NITS IMAGE HASHING DATABASE
Department of Electronics and Communication Engineering
National Institute of Technology Silchar
NITS IMAGE HASHING DATABASE was developed as part of B.Tech/Ph.D project under the Department of Electronics and Communication Engineering at National Institute of Technology, Silchar.
The database consists of 16000 perceptually similar image pairs, 200 visually distinct images, 300 global color changes, 180 local color changes, 646 large tampered pairs, 200 small tampered pairs and 200 images with corner changes. The database was developed with the intention that researchers in the field of image processing specifically working with image hashing and related areas can easily avail these images.
NITS perceptual similar image database
(after applying content-preserving operations)
Database Description: The database consist of 200 images which can be classified in 5 classes-
These images were collected from USC-SIPI, CASIA and Silchar Medical College(SMCS)
We have applied the following content-preserving operations
​
​
These 80 operations have been applied on 200 image. The entire database consist of 200x80 = 16000 perceptually similar images.
The database will be extended to 20,000 images later on.
NITS tampered image database
​
It consists of 646 large tampered image pairs taken from CASIAv2.0[1] dataset which consists of different images of size from 240 x 160 to 900 x 600 and 200 self constructed small tampered images of size from 256x384 to 800x600.
​
Directory Structure
Large Tampered(>65%)
original^ - It consists of 646 original untampered images
tampered^- It consists of 646 tampered images
Filename Structure^
<index>.<fileformat>​
-
index ranges from 164-809
-
fileformat - jpg
Small Tampered(20-65%)
original'- It consists of 200 original untampered images
small tampered '- It consists of 200 small tampered images
Filename Structure'
<index>.<fileformat>​
-
index ranges from 1-200
-
fileformat - jpg
​
​
​
​
​
NITS color changes image database
It consists of 160 source images taken from CASIAv2.0[1]. Out of these 160 images, 100 images have been used to create 300 global color forged images and from remaining 60 source images we have created 180 local color forged images. These 180 local forged images have forged area of sizes 24x24, 32x32 and 64x64 as follows-
​
Forged area size No. of images
24x24 36
32x32 84
64x64 60
Directory Structure-
Global Forgery
original* - It has 100 original images
global forgery** - It has 300 color tampered images (100 red channel, 100 blue channel, 100 green channel)
​
24x24 Local Forgery
original*- It has 36 original images
local forgery** - It has 108 local color tampered images (36 red channel, 36 blue channel, 36 green channel)
32x32 Local Forgery
original* - It has 84 original images
local forgery** - It has 252 local color tampered images (84 red channel, 84 blue channel, 84 green channel)
​
64x64 Local Forgery
original* - It has 60 original images
local forgery** - It has 180 local color tampered images (60 red channel, 60 blue channel, 60 green channel)
Filename Structure *-
Au_ani_<index>.<fileformat>
-
index ranges from 00001-10XXX
-
fileformat - jpg
Filename Structure **-
Au_ani_<index>_<channel>.<fileformat>
-
index ranges from 00001-10XXX
-
fileformat - jpg
-
channel - {r,g,b}
​
​
NITS corner changes image database
It consists of 200 self constructed corner tampered image pairs. Out of these 200 image pairs a few of them have only one/two/three corners tampered while others have all four corners tampered.
​
No. of corners tampered No. of image pairs
One corner 50
Two corners 50
Three corners 50
Four corners 50
​
Directory Structure-
Corner Forgery
original - It has 200 original untampered images
corner forgery - It has 200 corner tampered images
​
Filename Structure-
Au_ani_<index>.<fileformat>
-
index ranges from 00001-00200
-
fileformat - jpg
​
​
NITS visually different image database
It consists of 200 visually distinct images taken from Ground Truth Database[2].
Directory Structure-
Different
Filename Structure-
<index>.<fileformat>
-
index ranges from 1-200
-
fileformat - jpg
​