Automated urban area building extraction from high resolution stereo imagery

摘要

This paper describes a new automated building extraction algorithm developed for high resolution stereo imagery. The problems of urban area imagery for stereo matching were investigated. Buildings were found to create isolated regions and many blunders. To overcome the problem of isolated regions, a pyramidal matching algorithm with automatic seed points using a tile-based control strategy was developed. To remove possible blunders and extract buildings from other background objects, a series of 'smart' operations using linear elements from buildings was applied. A quantitative analysis of the accuracy of the algorithm was assessed in comparison with a previous algorithm and shown to have an RMS error of 3.65m in elevation compared to 13.46m.