House Semi-automatic Extraction Based on Integration of Geometrical Constraints and Image Segmentation

In this paper, a new approach was introduced to extract house from aerial image.This approach can reduce the interaction of operator using the semi-automatic extraction that integrates geometrical constraints and image segmentation.We use segmentation to find boundaries of building whose shape is a single or a composition of rectangular parallelepipeds (thus allowing "L", "T" and "I" shapes for examples).The segmentation is based on line extraction.Therefore, the house edge can be acquired exactly. The edge of building roof, whose shape likes a "人"-a Chinese character, can be also found through analyzing its structure.This analysis is based on result of segmentation of the image.The experimental results show that the methods introduced in this paper are practical and improve the work efficiency.