MSc dissertation proposal 2017/2018

Non-conventional Vision for Scenarios under Investigation (id 9272)

Objectives:

A crime scene under investigation provides initial data which does not suffice to answer the questions of the investigators. After some more data acquisition, completed with processing, many questions can be answered. In this thesis work is explored vision through hammered glasses paying special attention to enhance the visibility of a moving suspect that needs to be identified. Scenarios of interest include: (i) one fixed surveillance camera in a room separated from the crime scene by a hammered glass, (ii) considering one mobile phone instead of the fixed camera. The difference of the two cases is the lack of knowledge of the pose of the mobile camera with respect to the glass. In this thesis work is implemented imaging geometric correction to improve the imaging through a hammered glass.

Goals:

1) Review of “raxel” representations for cameras whose input light rays may have non-convergent 3D directions [Bergamasco13];

2) Mosaicking of a panning video sequence, shown by a computer LCD and acquired by a camera having in between a hammered glass

3) Segmenting and reconstructing a moving character in a static video sequence

References:

[Bergamasco13].Can a fully unconstrained imaging model be applied effectively to central cameras? Filippo Bergamasco, Andrea Albarelli, Emanuele Rodolà, and Andrea Torsello, In 2013 IEEE Conference on Computer Vision and Pattern Recognition, 2013

Requirements (grades, required courses, etc):

Interest for computer vision, video surveillance and imaging enhancement.

Place for conducting the work-proposal:

ISR / IST

Observations:

The increasing need of surveillance in public spaces and the recent technological advances on embedded video compression and communications made surveillance cameras ubiquitous. Even in places not equipped with surveillance cameras, cameras ubiquiti still happens - people use the cameras of their mobile phones and effectively acquire relevant forensic information. While the technological advances already allowed such a ubiquitous presence of cameras, the still infancy stage of video analytics and the high variability of forensic research challenges make per si the extraction of information from the cameras an active research area.

As referred in the objectives, in this thesis is implemented imaging-geometric-correction to improve the imaging through a hammered glass. The central problem consists in describing paths of 3D optical rays, from the target towards the camera, passing through the hammered glass. The description for the camera itself is the traditional pin-hole model. The paths from the glass towards the world are described as a 3D line per pixel.

The testing of the proposed system is based on a computer LCD. The LCD displays images (videos) which are captured by a camera placed behind a hammered glass. Displaying images as checkerboard patterns allows the direct identification of 3D trajectories of light rays travelling from the LCD, passing through the hammered glass and being captured by the camera.

Expected results:

At the end of the work, the students will have enriched their experience in computer vision applied to forensic applications.

More MSc dissertation proposals on Computer and Robot Vision in:

http://users.isr.tecnico.ulisboa.pt/~jag/msc/msc_2017_2018.html