An Automated Approach to Extracting River Bank Locations from Aerial Imagery Using Image Texture

A fundamental challenge in river analysis and modelling is the lack of readily available and reliable information on river bank geometry. Traditional survey methods are expensive and time consuming and often difficult to execute in many river systems because of hazardous terrain or lack of access. However, as high quality aerial and satellite imagery becomes available for more of the globe, it is increasingly possible to extract these bank locations directly from imagery. The most direct method of doing this involves manually designating edges based on visual criterion. This, however, is often prohibitively time consuming and labour intensive, and the quality is dependent on the individual doing the task. This paper describes a quick and fully automated method for locating water surface and river banks in high resolution aerial imagery without recourse to any multispectral information, by segmenting based on the local entropy of the image. This method is demonstrated on imagery of several rivers and its advantages and limitations are discussed. Published 2013.