![]() | Only 14 pages are availabe for public view |
Abstract Cluster analysis is an unsupervised learning technique, Clustering techniques have been applied to a wide variety of research problems, in the field of medicine, clustering diseases, cures for diseases, or symptoms of diseases can lead to very useful taxonomies. In the field of psychiatry, the correct diagnosis of clusters of symptoms such as paranoia, schizophrenia, etc. is essential for successful therapy. In archeology, researchers have attempted to establish taxonomies of stone tools, funeral objects, etc. by applying cluster analytic techniques. In general, whenever one needs to classify a ”mountain” of information into manageable meaningful piles, cluster analysis is of great utility. The main objective of this thesis is to study and evaluate some of the cluster analysis algorithms dedicated for digital image analysis. Cluster analysis techniqu s are a diverse collection of techniques that can be used to classify objects. The classification has the effect of reducing the dimensionality of a data table by reducing the number of rows (cases).are studied and compared various agglomerative and partitioning approaches. The underlying mathematics of most of these methods is relatively simple but requires a large number of calculations which needs an extensive computing power. Each classification technique is based upon a particular method, as it is possible to measure similarity and dissimilarity in a number of ways. Consequently there isn’t a special classification technique that may be considered as the sole correct technique, although there have been attempts to define concepts such as ’optimal’ classification. In the present work, the techniques implemented are the K-means and !SODATA techniques that are applied to nine images to classify these images, record the standard deviation of each class and examine results of both classifiers. The K means technique proved to be a very simple algorithm, whereas the !SODATA technique is highly successful at finding the spectral clusters that are inherent in the data. The ISODATA algorithm is similar to the k-means algorithm with the distinct difference that the ISODATA algorithm allows for different number of clusters while the k-means assumes that the number of clusters is known a priori, ISODATA is method includes merging clusters if they are too small and splitting clusters that are too large. So it does not need to know priori exactly number of clusters, it is enough to give a range of detected values. So the I SODAT A is considered advanced version of the k-means but it is time consuming. |