An Efficient Color Image Segmentation Using Texture Features and Improved Saliency Map
M. Sivasubramanian1, M. Sivajothi2, P. Kumar3
1M. Sivasubramanian, Research Scholar, Centre for Information Technology and Engineering, Manonmaniam Sundaranar University, Abishekapatti, Tirunelveli (Tamil Nadu), India.
2Dr. M. Sivajothi, Associate Professor, Department of Computer Science, Sri Parasakthi College for Women, Courtallam (Tamil Nadu), India.
3Dr. P. Kumar, Assistant Professor, Centre for Information Technology and Engineering, Manonmaniam Sundaranar University, Abishekapatti, Tirunelveli (Tamil Nadu), India.
Manuscript received on 22 April 2019 | Revised Manuscript received on 01 May 2019 | Manuscript Published on 08 May 2019 | PP: 52-56 | Volume-7 Issue-5S3 February 2019 | Retrieval Number: E12000275S19/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Image segmentation is the process of partitioning an image into many regions based on some characteristics like color, texture and intensity. It plays a very important role in image analysis. This paper presents an efficient technique for color image segmentation. Proposed technique utilizes Integer Wavelet Transform (IWT) and Self-Organizing Map (SOM) based Enhanced Adaptive Kernelized FCM (EAKFCM) algorithm. Low frequency components of color image are extracted using IWT. Five texture features are derived from the low frequency components. Improved saliency map is calculated. Texture features and ISM are used as an input to the SOM which is an unsupervised neural network. The segmentation of homogenous regions is obtained employing EAKFCM algorithm. Proposed segmentation technique is tested on natural images and Berkeley segmentation dataset. Efficiency of the proposed technique is measured using five widely used statistical measures such as precision, accuracy, recall, entropy and time. Results demonstrated that the efficiency of the proposed color image segmentation technique is superior to other methods in terms of evaluation metrics.
Keywords: Color Image Segmentation, Clustering, Lifting Scheme, Integer Wavelet Transform, Self-organizing Map, Improved Saliency Map.
Scope of the Article: Image Security