Texture Analysis



Nohelia Gonzalez

Han Jin Yang

Ekaterina Levkina

Saad Saheen


The aim of the texture analysis was to be able to automatically identify different types of vegetation, bare lands, agricultural areas, among others from the 3d model generated.

Different filters were applied to the aerial image in order to increase the differentiation between the textures and colors on the image.

2_Big Scale analysis

By using a grasshopper script, we were able to choose a color and build a mesh out of the areas with similar shade (controlled by a threshold). However the filters previously applied were not enough to make a recognizable difference and the resulting mesh was not able to cluster valuable information.


3_Big Scale results filter 1

We moved forward to a more specific image processing  algorithm by using the OpenCV framework. We realized that the different types of vegetation were more easily differentiated in the HSV (Hue, saturation, value) color space than in the original RGB space.  Therefore, we transformed the image to the HSV space and, applying a equalization of its histogram, we were able to use a thresholding operator to successfully identify the vegetation. Some postprocessing filters were used to remove noisy points (e.g. erosion, dilation and blur). In order to identify the paths, we applied a clustering algorithm that aggregates pixels with the same colors and applied the aforementioned postprocessing techniques to remove noisy blobs.

4_Big Scale Filtering_Light

In a big scale we were able to identify 2 vegetation colors and paths.

5_Big Scale results Light_Filter2 6_Big Scale Filtering_Paths 7_Big Scale results Path_Filter2 8_Big Scale Filtering_dark 9_Big Scale results Dark_Filter2

We chose an area for filtering in a higher resolution and from this one we were able to find 3 vegetation colors and the paths. Bare land areas were not automatically found.

Finally, the results were displayed on a particle-like mesh and projected on the original one.

10_Zoom scale_Section selection 11_Zoom scale_Path_Filter2 12_Zoom scale_path_results 13_Zoom scale_Dark green_Filter2 14_Zoom scale_dark green_results 15_Zoom scale_mid green_Filter2 16_Zoom scale_mid green_results 17_Zoom scale_Light green_Filter2 18_Zoom scale_Light green_results 19_Zoom scale_Results combined


This entry was posted in Uncategorized. Bookmark the permalink. Both comments and trackbacks are currently closed.