Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

Segmentation of Hyperspectral Image Using JSEG Based on Unsupervised Clustering Algorithms


Affiliations
1 Centre for Bioinformatics, Pondicherry University, India
2 Department of Computer Science and Engineering, Hindustan University, India
     

   Subscribe/Renew Journal


Hyperspectral image analysis is a complicated and challenging task due to the inherent nature of the image. The main aim of this work is to segment the object in hyperspectral scene using image processing technique. This paper address a novel approach entitled as Segmentation of hyperspectral image using JSEG based on unsupervised cluster methods. In the preprocessing part, single band is picked out from the hyperspectral image and then converts into false color image. The JSEG algorithm is segregate the false color image properly without manual parameter adjustment. The segmentation has carried in two major stages. To begin with, colors in the image are quantized to represent several classes which can be used to differentiate regions in the image. Besides, hit rate regions with cognate color regions merging algorithm is used. In region merging part, K-means, Fuzzy C-Means (FCM) and Fast K-Means weighted option (FWKM) algorithm are used to segregate the image in accordance with the color for each cluster and its neighborhoods. Experiment results of above clustering method could be analyzed in terms of mean, standard deviation, number of cluster, number of pixels, time taken, number of objects occur in the resultant image. FWKM algorithm results yields good performance than its counterparts.

Keywords

Cluster, Region Growing, Hit Ratio Region, Class-Map, Quantize.
Subscription Login to verify subscription
User
Notifications
Font Size

Abstract Views: 239

PDF Views: 2




  • Segmentation of Hyperspectral Image Using JSEG Based on Unsupervised Clustering Algorithms

Abstract Views: 239  |  PDF Views: 2

Authors

V. Saravana Kumar
Centre for Bioinformatics, Pondicherry University, India
E. R. Naganathan
Department of Computer Science and Engineering, Hindustan University, India

Abstract


Hyperspectral image analysis is a complicated and challenging task due to the inherent nature of the image. The main aim of this work is to segment the object in hyperspectral scene using image processing technique. This paper address a novel approach entitled as Segmentation of hyperspectral image using JSEG based on unsupervised cluster methods. In the preprocessing part, single band is picked out from the hyperspectral image and then converts into false color image. The JSEG algorithm is segregate the false color image properly without manual parameter adjustment. The segmentation has carried in two major stages. To begin with, colors in the image are quantized to represent several classes which can be used to differentiate regions in the image. Besides, hit rate regions with cognate color regions merging algorithm is used. In region merging part, K-means, Fuzzy C-Means (FCM) and Fast K-Means weighted option (FWKM) algorithm are used to segregate the image in accordance with the color for each cluster and its neighborhoods. Experiment results of above clustering method could be analyzed in terms of mean, standard deviation, number of cluster, number of pixels, time taken, number of objects occur in the resultant image. FWKM algorithm results yields good performance than its counterparts.

Keywords


Cluster, Region Growing, Hit Ratio Region, Class-Map, Quantize.