Measurement, modeling and visualization of the shape of the cornea
F.M. Vos et al.

Section 1: Introduction

Available instruments employing this technique are known not to work ideally (antalis93). We have developed a prototype stimulus that exploits the virtues of Pseudo Random Binary Arrays (PRBA's) (vos96), which, encoded in a coloured stimulus-pattern, contributes to a very robust measurement-technique with uniquely characterized points. Exploiting this approach we are able to overcome the shortcomings of traditional devices. A more detailed description of the instrument is given in vos96.

The prototype-instrument devised by us is depicted in figure 2. It consists of a cylinder closed at one side, containing a pattern (called stimulus or stimulator) brightly lit from the back. The object to be measured is positioned in the front of the open end; the reflected image is registered with a camera.

The essence of the novel technique lies in the use of uniquely characterized positions both on the stimulus and in the recorded reflection. In this way a one-to-one correspondence between points on the stimulus and on the registering device is established which can be used to reconstruct the corneal shape.

In this paper we will discuss various aspects of the instrument design especially in relation to the added value of visualization of (intermediate) results. The measurement process consists of three steps:

1. First the actual image is registered. For subsequent steps it is of crucial importance to have an accurately focussed image. This requires a real-time focussing procedure. We will not discuss details of this. It suffices here to say that we have developed a set-up that runs on a dedicated PC, grabs in real-time images and calculates a measure-of-focus (see e.g. Firestone91).
   Of this series the one with the highest focus value is stored. The result of the cycle is the creation of an TIFF image which serves as input to the rest of the analysis software.
2. Extraction of the graph and the PRBA (color) information from the image makes up the second step of the process. This is discussed in section two of this paper and falls apart into two sub problems. First the positions of the grid have to be recovered and then to be connected into a graph.
3. The third step is the modeling of the corneal surface based on the information thus deduced. Some elements of this process will be discussed in section three of this paper.