Optimized Segmentation of Oil Spills from SAR Images Using Adaptive Fuzzy K-Means Level Set Formulation

Abstract

With the increasing amount and complexity of remote sensing image data and the difficulties faced in processing the data, the development of large-scale image segmentation analysis algorithms could not keep pace with the need for methods that improve the final accuracy of object recognition. So, the development of such methods for large-scale images poses a great challenge nowadays. Traditional level set segmentation methods which are Chan-Vese (CV), image and vision computing (IVC) 2010, ACM with SBGFRLS and online region-based ACM (ORACM) were suffered with more amounts of time complexity, as well as low segmentation accuracy due to the large intensity homogeneities and the noise. The robust region-based segmentation is impossible in remote sensing images is a tedious task because due to lack of spatial information and pixel intensities are non-homogenous. For this reason, we proposed a novel hybrid approach called adaptive particle swarm optimization (PSO)-based Fuzzy K-Means clustering algorithm. The proposed approach is diversified into two stages: in stage one, pre-processing the input image to improve the clustering efficiency and overcome the obstacles present in traditional methods by using particle swarm optimization (PSO) and Fuzzy K-Means clustering algorithm. With the help of PSO algorithm, we get the “optimum” pixels values that are extracted from the input SAR images; these optimum values are automatically acted as clusters centers for Fuzzy K-Means clustering instead of random initialization from original image. The pre-processing segmentation result improved the clustering efficiency but suffers from few drawbacks such as boundary leakages and outlier’s even particle swarm optimization is used. To overcome the above drawbacks, post-processing is necessary to facilitate the superior segmentation results by using level set method. Level set method utilizes an efficient curve deformation is driven by external and internal forces in order to capture the important structures (usual edges) in an image as well as curve with minimal energy function is defined. The combined approach of both pre-processing and post-processing is called as Adaptive Particle Swarm Optimization-based Fuzzy K-Means (AFKM) clustering via level set method. The proposed method is successfully implemented on large-scale remote sensing imagery, and the dataset are taken from the open source NASA earth observatory database for segmenting the oil slicker creeps, oil slicker regions, etc. So here in this, the proposed new hybrid method had feasibility and the efficiency which could attain the high accurate segmentation results when compared with traditional level set methods …