Due to incredible growth of high dimensional dataset, conventional data base querying methods are inadequate to extract useful information, so researchers nowadays is forced to develop new techniques to meet the raised requirements. Such large expression data gives rise to a number of new computational challenges not only due to the increase in number of data objects but also due to the increase in number of features/attributes. Hence, to improve the efficiency and accuracy of mining task on high dimensional data, the data must be preprocessed by an efficient dimensionality reduction method. Recently cluster analysis is a popularly used data analysis method in number of areas. K-means is a well known partitioning based clustering technique that attempts to find a user specified number of clusters represented by their centroids. But its output is quite sensitive to initial positions of cluster centers. Again, the number of distance calculations increases exponentially with the increase of the dimensionality of the data. Hence, in this paper we proposed to use the Principal Component Analysis (PCA) method as a first phase for K-means clustering which will simplify the analysis and visualization of multi dimensional data set. Here also, we have proposed a new method to find the initial centroids to make the algorithm more effective and efficient. By comparing the result of original and new approach, it was found that the results obtained are more accurate, easy to understand and above all the time taken to process the data was substantially reduced.

Criterion based thresholding algorithms are simple and effective for two-level thresholding. However, if a multilevel thresholding is needed, the computational complexity will exponentially increase and the performance may become unreliable. In this approach, a novel and more effective method is used for multilevel thresholding by taking hierarchical cluster organization into account. Developing a dendogram of gray levels in the histogram of an image, based on the similarity measure which involves the inter-class variance of the clusters to be merged and the intra-class variance of the new merged cluster . The bottom-up generation of clusters employing a dendogram by the proposed method yields good separation of the clusters and obtains a robust estimate of the threshold. Such cluster organization will yield a clear separation between object and background even for the case of nearly unimodal or multimodal histogram. Since the hierarchical clustering method performs an iterative merging operation, it is extended to multilevel thresholding problem by eliminating grouping of clusters when the pixel values are obtained from the expected numbers of clusters. This paper gives a comparison on Otsu’s & Kwon’s criterion with hierarchical based multi-level thresholding.