PHYS34611: Project
List of Projects Available 2015-16
For students on BSc Physics with Forensic Applications
Title:Ballistics simulations.
Project Type: Experimental, including building equipment.
Involves:Designing & testing apparatus, including factoring in safety considerations, plus developing a theoretical interpretation.
Description:Design & build a ballistic pendulum to measure the velocities of balls fired from a BB gun. Theory required to interpret data to be deduced from first principles. Scope for multiple design solutions. Compare the results with data obtained from the “Ballistics & Firearms” module experiment using light gates. Test reproducibility. Possible extension of project to measure dispersion of shot.
Title:Forensic Analysis of torch bulbs and cycle lamps following an accidents.
Project Type:Experimental.
Involves:(Safe) destructive testing of bulbs, optical and electron microscopy.
Description:It has been shown that microscopic analysis of car headlamps can show whether the lamps were switched on, switched off or already broken at the time of a vehicle accident. This is due to the different behaviour of the bulb filament material under mechanical stress or impact when it is hot compared to when it is cold, and it can be important to assigning practical and legal blame, or in criminal investigations. In this project you will investigate whether similar techniques can be used to identify the status (on off or broken) of lower powered bulbs, such as hand-held torch bulbs, or bulbs from cycle lamps. This will first involve devising methods of safely impact shattering bulbs while switched off or on, (as well as “blowing” bulbs through current overload,) and then using optical microscopy to investigate the structure of the bulb filaments, (and possibly the glass fragments,) and the damage done. Interesting sample bulbs can then be investigated at ultra-high resolution using the school’s Scanning Electron Microscopy (SEM) in ERD125.
Title:A biometric system based on hand writing recognition.
Project Type:Experimental.
Involves:Recruiting volunteers, collecting hand writing data, statistical analysis.
Description:Physical measurements on humans can sometimes comprise of information that is highly unique to an individual. This is the case for human hand writing.
In this project an accelerometer will be secured to a pen, and connected to the sound card of a PC for continuous logging of the time course of the pen’s 3D acceleration as volunteers are asked to write a short statement of their choice.
Software will be applied to extract a signature from the recording, unique to the volunteer under investigation. The accuracy of the system will be clarified.
Title:Human Face recognition using Principal Component Analysis
Project Type: Experimental.
Involves:Recruiting volunteers, collecting face photos, statistical analysis.
Description:In this project a simple set up will be made by the student in order to capture photos of human faces (volunteers), and process them so as to have the computer to recognize somebody from their face. The project will involve setting up a lab space for the photos to take place, recruit volunteers for the collection of the data base, and use matlab software to achieve the automatic recognition of the individuals.
Title: What can we learn from a blood droplet?
Project Type:Experimental & Image Processing.
Involves:Image analysis and macro “plug-in” programming in ImageJ or Matlab.
Description:You will investigate how the pattern formed by a dried blood droplet varies between different subjects. Using blood taken and characterised in various experiments around the school, you will take images of dried droplets and develop image processing techniques to characterise the patterns seen. You will look for statistical differences between the droplets which will hopefully correlate with the other measured parameters of the blood.
Title: Digital Separation of Overlapping Fingerprints.
Project Type:Image processing
Involves:Image analysis and macro “plug-in” programming in ImageJ or Matlab.
Description:The aim is to investigate whether two overlayed fingerprints can be separated using image processing techniques. Such prints are not normally considered useable, since ridges from one print will cross ridges from the other, making it hard to assign identifiable features to one print or to the other. However, it might be possible to use them for identification if a technique were developed to separate them. In principle this could be done using image processing techniques, such as using a series of FFTs of small subsections of the image or skeletonising the image and then using morphological processing techniques to trace out the primary ridges. This project would investigate the feasibility of one or more of these (or other) approaches, and then to try to write an ImageJ “plug-in” macro, or possibly a Matlab programme, to automate the process.