| Summary: | 碩士 === 淡江大學 === 電機工程學系 === 87 === Cluster analysis is a tool for exploring the underlying structure of the data set to be analyzed and is being applied in a variety of engineering and scientific fields such as computer vision, biology, psychology, medicine and marketing. The aim of cluster analysis is to partition a data set into several clusters such that the degree of similarity is high among members of the same cluster and low between members belonging to different clusters. However, clusters can be of arbitrary shapes and sizes in a multidimensional pattern space. For the time being, clustering algorithms that can deal with all situations are not yet available. Each clustering criterion imposes a certain structure on the data, and if the data happen to conform to the requirements of a particular criterion, the true clusters are recovered. Unfortunately, there is no general guideline for choosing one criterion over the other since we usually do not have prior knowledge about the geometric characteristics of the data.
In this thesis, we propose a modified version of the K-means algorithm and a modified version of the competitive learning to cluster data. The proposed algorithms adopt a new distance measure based on the idea of "symmetry". We intend to trade-off flexibility in clustering data with computational complexity. By employing the new measure, the algorithms are capable of grouping a given data set into a number of clusters of different geometrical structures. Several data sets are tested to illustrate the effectiveness of our proposed algorithms. In addition, the second goal of this thesis is to use the concept of symmetry to detect human faces in a complex background. First, the symmetry of human faces is used to quickly locate all the candidate of human faces with all possible sizes and locations. Then two associate memories are used to decide whether or not a human face exits at the locations. Some experimental results are given and simulation results are very encouraging.
|
|---|
