Roland's Master Thesis Page (Mathematics)

"Neural Network and Gradient Based Edge Detection"

Abstract:

Edge detection is one of the essential methods in early vision of the human visual system (HVS) to extract important information from the projected image of the environment. Among all the methods which have been developed for the purpose of simulating edge detection of the HVS on the computer there are two interesting approaches. First the derivative based methods which use the first and second -- directional and direction invariant -- derivative to extract intensity information from the image. The second method is based on artificial neural networks and proofs to be a powerful, biologically inspired tool for the extraction of grey level discontinuities. To reduce the amount of information to be processed, the output from the edge detection methods has to be transformed into a representation which has the same information contents but less superfluous information. For this purpose a thinning algorithm based on mathematical morphological methods is applied. In our thesis we have examined the two methods described above for edge detection in grey level images. Furthermore the overall thinning algorithm is discussed in detail.

Literature on Neural Network and Gradient Based Edge Detection

• Articles Bibliography by Roland Schwaiger (.bib)