Search for Articles:

Recently Downloaded: Paper ID: ART20194760, Total 5765 Articles Downloaded Today

Hybrid Image Segmentation Using Mean Shift and Predicate Algorithms

Amit Patel, Roopesh Kumar Yadav, Kavitha Ambrelli

Abstract: This paper addresses the problem of image segmentation by combing different segmentation algorithms in such a way that the hybrid approach results in feasible segmentation of digital images. Segmentation is generally the first stage in any attempt to analyze or interpret an image automatically. Various segmentation algorithms like graph based, cluster based, mean shift based, intensity based, discontinuity based and predicate based are proposed so as to achieve image segmentation process. Each algorithm finds application in versatile real time problems. Although there are many segmentation algorithms, graph based, mean shift based and cluster based methods are efficient and easier to implement. As part of our work we considered hybrid combination of mean shift and predicate using algorithms. The results of mean shift segmentation are then processed by normal cuts and predicate using algorithm and the results are compared. Mean shift segmentation is robust feature space analysis algorithm capable of storing and saving the discontinuity, edge preservation of image features, it smoothens the image after segmentation. Predicate based algorithms use a function that can be again a segmentation method or a logical predicate based on the image features. Normalized cuts segmentation is the most widely used segmentation algorithm and is parameter sensitive. Cluster based algorithms clusters similar features of image to segment the image

Keywords: Hybrid segmentation, Mean shift, Normalized cuts, Predicate using Mean shift

Country: India, Subject Area: Computer Engineering

Pages: 696 - 701

Edition: Volume 8 Issue 3, March 2019

How to Cite this Article?

Amit Patel, Roopesh Kumar Yadav, Kavitha Ambrelli, "Hybrid Image Segmentation Using Mean Shift and Predicate Algorithms", International Journal of Science and Research (IJSR),, Volume 8 Issue 3, March 2019, 696 - 701

Download PDF

Viewed 68 times.

Downloaded 21 times.