An Enhanced Algorithm for Mining Color Images – A Novel Approach

Abstract

Image mining is not mere an extension of data mining to image domain. Image mining is a technique normally used to extract knowledge and recognize objects directly from images. Image segmentation will normally be the first step in image mining. Image segmentation is difficult, but it is important problem in computer vision and machine perception. We can treat image segmentation as graph partitioning problem. The minimum spanning tree algorithm is capable of detecting clusters with irregular boundaries to mine images. This paper proposes the minimum spanning tree based clustering algorithm to detect color images using weighted Euclidean distance for edges, which is key element in building the graph from image. The algorithm produces n clusters with segments. An important characteristic of this method is its capacity to conserve information in low variability image regions while omitting detail in high-variability regions. The proposed algorithm has been employed using MATLAB. The implemented system produces promising results.