datasets/Aerial64x64Tiles/FeatureVectors/NormalizedTexture/index.htm

Home
People
Research
Publications
Courses
Seminars
Links

 

Aerial photos ( 64 x64 tiles )

Each large aerial photo is divided into 64x64 tiles. The total number of tiles is 275,465.

These texture feature vectors are extracted from the 64x64 pixel tiles of 40 large aerial images. The total number of tiles is 275,465. The feature vectors have dimension 60. They are the mean and standard deviations of 30 Gabors filters (5 scales and 6 orientations). The exact ordering of the elements is not known. There are 40 feature files. 

 

The features are stored in a text file (byte swapping done). There are raw and normalized versions of the file: 

Aerial40.txt Aerial_norm40.txt Extract_feature.c

 

The features are stored as floats in binary format, one feature at a time. The features were extracted on a non-Intel machine so will need to be byte-swapped at the float level if read on an intel machine. These are normalized feature file.  Normalized feature vector list follows:

c209_a1   haan1_30   hbiu_141 
c209_a4   haan1_31  hbiu_142 
c209_b1   haan1_32  hbiu_143 
c209_b3  haan1_33  pwsb1_18
c209_c1   haan1_34  pwsb1_19 
c209_c5   haan1_126 pwsb1_20 
c209_d1 haan1_127 pwsb1_21 
c209_d3   haan1_128 pwsb1_34 
c209_d5   haan1_129 pwsb1_35 

c4950_sf18  

haan1_130 pwsb1_36 
c4950_sf19  

haoi_88 

pwsb1_37 
c4950_sf20 haoi_90 
gsem6_105 haoi_91  fea_mean
gsem6_106 haoi_92  fea_std
gsem6_107  
Aerial photos ( 128 x 128 tiles )