An improved Chebyshev distance metric for clustering medical images

A metric or distance function is a function which defines a distance between elements of a set. In clustering, measuring the similarity between objects has become an important issue. In practice, there are various similarity measures used and this includes the Euclidean, Manhattan and Minkowski. In this paper, an improved Chebyshev similarity measure is introduced to replace existing metrics (such as Euclidean and standard Chebyshev) in clustering analysis. The proposed measure is later realized in analyzing blood cancer images. Results demonstrate that the proposed measure produces the smallest objective function value and converge at the lowest number of iteration. Hence, it can be concluded that the proposed distance metric contribute in producing better clusters.