Waterbody Extraction from Aerial Image Using Clustering Based on SCDAE Approach

Extraction of water bodies from satellite image is essential in a variety of disciplines, including lake coastal zone management, monitoring of erosion and coastal change, flood prediction, and resource evaluation. In this paper, a novel machine learning technique called clustering-based SCDAE method has been proposed which is used to extract the water bodies in the input satellite images. The satellite images are given as input to the pre-processed stage where pre-processing is done by using Scalable Range Adaptive Bilateral Filter (SCRAB) and CLAHE filter. SCRAB filter is used to reduce the noise from the input aerial images. Then, the CLAHE technique is used to enhance the various changes in the images. The Stacked Convolutional Denoising Auto-Encoder (SCDAE) is used to extract the features from the satellite images. Finally, the fuzzy K-means clustering is used to segment the water bodies in the images. The performance of the proposed method has been evaluated in terms of specific parameters such as accuracy, precision, recall, specificity, and F1-score, and the proposed method shows better results than existing techniques. The proposed clustering-based SCDAE method achieves overall accuracy of 98.54%, and the existing methods such as C means clustering, fuzzy K-means, pixel-wise, and KNN clustering achieve 92.45%, 93.56%, 91.37%, and 94.38%, respectively.