| dc.description.abstract | The advent of multimedia computing has led to increased demands for digital images. Large number of people has not been satisfied with the result of image that has been retrieved. Image clustering is one of the crucial process in content-based image retrieval (CBIR). This research implement Self-Organizing Map (SOM) for clustering , not only by color-based image retrieval but also combine it with shape-based image retrieval. As a comparison, this research compare method without clustering with method that using SOM as a clustering method. Recall and precision is used to evaluate the retrieval performaces. According to this research, cluster method SOM for color-based image retrieval has average precision 88.98 %, while method without cluster has 75%. The better result is accomplished by combining the color-based with shape-based image retrieval. By using the combination, average precision for cluster method SOM has increased to 89.25 %. | |
