Institute of Bio-Sensing Technology

Photometric stereo

photometric_devicePhotometric stereo is a machine vision method of analyzing and detailing the shape and reflectivity of a surface in 3D space.  

The method involves directing external light sources onto a surface, and gathering multiple images based on the illumination appearance of a given object. The technique allows the determination of both the orientation and reflectance properties of the surface at multiple points.

Photometric stereo has diverse applications including facial recognition, healthcare in particular- tele-dermatology, pharmacology and cosmetology. 

This novel imaging technology has enabled the team to realise devices to measure skin features (such as colour and shape/texture) with high accuracy and sensitivity required for applications in computer aided diagnosis and tele-dermatology. It can be used to achieve highly realistic and interactive 3D visualisations over the Web in remote locations – offering potential application in developing countries.

Examples of projects

PhotoFace- Face recognition project

PhotoFace (Photometric stereo for face recognition) realised a low-cost demonstrator for 3D face recognition based on the acquisition and analysis of surface normals. The project, which was EPSRC funded and undertaken in collaboration with Imperial College and the Home Office, also generated a publically available photometric face database. A second follow-on EPSRC grant is now underway that aims to develop new 3D computer vision methods for estimating the 3D shape and reflectance of a human face and other body parts using high-speed photometric stereo (HSPS).

Skin Analyser- 3D and 2D imaging for tele-diagnostics

The Skin Analyser has been developed as part of a TSB funded project to enable remote realistic and interactive skin visualisations, through use of a vision device that can be used in the community (e.g. in primary healthcare).
A new level of realism and interactivity in remote skin visualisation offers great potential for early detection of a range of skin conditions e.g. skin cancer such as melanoma. The device allows accurate measurements to be taken from the skin for use in computer aided diagnosis (CAD) and in monitoring.

Novel Non-invasive Assessment of Respiratory Function (NORM)

The NORM project developed a new type of medical respiratory monitoring system, using non-contact, non-invasive optical and dynamic photometric stereo techniques, to monitor movements of the thorax and abdomen in real-time. The data gathered enabled the construction of a 4D (space & time) model of the patient, which will enable the diagnosis of respiratory diseases & conditions that compromise normal respiratory function. The NIHR funded NORM project acted as a pilot for a three-year study now underway into computer vision based pulmonary functional testing that is already showing  promising correlation (0.98) with spirometer ground truth data.

Find out more

Melvyn Smith Lead researcher
Professor Melvyn Smith, University of the West of England

For more information about Photometric stereo, please visit the Machine Vision Laboratory website.