Analog and Digital Signal Processing in an Optical Application

The thesis consists of two parts. Registration of Digital Images: Registration of images is a fundamental task in image processing. This is to match two or more images taken, for example, at different times, from different sensors or from different points of view. The methods used to register images are striving for a similarity metric indicating when the images are correctly registered. A similarity metric gives a measure of the degree of similarity between an image and a template. The approaches to find a good metric have been done looking at statistics in the images, e.g. correlation between them. A new similarity metric, the Focus Similarity Metric, invented by personnel at CellaVision AB is here being introduced. There are patents pending. Below, it will be shown that it is a very robust algorithm. It is a similarity metric based on a focus function, i.e. two added images are most aligned when they, together, are as focused as possible according to a focus function. Given some tuning, the images can be registered using the Focus Similarity Metric if the differ in scale or have been translated or rotated. They can even have been manipulated and noisy. The metric is also insensitive to intensity differences between the images and gives a very well defined maximum. Furthermore, it is fast and easy to implement. Control of CD-pickup: A CD-pickup is used to focus for example a diode laser beam at a CD-record. The pickup is moved using magnetic coils. It has a small mass, which makes it very sensitive to external disturbances. In this application the problem is to control the position of the pickup insensitive to disturbances. In this application the problem is to control the position of the pickup insensitive to disturbances. In the thesis identification and analysis of the pickup is made. A controller is selected and realized with an analog circuit