Rectangular features extraction from aerial image of urban area

Deshmukh Nilesh K., Khamitkar Santosh D., Bhalchandra Parag U.


We purpose a model based approach to extraction of rectangular features of aerial image of urban area. We focus on reconstruction strategy that is restricted to constant rectangle which is defined by its interior’s property of having constant or uniform intensity pixel. Examples of such rectangles appearing in input imagery are building’s roofs, car parks and field boundaries. Extractions of geometrical objects are easier due to availability of many well known geometrical techniques. The model proposed in this paper makes use of the basic image processing techniques. Noise removal and image sharpening techniques are used to boost the input image. Then, the edges are extracted from the image using the Canny edge detection technique. The edges obtained are composed of discrete points. These discrete points are vectorized in order to produce straight line segments. This is performed with the use of the Hough transform and the perceptual grouping techniques. The straight line segments become the basic structures of the buildings. The straight line segments are grouped to apply segment pair matching technique. Finally, rectangles are constructed by using harries corner detection technique. In addition to rectangle construction, the rectangle size and the parameters used in the Hough transform and the perceptual grouping stages also affect the success of the proposed method. Due to the explicit representation of well defined processing states in terms of model based rectangle descriptions at all levels of modeling and data aggregation, our approach reveals a great potential for reliable rectangle extraction for building extraction



Keywords: Rectangle extraction; Building extraction; Canny edge detection; Harris corner detection; Noise removal techniques; Morphological filter; Smoothening

Full Text:




  • There are currently no refbacks.

Copyright (c) 2016 International Journal of Advanced Research in Computer Science