Focusing Attention in Robot Vision using Evolutionary Computation
Title: Focusing Attention in Robot Vision using Evolutionary Computation
Supervisors: NH Siddique, JV Condell
Cognitive Robotics Team
Focusing attention refers to the ability to respond discretely to specific visual, auditory or tactile stimuli. Anne Treisman [1] developed the highly influential feature integration theory. According to this model, attention binds different features of an object. It is concluded from many experiments that color, orientation, and intensity are primitive features, for which feature searches can be performed. There are distinguishable two kinds of visual search tasks: feature searches and conjunction searches. Feature searches can be performed fast and pre-attentively for targets defined by primitive features. Conjunction searches are the serial searches for targets defined by a conjunction of primitive features. These are much slower and require conscious attention.
Image enhancement visually enhances certain parts of images. Domain experts are not always experts of image processing and cannot always design image enhancement filters. At the same time, image processing experts may not have the domain knowledge. Interactive Evolutionary Computation (IEC) can be applied to visually generate descriptions and equations which determine the characteristics of the filters. An example of such an application is the design of a colour filter to enhance MRI images to aid decision making of medical doctors. This could involve the design of an image synthesis filter that would help the doctor to see when two echo-cardiograms are mapped onto one. The key issues that this project will investigate are: the development of an image filter for robot vision system - conventional (mathematical), nonconventional (e.g. fuzzy, neural networks) or hybrid approaches may be explored in designing the image filter, development of an IEC to improve the performance of the image filter, and finally the definition of a fitness function to be used by the IEC that incorporates a measure of attention.
It is envisaged that the main outcome of the project will be an image filter using an IEC within the design loop. This model can be deployed to various applications such as enhancing MRI images for medical doctors. The achievements and results from this project will be disseminated by publishing research outputs to conferences and journals of international repute in the field.
Resources needed: The project may need to use the Pioneer II/SCITOS robots equipped with a vision system available in the new robotics laboratory in ISRC. The researcher will also need access to the MATLAB suite and Toolboxes.
Ethical issues: None
- Treisman, A. M. and Gelade, G., “A feature-integration theory of attention”, Cognitive Psychology, Vol. 12, No. 1, pp. 97-136, 1980.
- Treisman, Anne M., and Nancy G.Kanwisher, "Perceiving visually presented objects: recognition, awareness, and modularity," Current Opinion in Neurobiology, (1998), pp 218-226
- Burbridge, C., Spacek, L., Condell, J.V., Nehmzow, U.(2008) Monocular Omnidirectional Vision based Robot Localisation and Mapping. In Proceedings of TAROS 2008. Editors: Ramamoorthy, S., Hayes, G. M., IEEE Computer Society Press, 135-141, 2008.
- Hayashida N. and Takagi H., “Acceleration of EC Convergence with Landscape Visualization and Human Intervention,” Applied Soft Computing, Elsevier Science.
- Ikuta, A; Ohta, M; and Siddique, M.N.H. (2005). Prediction of Probability Distribution for the Psychological Evaluation of Noise in the Environment Based on Fuzzy Theory, International Journal of Acoustics and Vibration, Vol.10, No.3, pp.107-114.
Personnel Involved
First Supervisor: Siddique, N Dr
Second Supervisor: Condell, J Dr
Collaboration: This project does not involve collaboration with another establishment