Bayesian classification of eigencells
Saeid Sanei, Tim K. Lee · Proceedings - International Conference on Image Processing · 2003
A new method for identification of blood cells based on the Bayesian classification of eigencells is introduced. The work by M. Turk and A. Pentland on face recognition (see J. Cognitive Neuroscience, vol.3, no.1, p.71-86, 1991) has been modified and used for cell recognition. Their method lacks a suitable means of eigenvector selection. Also, only monochrome images have been considered and the method is not tolerant enough to geometrical changes. We extend the idea to colour patterns. A fast method in size and rotation adjustment preprocesses the images, the eigencells are selected based on minimisation of similarities among various sets and, finally, a classifier identifies cell types by looking at three-fold intensity-colour information. This overcomes many problems in cell classification where either certain cells are recognised or some constraints such as geometrical variations are involved.