Iterative CT Reconstruction
26.02.2019Weilin FuOpens external link in new windowFast End-to-End Trainable Guided Filter
12.03.2019Mario AmrehnRobot User Comparison and Opens external link in new windowUnderstanding Disentangling in β-VAE
from 12.2.to 23.4.2019no lectures, colloquium on demand (e.g. thesis talks)

Opens internal link in current windowPrior sessions

 

Mailing list subscription management page for Opens external link in new windowstudents and Opens external link in new windowguests.

In case a remote participation is needed, please Opens internal link in current windowcontact the organizer of the colloquium.

Running Projects


3-D Reconstruction of Historical Documents

This project focuses on the reconstruction of historical documents that can not be opened or page-turned anymore. We perform X-ray CT scans and develop algorithms to extract and visualize pages from the 3-D volume. This is possible because historical documents were commonly written with ancient inks consisting of metallic particles.

Parameter-Optimization for DBT Imaging Systems

DBT can be used for screening of the human breast which implies a specific demand of best possible image quality w.r.t. diagnostic value. One diagnostic measure is the detectability of lesions which is connected with the concept of model observers. Ultimately, we use these kinds of measure to optimize DBT reconstruction parameters.

Image Quality Assessment with Human Observers

The penalized least-square reconstruction with statistical weights is a popular model-based iterative reconstruction method for CT. In our project we examine the impact of the statistical weights on image quality. We conduct human observer studies to test the lesion detection performance on simulated phantom data. The final image quality assessment is based on the analysis of the localization receiver operating characteristic.

Airway Segmentation

Airway Segmentation is a challenging task, because the radius range of the airway can be large, and image quality can be bad. We are using differen methods to segment airway out, including region growing, Frangi tubeness, cavity enhancement filtering, circle detection filtering, gradient vector flow, and super pixel.

Image Procesing for Ophthalmic Optical Coherence Tomography

Optical Coherence Tomography is a widely used imaging modality in ophthalmology. Ocular diseases such as diabetic retinopathy and age-related macular degeneration are becoming more common due to the increasing prevalence of diabetes and increasing life expectancies in the population. This project aims to aid in the diagnosis and mointoring of ophthalmic dieases.

Finished Projects

Segmentation of Fat and Fascia Layers in Ultrasound Images

One recent research area in zoology is to measure the thickness of the fat and fascia layer within ultrasound images. Our goal is to develop an algorithm for fully automatic separation of those layers. Furthermore, we want to provide a GUI for specialist such that there is no more need to manually measure the layers.

Machine Learning-based Material Decomposition for Spectral X-ray Imaging

Benefiting from multi-energy X-ray imaging technology, material decomposition facilitates the differentiation of different materials in X-ray imaging. We propose a novel machine learning-based pipeline to perform material decomposition using machine learning algorithms. Feature extraction is involved into the pipeline to improve the performance of material decomposition.