Building detection from google map.
Maung Naing Lin Tun.
Date of Issue2009
School of Electrical and Electronic Engineering
In urban areas, buildings are often used as landmarks for localization. Reliable and efficient recognition of building is crucial for enabling this localization functionality. This FYP project represents a content-based image retrieval system for buildings images. In this context, the ultimate aim is to develop a simple building retrieval system that can be implemented in any imaging devices. The first stage of the building recognition is to detect the building using consistent-line clustering technique. Edges of the image are clustered according to the color and spatial features. Line segments from other non-building objects are eliminated in this stage and the final consistent-line cluster has been used to represent the main building. In the second stage, the color feature of the building are extracted using the consistent-line cluster obtained in first stage. The color indexing vector has been constructed using the colors from left and right region of each line segment. Finally, for effective image retrieval, a simple matching function is defined based on the Chi-square of measure of similarity between the indexing vectors from query image and database images. The performance of the retrieval system is evaluated using an image database of 30 buildings taken from the publicly available database, ZuBuD, and 150 query images have been used to test the system.
DRNTU::Engineering::Electrical and electronic engineering
Final Year Project (FYP)
Nanyang Technological University