Segmentation Evaluation of Salient Object Extraction Using K-Means and Fuzzy C-Means Clustering

The purpose of current paper is to compare K-Means (KM) clustering and Fuzzy C-Means (FCM) clustering accuracy and performance for segmentation of salient object in the image. The proposed approaches initially emphasize on preprocessing of the image for image refinement. Then use of Discreet Wavelet Transform (DWT) and statistical parameter for extracting the features in an image is investigated. The image is then subdivided into similar areas using the hard K means and Fuzzy clustering approaches. The algorithms are tested on various images from the Berkeley database. The segmentation images obtained are assessed with the help of ground truth and the results show that FCM sensitivity is more than K-means. The convergence and execution time for each algorithm is also analyzed and the results are compared with one another.