Introduction - If you have any usage issues, please Google them yourself
Step 1: train the network. Use training samples for training. (this program can also be untrained because the author has saved the trained network parameters and can be identified directly by the reader)
Step 2: identify.
First, open the image (256 colors)
Again, normalization. Click "one-time processing"
Finally, click "R" or use the menu to find the corresponding item to identify
The results of the identification are displayed on the screen and are also exported to the file result.txt
The recognition rate of the system is usually 90%
In addition, the image preprocessing work can be taken step by step. But be aware that each step can only be performed once and in order. The steps are: "256 color bitmap to grayscale graph" "grayscale graph binary" "de-noising" "tilt correction" "segmentation" "standard size" "austerity rearrangement"
Also note that the image to be identified is in the same directory as win.dat and whi. Dat. The two files hold the weight parameters of the network after training
Packet : 67506252digitrec.rar filelist
MainFrm.h
DIBAPI.CPP
Bp.h
DigitRec.dsp
使用说明.txt
DigitRec.cpp
INPUT1.cpp
INPUT1.h
DBpParamater.cpp
DBpParamater.h
DigitRec.dsw
DigitRec.h
ReadMe.txt
StdAfx.cpp
StdAfx.h
DIBAPI.H
Release\DigitRec.exe
Release\图片\6.bmp
Release\图片\7.bmp
Release\图片\num
Release\图片\result.txt
Release\图片\Thumbs.db
Release\图片\whi.dat
Release\图片\win.dat
Release\图片\已经训练好的网络参数.rar
Release\图片\测试图片1.bmp
Release\图片\测试图片2.bmp
Release\图片\测试图片3.bmp
Release\图片\测试图片4.bmp
Release\图片\测试图片5.bmp
Release\图片\测试图片6.bmp
Release\图片\测试图片7.bmp
Release\图片\训练样本.bmp
res\DigitRec.ico
res\DigitRec.rc2
res\Toolbar.bmp
DigitRec.clw
ChildView.cpp
DigitRec.aps
DigitRec.rc
mydiblib.h
Resource.h
ChildView.h
MainFrm.cpp
Release\图片
Release
res