Aim

The principle aim of this project is to extend the existing analogue processing algorithms to improve their performance. Specifically, this thesis will going to examine the use of resistive fuse grids for image smoothing and segmentation. Such resistive fuse grids have been extensively analysed in the past [17]. This project uses an extended model of interconnectivity to improve the performance of the resistive grid.

The performance of a resistive fuse grid can be defined as its ability to take an image and define regions of the image as belonging to discrete objects, in a manner as close as possible to the operation of a human eye. Ultimately, the system should be able to construct some high-level internal representation of the scene it is viewing, and to do this it needs to move from a pixel-by-pixel representation to an object-by-object representation. This is the purpose of the resistive fuse grid circuit. The circuit should, as far as possible, use this information to eliminate any noise that was introduced into the image by the image capture elements.

It is hoped that by designing such a circuit it may be possible to improve designs in such fields as machine and robot vision. It may also suggest insights into the operation of human vision. The ultimate goal of the Visual Communications Research Group is to design a video compression system that can operate in real time for a relatively low cost. This will be necessary in the near future because the world is switching to analogue rather than digital transmission systems, but so far no affordable technology has been developed that is the equivalent of analogue video cameras. An effective and efficient segmentation system will be useful for the design of this system. Furthermore, effective segmentation is useful for other types of image analysis [10].