Figure 1: Location of the study area and its subareas
Firstly, lacunariry values were extracted from 168 QuickBird image samples of this area, with 75 x 75 meters in size and 0.60 meters in spatial resolution. Then, these values were assigned to georeferenced centroid points of each image sample and interpolated by Ordinary Kriging. Finally, the generated image was classified through a K-Means algorithm. The lacunarity analysis was run considering 5 different box sizes, and the classified images with each box size were overlaid to describe the morphological pattern through several scales.
Figure 2: Image generated through the overlay of the 5 classified images
The results showed that the lacunarity values were very sensitive to the texture variation of the selected images. Image samples with high lacunarity correspond to intra-urban areas composed of relatively big buildings and large roads, whereas image samples with low lacunarity correspond to denser and more compact intra-urban areas composed by small buildings, narrow and tortuous alleys. In the end it was possible to detect different intra-urban morphological patterns between the formal and informal areas, but within the informal areas as well.
The paper was presented on the 11th International Conference on Computers in Urban Planning and Urban Management - CUPUM which was held from 16 to 18 June 2009, in Hong Kong.
