Comparative Analysis for the Detection of Marine Vessels from Satellite Images Using FCM and Marker-Controlled Watershed Segmentation Algorithm

The security of maritime activity is enhanced by the detection of marine vessels. Satellite images are used to detect the marine vessels irrespective of extreme weather conditions. Marine vessels can be detected efficiently using image segmentation algorithms. Many researchers have applied Haar-like classifier, convolution neural network, artificial neural network techniques to detect the marine vessels. In this work two different methodologies such as fuzzy C means (FCM) and marker-controlled watershed segmentation algorithms are developed and demonstrated to detect the marine vessels from satellite images. The marker-controlled watershed algorithm can effectively visualize an image in three dimensions and easily segments three-dimensional images. On the other hand, the number of iterations needed to achieve a specific clustering exercise in FCM is very less. It calculates the distance between the pixels and the cluster centres in the spectral domain to calculate the membership function. Experiments are carried out using IKONOS image of 4-m resolution. The average users accuracy of FCM algorithm and marker-controlled watershed algorithm is 91.29% and 95.79%, respectively. The results obtained show that there is an increase in accuracy for marker-controlled watershed algorithm when compared to FCM algorithm.