Online from: 1981
Subject Area: Electrical & Electronic Engineering
|Title:||Computer screen pointer using intelligent colour sensing systems|
|Author(s):||S.D. Urwin-Wright, (MotionTouch Ltd, Cranleigh, UK), D.A. Sanders, (University of Portsmouth, Portsmouth, UK), G.E. Tewkesbury, (University of Portsmouth, Portsmouth, UK)|
|Citation:||S.D. Urwin-Wright, D.A. Sanders, G.E. Tewkesbury, (2005) "Computer screen pointer using intelligent colour sensing systems", Sensor Review, Vol. 25 Iss: 2, pp.125 - 133|
|Keywords:||Colour, Computer peripheral equipment, Neural nets, Sensors|
|Article type:||General review|
|DOI:||10.1108/02602280510585718 (Permanent URL)|
|Publisher:||Emerald Group Publishing Limited|
Purpose – A novel sensor system is described that can replace the need to use a touch screen, mouse or keyboard for use with either a thin film transistor (TFT) or cathode ray tube (CRT) screen.
Design/methodology/approach – A tri-colour photodiode in a colour sensor pen (CSP) is used as an input to depict where a user is pointing by detecting a dynamic kaleidoscope of colours flashed onto the screen. The most easily detected colours are placed around the predicted position of the sensor. The output from the sensor is filtered and transmitted to a computer where colour is translated into a corresponding co-ordinate. An artificial neural network (ANN) ensures that optimum colours are placed closest to the sensor by predicting future pointer position.
Findings – The CSP successfully provided an input that depicted where a user was pointing. Sensors tested were not able to differentiate more than 28 colours on a TFT screen but an ANN system made them more accurate. Hue and Saturation were successfully used to model colour in a two-dimensional colour grid as they were unaffected by brightness. The sensor output was successfully converted into a co-ordinate and used to move a cursor. The kaleidoscope of colours was dynamic as it changed size and colour to increase accuracy. The ANN successfully predicted future positions to assist the sensor in making optimum colour detections. The best predictions of future positions were obtained with four and eight hidden neurons and there did not appear to be any significant advantage in training with more than 2,000?epochs. The system has performed satisfactorily for various inputs.
Research limitations/implications – The novel light pen works on TFT and CRT screens. Problems included the time that colours needed to be left on a screen to be detected and the limited number of colours that could reliably be detected. These were overcome by using a dynamic kaleidoscope of colours to improve colour detection and an ANN to predict future pointer positions, based on previous locations. This prediction was used to place optimum colours closest to the sensor in order to increase accuracy. Further testing is taking place to confirm that the sensor can be used at any angle.
Practical implications – The novel light pen type peripheral and intelligent colour sensing system can replace the need to use a touch screen, mouse or keyboard.
Originality/value – The new intelligent colour sensing system works on either a TFT or CRT screen. The use of the colour translation and sensor and the ANN to predict future pointer position are all novel.
Existing customers: login
to access this document
Downloadable; Printable; Owned
HTML, PDF (449kb)
Due to our platform migration, pay-per-view is temporarily unavailable.
To purchase this item please login or register.
Complete and print this form to request this document from your librarian