通用图像识别的神经网络代码描述

　2006-02-04 14:08:53　来源：WEB开发网　　　

核心提示： 写人脸检测程序的时候顺带写的，网络格式是靠读入一个文件定义的，通用图像识别的神经网络代码描述，文件的格式如下：输入图像长输入图像宽隐层神经元个数输出神经元个数不同网络结构数量[连接位置不同的隐层神经元的个数连接的隐层神经元个数][隐层神经元连接的输入神经元的位置表]下面是一个例子：24 28 52 1316

写人脸检测程序的时候顺带写的，网络格式是靠读入一个文件定义的，文件的格式如下：

输入图像长输入图像宽隐层神经元个数输出神经元个数
不同网络结构数量
[连接位置不同的隐层神经元的个数连接的隐层神经元个数]
[隐层神经元连接的输入神经元的位置表]

下面是一个例子：

24 28 52 1
3
16 32
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 4 4 4 4 4 4
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
5 5 5 5 5 5 6 6 6 6 6 6 7 7 7 7 7 7 8 8 8 8 8 8
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
9 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 11 11 12 12 12 12 12 12
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
13 13 13 13 13 13 14 14 14 14 14 14 15 15 15 15 15 15 16 16 16 16 16 16
4 8
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4
6 12
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

下面是程序代码：

type

TSingleExtendedArray = array of extended;
TDoubleExtendedArray = array of array of extended;

TSamples = packed record
　 Ins: TSingleExtendedArray;
　 Outs: TSingleExtendedArray;
end;

type

TGraphicBpnn = class
PRivate
　 procedure BackPropagate(t: TSingleExtendedArray; n, m: extended);
　 function UpDate(inputs: TSingleExtendedArray): extended;
public
　 samplecounts, TestCounts: longint;
　 procedure AddToTrain(Ins, Outs: TSingleExtendedArray);
　 procedure AddToTest(Ins, Outs: TSingleExtendedArray);
　 procedure SaveToFile(FileName: string);
　 procedure LoadFromFile(FileName: string);
　 procedure Train(n, m: extended);
　 function Init(FileName: string): boolean;
　 function Predict(Ins: TSingleExtendedArray): extended;
　 function Test: extended;
　 destructor Destroy; override;
private
　 nI, nH, nO: longint;
　 aI, aH, aO, Output_Deltas, Hidden_Deltas: TSingleExtendedArray;
　 wI, wO, cI, cO: TDoubleExtendedArray;
　 Connections: array of array of boolean;
　 Samples: array of TSamples;
　 TestSet: array of TSamples;
end;

implementation

function TGraphicBpnn.Init(FileName: string): boolean;
var
i, j, k, fi, fj: longint;
nIw, nIh, RopMax, RopNum, RopTypes: longint;
RopMap: array of longint;
begin
AssignFile(Input, FileName);
ReSet(Input);
Readln(Input, nIw, nIh, nH, nO);
nI := nIw * nIh;
setlength(aI, nI);
setlength(aH, nH);
setlength(aO, nO);
for i := 0 to nI - 1 do aI[i] := 1;
for i := 0 to nH - 1 do aH[i] := 1;
for i := 0 to nO - 1 do aO[i] := 1;

setlength(wI, nI, nH);
setlength(wO, nH, nO);
setlength(cI, nI, nH);
setlength(cO, nH, nO);
setlength(Connections, nI, nH);

for i := 0 to nI - 1 do
　 for j := 0 to nH - 1 do
　　 Connections[i, j] := False;

Readln(RopTypes); fj := 0;
for k := 1 to RopTypes do begin
　 Readln(RopMax, RopNum);
　 setlength(RopMap, nI);
　 fi := 0;
　 for i := 1 to nIh do begin
　　 for j := 1 to nIw do begin
　　　 Read(RopMap[fi]);
　　　 Inc(fi);
　　 end;
　　 Readln;
　 end;
　 fi := 0;
　 for i := 1 to RopNum do begin
　　 Inc(fi);
　　 if fi > RopMax then fi := 1;
　　 for j := 0 to nI - 1 do
　　　 if RopMap[j] = fi then Connections[j, fj] := true;
　　 Inc(fj);
　 end;
end;

setlength(Output_Deltas, nO);
setlength(Hidden_Deltas, nH);

randomize;
for i := 0 to nI - 1 do
　 for j := 0 to nH - 1 do begin
　　 cI[i, j] := 0;
　　 wI[i, j] := random(40000) / 10000 - 2;
　 end;

for i := 0 to nH - 1 do
　 for j := 0 to nO - 1 do begin
　　 cO[i, j] := 0;
　　 wO[i, j] := random(40000) / 10000 - 2;
　 end;

setlength(Samples, $100); setlength(TestSet, $100);
samplecounts := 0; TestCounts := 0;
CloseFile(Input);
end;

procedure TGraphicBpnn.BackPropagate(t: TSingleExtendedArray; n, m: extended);
var
i, j, k: Longint;
Sum, Change: extended;
begin
for i := 0 to nO - 1 do
　 Output_Deltas[i] := aO[i] * (1 - aO[i]) * (t[i] - aO[i]);

for j := 0 to nH - 1 do begin
　 Sum := 0;
　 for k := 0 to nO - 1 do
　　 Sum := Sum + Output_Deltas[k] * wO[j, k];
　 Hidden_Deltas[j] := aH[j] * (1 - aH[j]) * Sum;
end;

for j := 0 to nH - 1 do
　 for k := 0 to nO - 1 do begin
　　 Change := Output_Deltas[k] * aH[j];
　　 wO[j, k] := wO[j, k] + n * Change + m * cO[j, k];
　　 cO[j, k] := Change;
　 end;

for i := 0 to nI - 1 do
　 for j := 0 to nH - 1 do
　　 if Connections[i, j] then begin
　　　 Change := Hidden_Deltas[j] * aI[i];
　　　 wI[i, j] := wI[i, j] + n * Change + m * cI[i, j];
　　　 cI[i, j] := Change;
　　 end;

end;

function TGraphicBpnn.UpDate(inputs: TSingleExtendedArray): extended;
var
i, j, k: Longint;
Sum: extended;
begin
for i := 0 to nI - 1 do
　 aI[i] := Inputs[i];
for j := 0 to nH - 1 do begin
　 Sum := 0;
　 for i := 0 to nI - 1 do
　　 if Connections[i, j] then
　　　 Sum := Sum + aI[i] * wI[i, j];
　 aH[j] := 1 / (1 + Exp(-Sum));
end;
for k := 0 to nO - 1 do begin
　 Sum := 0;
　 for j := 0 to nH - 1 do
　　 Sum := Sum + aH[j] * wO[j, k];
　 aO[k] := 1 / (1 + Exp(-Sum));
end;
UpDate := aO[0];
end;

procedure TGraphicBpnn.Train(n, m: extended);
var i: Longint;
begin
for i := 0 to samplecounts - 1 do begin
　 UpDate(Samples[i].Ins);
　 BackPropagate(Samples[i].Outs, n, m);
end;
end;

procedure TGraphicBpnn.AddToTrain(Ins, Outs: TSingleExtendedArray);
var i: longint;
begin
if samplecounts > High(Samples) then setlength(Samples, samplecounts + $100);
setlength(Samples[samplecounts].Ins, nI);
setlength(Samples[samplecounts].Outs, nO);
for i := 0 to nI - 1 do Samples[samplecounts].Ins[i] := Ins[i];
for i := 0 to nO - 1 do Samples[samplecounts].Outs[i] := Outs[i];
Inc(samplecounts);
end;

procedure TGraphicBpnn.AddToTest(Ins, Outs: TSingleExtendedArray);
var i: longint;
begin
if TestCounts > High(TestSet) then setlength(TestSet, TestCounts + $100);
setlength(TestSet[TestCounts].Ins, nI);
setlength(TestSet[TestCounts].Outs, nO);
for i := 0 to nI - 1 do TestSet[TestCounts].Ins[i] := Ins[i];
for i := 0 to nO - 1 do TestSet[TestCounts].Outs[i] := Outs[i];
Inc(TestCounts);
end;

procedure TGraphicBpnn.SaveToFile(FileName: string);
var
i, j, k: longint;
SaveStream: TMemoryStream;
begin
SaveStream := TMemoryStream.Create;
SaveStream.Seek(0, 0);
for i := 0 to nI - 1 do
　 for j := 0 to nH - 1 do begin
　　 SaveStream.Write(wI[i, j], sizeof(wI[i, j]));
　　 SaveStream.Write(cI[i, j], sizeof(cI[i, j]));
　 end;
for j := 0 to nH - 1 do
　 for k := 0 to nO - 1 do begin
　　 SaveStream.Write(wO[j, k], sizeof(wO[j, k]));
　　 SaveStream.Write(cO[j, k], sizeof(cO[j, k]));
　 end;
SaveStream.SaveToFile(FileName);
SaveStream.Free;
end;

procedure TGraphicBpnn.LoadFromFile(FileName: string);
var
i, j, k: longint;
ReadStream: TMemoryStream;
begin
ReadStream := TMemoryStream.Create;
ReadStream.LoadFromFile(FileName);
ReadStream.Seek(0, 0);
for i := 0 to nI - 1 do
　 for j := 0 to nH - 1 do begin
　　 ReadStream.Read(wI[i, j], sizeof(wI[i, j]));
　　 ReadStream.Read(cI[i, j], sizeof(cI[i, j]));
　 end;
for j := 0 to nH - 1 do
　 for k := 0 to nO - 1 do begin
　　 ReadStream.Read(wO[j, k], sizeof(wO[j, k]));
　　 ReadStream.Read(cO[j, k], sizeof(cO[j, k]));
　 end;
ReadStream.Free;
end;

function TGraphicBpnn.Predict(Ins: TSingleExtendedArray): extended;
begin
try
　 Predict := Update(Ins);
except
　 Predict := 0;
end;
end;

function TGraphicBpnn.Test: extended;
var
PreRet: extended;
i, Counts, Ret: longint;
begin
Counts := 0;
for i := 0 to TestCounts - 1 do begin
　 PreRet := Predict(TestSet[i].Ins);
　 if PreRet > 0.5 then Ret := 1 else Ret := 0;
　 if Ret = TestSet[i].Outs[0] then Inc(Counts);
end;
Result := Counts / TestCounts;
end;

destructor TGraphicBpnn.Destroy;
begin
setlength(aI, 0);
setlength(aH, 0);
setlength(aO, 0);
setlength(Output_Deltas, 0);
setlength(Hidden_Deltas, 0);
setlength(wI, 0, 0);
setlength(wO, 0, 0);
setlength(cI, 0, 0);
setlength(cO, 0, 0);
setlength(Connections, 0, 0);
setlength(Samples, 0);
inherited;
end;