14.6 Stamp – Translation Symmetry
The images used in this section are in the directory “.\ex_stamp\”. In this section, we try to identify 2 stamps. Rather than use an existing image to search, we will focus on building a sample image for matching.
14.6.1 Example 1
The first example retrieves images like the following:
We will build a sample image as follows:
There are two ways to run this example:
-
Batch
-
Manual
The Batch Run takes only two clicks:
-
Click “Examples/Neural Net/Stamp 1 - T” ;
-
Click “Batch/Run”.
The Manual Run requires a few more clicks:
Input:
Training: .\ex_stamp\class1.jpg
Search: .\ex_stamp\
Parameters
Edge Filter: (Image Processing---------------1st box)
None
Threshold Filter: (Image Processing---------------2nd box)
Default
Clean Up Filter: (Image Processing---------------3rd box)
None
NeuralNet Filter Parameters: (Neural Net Parameters)
Translation Symmetry
Blurring = 8
Sensitivity = 45
InternalCut = 40 %
Operation
-
Click the "Key Segment" button and select ".\ex_stamp\class 1.jpg";
-
Click the "Source" button and select ".\ex_stamp\";
-
Set the Parameters as specified above;
-
Click the "Neural Net\Train" button to train the filter;
-
Click the "NeuralNet\matching\1:N (key vs Source)" button to make a search.
Results
ID Name Path Score X Y W H
CLASS1 CLASS1.JPG C:\...\ex_stamp\ 128000000 36 24 244 272
CLASS1_1 CLASS1_1.JPG C:\...\ex_stamp\ 39680 64 24 244 272
class1_10 class1_10.jpg C:\...\ex_stamp\ 18112 12 28 244 272
CLASS1_2 CLASS1_2.JPG C:\...\ex_stamp\ 44160 44 20 244 272
CLASS1_3 CLASS1_3.JPG C:\...\ex_stamp\ 19584 4 24 244 272
CLASS1_4 CLASS1_4.JPG C:\...\ex_stamp\ 45120 68 24 244 272
CLASS1_5 CLASS1_5.JPG C:\...\ex_stamp\ 53312 36 24 244 272
CLASS1_6 CLASS1_6.JPG C:\...\ex_stamp\ 43136 68 36 244 272
CLASS1_7 CLASS1_7.JPG C:\...\ex_stamp\ 40960 56 52 244 272
CLASS1_8 CLASS1_8.JPG C:\...\ex_stamp\ 43648 44 44 244 272
CLASS1_9 CLASS1_9.JPG C:\...\ex_stamp\ 28544 88 24 244 272
Summary
# Images = 104
# To be retrieved = 11
# Retrieved Correctly = 11
# Missed = 0
Hit Ratio = 100%
14.6.2 Example 2
The second example retrieves images like the following:
We will build a sample image as follows:
There are two ways to run this example:
-
Batch
-
Manual
The Batch Run takes only two clicks:
-
Click “Examples/Neural Net/Stamp 2 - T” ;
-
Click “Batch/Run”.
The Manual Run requires a few more clicks:
Input:
Training: .\ex_stamp\class7.jpg
Search: .\ex_stamp\
Parameters
Edge Filter: (Image Processing---------------1st box)
None
Threshold Filter: (Image Processing---------------2nd box)
Default
Clean Up Filter: (Image Processing---------------3rd box)
None
NeuralNet Filter Parameters: (Neural Net Parameters)
Translation Symmetry
Blurring = 12
Sensitivity = 40
InternalCut = 70
Operation
-
Click the "Key Segment" button and select ".\ex_stamp\class 7.jpg";
-
Click the "Source" button and select ".\ex_stamp\";
-
Set the Parameters as specified above;
-
Click the "Neural Net\Train" button to train the filter;
-
Click the "NeuralNet\matching\1:N (key vs Source)" button to make a search.
Results
ID Name Path Score X Y W H
CLASS7 CLASS7.JPG C:\...\ex_stamp\ 128000000 36 24 272 256
CLASS7_1 CLASS7_1.JPG C:\...\ex_stamp\ 61184 36 24 272 256
class7_10 class7_10.jpg C:\...\ex_stamp\ 29504 48 36 272 256
CLASS7_2 CLASS7_2.JPG C:\...\ex_stamp\ 26240 56 28 272 256
CLASS7_3 CLASS7_3.JPG C:\...\ex_stamp\ 22336 40 52 272 256
CLASS7_4 CLASS7_4.JPG C:\...\ex_stamp\ 26880 56 28 272 256
CLASS7_5 CLASS7_5.JPG C:\...\ex_stamp\ 21184 52 32 272 256
CLASS7_6 CLASS7_6.JPG C:\...\ex_stamp\ 21568 56 32 272 256
CLASS7_7 CLASS7_7.JPG C:\...\ex_stamp\ 50688 72 4 272 256
CLASS7_8 CLASS7_8.JPG C:\...\ex_stamp\ 33536 28 0 272 256
CLASS7_9 CLASS7_9.JPG C:\...\ex_stamp\ 44096 48 0 272 256
Summary
# Images = 104
# To be retrieved = 11
# Retrieved Correctly = 11
# Missed = 0
Hit Ratio = 100%
Copyright (c) 1998 - 2006 Attrasoft, Inc. All rights reserved.