Temporal Data Mining

112    ◾    Temporal Data Mining

In [Nie06], multimedia data, such as images, are clustered by convert-

ing them to time series data. is results in signicant compression of the

data storage requirements. e shapes of objects in the image, such as a

face or a leaf, are extracted and then transformed into a time series. e

time series are clustered using a K-Medoids algorithm and the Euclidean

and Dynamic Time Warping distance measures are utilized. e images

are also clustered using the K-Medoids algorithm on the color histograms

of the images. e results showed that for most images, the time series

representation yields a higher maximum