Although significant research progress has been made in the field of data mining, which aims
to extract essential information implicitly stored in large data archives [1], research in image
information mining (IIM) is still in its infancy. Mining useful information from image archives is
very much an interdisciplinary endeavor that draws upon expertise in image processing, database
organization, pattern recognition, information retrieval, and data mining [2]. Some important
research issues and system frameworks for IIM have been proposed [3], [4]. First, IIM is different
from low-level image processing techniques as it deals with the extraction of useful patterns that
are previously unknown from a large collection of images, referred to as an image database,
whereas image processing generally focuses on extracting and understanding features within a
single image. Second, some overlap exists between the concept of IIM and content-based image
retrieval (CBIR) [5], [6] regarding retrieval of images relevant to user requests from image
databases. CBIR is characterized by the ability of the system to retrieve relevant images based on
their semantic and visual contents rather than by using atomic attributes or keywords assigned to
them. IIM emphasizes the process of discovering significant and potentially useful hidden
patterns from large image databases, whereas the set of relevant images is dynamic, subjective,
and even completely unknown. Therefore, an IIM system should be adaptive and process queries
from the viewpoint of the user’s interpretation of the image content and domain semantics