Search

The automation in man-made object extraction such as building habitation from urban area imagery has become a challenging task for photogrammetry, computer vision, and remote sensing. This study aims to automatically extract building of an urban area using high resolution intensity data and fuzzy membership logic to classify the image object by using e-Cognition software. The object oriented method was implemented and high resolution Quick-Bird imagery was used for automatic building extraction in Dehradun city of Uttarakhand district, India. We have further evaluated the performance of this automated extracted building feature by using accuracy completeness (89.74%), correctness (94.50%), and the quality (85.29%). The study however, clearly shows that the segmentation-based classification is much superior to the traditional per-pixel methods mainly used on high resolution images. It also shows that high spatial resolution satellite data and appropriate data processing play not only an important role in modern urban planning but also reduce the cost of manpower and saves time.