Detection of Building Changes from Aerial Imageries Through Information Fusion

1) the density of building is low 2) image resolution is high 3) the color or gray scale for building's surface In this paper, an improved approach of are 'lose to uniform detecting building changes with aerial 4) the shape of building's roof is of polygon imageries has been proposed. hi^ approach But since most of the cities in Japan are first makes uses of image pixel's color, densely inhabited, the existing approaches are structure's edge information and their hardly applicable. correlation to filter out the candidates at relatively high speed. Then existing vector topographical map information and previously processed data are used for more precise analysis to get rid of unrelated changes. Experiments with actual data shows that the proposed approach has obtained success rate of above 90% while keeping the retrieval rate above 70%.

• Type: Conference
• Year: 1998
• Where: MVA
• Authors: Wei Lu, Takeshi Doihara, Yoshino Matsumoto
• Keywords: Building Changes, Building's Roof, Improved Approach, MVA 1998, MVA 2007