A Basic Study on Detection of Debris of Damaged Area Based on Outline Information of Vertical Aerial Photographs

Abstract

This paper deals with automated detection method of collapsed wooden house areas based on outline information of aerial photographs taken just after earthquakes. In addition to original outline images, thinning and thickening of those images were made. These three kinds of images were divided into square meshes with proper sizes respectively. Fractional dimension, number of pixels which compose outlines and number of elements divided by outlines were considered as characteristic parameters of outline of each mesh. From analytical results, it was found that debris areas were accurately detected based on the fractional dimension of outline images thickened once.