Extraction and classification of beach features based on object-oriented analysis from video-images

Abstract

In this work a methodology for extraction and classification of coastal features from video images is applied and tested. The approach is based on object-oriented analysis applied over a time series of video images to identify different coastal features on the beach. The identification and quantification of morphological features and some beach parameters are extracted from the classification obtained by segmentation techniques and fuzzy classification. The images were acquired by an Argus system (WL|Delft Hydraulics) installed on Trafalgar beach (Spain) and were geometrically corrected and rectified. From visual interpretation and beach knowledge, some features were identified to generate rules of interpretation for each one of the resulting classes. The class descriptions were done using a fuzzy approach of nearest neighbor on object features, defined by membership functions based on representative samples. Different criteria for the attributes and different intrinsic object variables were used to identify the beach features.