Supervised classification and improved filtering method for shoreline detection

Abstract

Shoreline monitoring is important to overcome the problems in the measurement of the shoreline. Recently, many researchers have directed attention to methods of predicting shoreline changes by the use of multispectral images. However, the images being captured tend to have several problems due to the weather. Therefore, identification of multi class features which includes vegetation and shoreline using multispectral satellite image is one of the challenges encountered in the detection of shoreline. An efficient framework using the near infrared–histogram equalisation and improved filtering method is proposed to enhance the detection of the shoreline in Tanjung Piai, Malaysia, by using SPOT-5 images. Sub-pixel edge detection and the Wallis filter are used to compute the edge location with the subpixel accuracy and reduce the noise. Then, the image undergoes image classification process by using Support Vector Machine. The proposed method performed more effectively and reliable in preserving the missing line of the shoreline edge in the SPOT-5 images.