| 1 |
"""
|
| 2 |
|
| 3 |
Document Oculus
|
| 4 |
|
| 5 |
Created by
|
| 6 |
Emanuel Borges
|
| 7 |
02.21.2015
|
| 8 |
|
| 9 |
This program will help in the detection of near duplicates in the paper world.
|
| 10 |
|
| 11 |
"""
|
| 12 |
import os, shelve
|
| 13 |
from PIL import Image
|
| 14 |
|
| 15 |
def CreateNewMatrix(imagesPath):
|
| 16 |
matrix = {}
|
| 17 |
for root, dirs,files in os.walk(imagesPath):
|
| 18 |
for name in files:
|
| 19 |
if os.path.splitext(name)[1] == '.tif':
|
| 20 |
image = Image.open(os.path.join(root,name))
|
| 21 |
h = str(imagehash.dhash(image))
|
| 22 |
if h in matrix.keys():
|
| 23 |
matrix[h].append(os.path.join(root,name))
|
| 24 |
else:
|
| 25 |
matrix[h] = [os.path.join(root,name),]
|
| 26 |
|
| 27 |
print "there are %s groups"% str(len(matrix.keys()))
|
| 28 |
return Matrix
|
| 29 |
|
| 30 |
def GetMatrixFromShelve(shelvePath):
|
| 31 |
matrix = shelve.open(shelvePath)
|
| 32 |
## Dont forget to close the file matrix
|
| 33 |
return matrix
|
| 34 |
|
| 35 |
def SaveMatrixToShelve(shelvePath,matrix):
|
| 36 |
s = shelve.open(shelvePath)
|
| 37 |
s.update(matrix)
|
| 38 |
s.close()
|
| 39 |
|
| 40 |
|
| 41 |
if __name__ == '__main__':
|
| 42 |
matrix = {}
|
| 43 |
imagesPath = r'C:\Test\imageTest'
|
| 44 |
|
| 45 |
|
| 46 |
|
| 47 |
for root, dirs,files in os.walk(imagesPath):
|
| 48 |
for name in files:
|
| 49 |
if os.path.splitext(name)[1] == '.tif':
|
| 50 |
image = Image.open(os.path.join(root,name))
|
| 51 |
h = str(imagehash.dhash(image))
|
| 52 |
if h in matrix.keys():
|
| 53 |
matrix[h].append(os.path.join(root,name))
|
| 54 |
else:
|
| 55 |
matrix[h] = [os.path.join(root,name),]
|
| 56 |
|
| 57 |
print "there are %s groups"% str(len(matrix.keys()))
|
| 58 |
|
| 59 |
testList = []
|
| 60 |
for i in matrix.keys():
|
| 61 |
if len(matrix[i]) > 1:
|
| 62 |
testList.append(i)
|
| 63 |
|
| 64 |
print "There are %s groups of 2 or more"% str(len(testList))
|
| 65 |
|
| 66 |
highNo= 1
|
| 67 |
for i in matrix.keys():
|
| 68 |
if len(matrix[i]) > highNo:
|
| 69 |
highNo = len(matrix[i])
|
| 70 |
|
| 71 |
print "The highest group is %s."% str(highNo)
|
| 72 |
|
