int main()
{
const int sampleTypeCount = 7; //共有几种字体
const int sampleCount = 50; //每种字体的样本数
const int sampleAllCount = sampleCount*sampleTypeCount;
const int featureCount = 256; //特征维数
CvANN_MLP bp;// = CvANN_MLP(layerSizes,CvANN_MLP::SIGMOID_SYM,1,1);
string str_dir[sampleTypeCount];
str_dir[0] = "A水滴渍";
str_dir[1] = "B水纹";
str_dir[2] = "C指纹";
str_dir[3] = "D釉面凹凸";
str_dir[4] = "X凹点";
str_dir[5] = "Y杂质";
str_dir[6] = "Z划痕";
float trainingData[sampleAllCount][featureCount] = { 0 };
float outputData[sampleAllCount][sampleTypeCount] = { 0 };
int itemIndex = 0;
for (int index = 0; index < 7; index++)
{
for (int i = 1; i <= 50; i++)
{
outputData[itemIndex][index] = 1;
cout << str_dir[index] << "_" << i << endl;
stringstream ss;
char num[4];
sprintf(num, "%03d", i);
ss << "特征样本库\\" << str_dir[index] << "\\" << num << ".jpg";
string path;
ss >> path;
//读取灰度图像以便计算灰度直方图
cv::Mat f = cv::imread(path, 0);
cv::Mat grayHist;
// 设定bin数目，也就是灰度级别，这里选择的是0-255灰度
int histSize = 256;
//cv::equalizeHist(f, f);
cv::normalize(f, f, histSize, 0, cv::NORM_MINMAX);
//cv::bitwise_xor(f, cv::Scalar(255), f);//反相
FeatureMaker::GetGrayHist(f, grayHist, histSize);
for (int j = 0; j < 256; j++)
{
trainingData[itemIndex][j] = grayHist.ptr<float>(j)[0];
}
itemIndex++;
}
}
//创建一个网络
cv::Mat layerSizes = (cv::Mat_<int>(1, 3) << featureCount, 25, sampleTypeCount);//创建一个featureCount输入 IDC_EDIT_YinCangCount隐藏 sampleTypeCount输出的三层网络
CvANN_MLP_TrainParams param;
param.term_crit = cvTermCriteria(CV_TERMCRIT_ITER + CV_TERMCRIT_EPS, 50000, 0.002);
param.train_method = CvANN_MLP_TrainParams::BACKPROP;
param.bp_dw_scale = 0.01;//权值更新率
param.bp_moment_scale = 0.03;//权值更新冲量
cv::Mat inputs(sampleAllCount, featureCount, CV_32FC1, trainingData);//样品总数，特征维数，储存的数据类型
cv::Mat outputs(sampleAllCount, sampleTypeCount, CV_32FC1, outputData);
bp.create(layerSizes, CvANN_MLP::SIGMOID_SYM);
bp.train(inputs, outputs, cv::Mat(), cv::Mat(), param);
bp.save("ANN_mlp.xml");
itemIndex = 0;
int zhengque = 0;
for (int index = 0; index < 7; index++)
{
for (int i = 1; i <= 50; i++)
{
cv::Mat sampleMat(1, featureCount, CV_32FC1, trainingData[itemIndex]);//样品总数，特征维数，储存的数据类型
cv::Mat nearest(1, sampleTypeCount, CV_32FC1, cv::Scalar(0));
bp.predict(sampleMat, nearest);
float possibility = -1;
int outindex = 0;
for (int i = 0; i < nearest.size().width; i++){
float x = nearest.at<float>(0, i);
if (x>possibility){
possibility = x;
outindex = i;
}
}
if (outindex == index)
zhengque++;
cout << str_dir[index] << "_" << i << "：" << outindex << "->" << possibility << "->" << str_dir[outindex] << endl;
itemIndex++;
}
}
cout << "正确率" << ((double)zhengque / (double)sampleAllCount);
return 0;
}
QList<int> Skeleton::possibleNumbers(QList<int> digitsOnField){
QList<int> list;
/*if (listHoles.size() > 2 || listJunctions.size() > 5 || listLineEnds.size() > 5){
// certainly a bad 'catch'
}
if (listHoles.size() == 0){
if (listLineEnds.size() >= 2){
if (listLineEnds[0].x < 0.3 && listLineEnds[0].y < 0.3 && listLineEnds[1].x > 0.7 && listLineEnds[1].y > 0.7){
list.push_back(2);
}
else if (listLineEnds[0].x > 0.7 && listLineEnds[0].y < 0.3 && listLineEnds[1].y > 0.5){
list.push_back(5);
}
else if (listLineEnds[1].x < 0.4){
list.push_back(7);
}
else {
list.push_back(1);
}
}
else {
list.push_back(1);
list.push_back(2);
list.push_back(5);
list.push_back(7);
}
}
else if (listHoles.size() == 1){
if (listHoles[0].x > 0.35 && listHoles[0].x < 0.65 && listHoles[0].y > 0.35 && listHoles[0].y < 0.65){
if (listJunctions.size() + listLineEnds.size() < 3){
list.push_back(0);
}
else {
list.push_back(4);
}
}
else if (listHoles.size() == 1 && listJunctions.size() >= 1 && listLineEnds.size() >= 1){
if (listHoles[0].y > listLineEnds[0].y){
list.push_back(6);
}
else {
list.push_back(9);
}
}
else {
list.push_back(0);
list.push_back(4);
list.push_back(6);
list.push_back(9);
}
}
else if (listHoles.size() == 2){
list.push_back(8);
}
else { // if the 'number' is close to a 'normal' number, but not a perfect match, we try all of them
for (int i = 0; i < TEMPLATES_COUNT; i++){
list.push_back(i);
}
}*/
if (HOLE_SEPARATION){
if (listHoles.size() == 0){
QList<int> zeroHole;
zeroHole.push_back(1);
zeroHole.push_back(2);
zeroHole.push_back(3);
zeroHole.push_back(5);
zeroHole.push_back(7);
int dim = Skeleton::getDim(M0);
QList<double> vect = vectorization(M0);
float sampleData[dim];
for (int i = 0; i < dim; i++){
sampleData[i] = vect[i];
}
cv::Mat sampleMat(1, dim, CV_32FC1, sampleData);
int maxVote = 0;
int intMaxVote = -1;
for (int i = 0; i < 10; i++){
int vote = 0;
for (int j = 0; j < 10; j++){
if (i != j && zeroHole.contains(i) && zeroHole.contains(j)){
float response = Skeleton::machines.m[min(i,j)][max(i,j)]->predict(sampleMat);
if ((response == 0.0 && i == min(i,j)) || (response == 1.0 && i == max(i,j))){
vote++;
}
}
}
//qDebug() << i << ": " << vote;
if (vote > maxVote){
maxVote = vote;
intMaxVote = i;
}
}
if (intMaxVote == -1){
qDebug() << "Error : No vote is greater than 0";
}
else {
list.push_back(intMaxVote);
}
}
else if (listHoles.size() == 1){
QList<int> oneHole;
oneHole.push_back(0);
oneHole.push_back(4);
oneHole.push_back(6);
oneHole.push_back(9);
int dim = Skeleton::getDim(M1);
QList<double> vect = vectorization(M1);
float sampleData[dim];
for (int i = 0; i < dim; i++){
sampleData[i] = vect[i];
}
cv::Mat sampleMat(1, dim, CV_32FC1, sampleData);
int maxVote = 0;
int intMaxVote = -1;
for (int i = 0; i < 10; i++){
int vote = 0;
for (int j = 0; j < 10; j++){
if (i != j && oneHole.contains(i) && oneHole.contains(j)){
float response = Skeleton::machines.m[min(i,j)][max(i,j)]->predict(sampleMat);
if ((response == 0.0 && i == min(i,j)) || (response == 1.0 && i == max(i,j))){
vote++;
}
}
}
//qDebug() << i << ": " << vote;
if (vote > maxVote){
maxVote = vote;
intMaxVote = i;
}
}
if (intMaxVote == -1){
qDebug() << "Error : No vote is greater than 0";
}
else {
list.push_back(intMaxVote);
}
}
else if (listHoles.size() == 2){
list.push_back(8);
}
else { // if the 'number' is close to a 'normal' number, but not a perfect match, we try all of them
for (int i = 0; i < TEMPLATES_COUNT; i++){
list.push_back(i);
}
}
}
else {
int dim = Skeleton::getDim(M0);
QList<double> vect = vectorization(M0);
float sampleData[dim];
for (int i = 0; i < dim; i++){
sampleData[i] = vect[i];
}
cv::Mat sampleMat(1, dim, CV_32FC1, sampleData);
int maxVote = 0;
int intMaxVote = -1;
for (int i = 0; i < 10; i++){
int vote = 0;
for (int j = 0; j < 10; j++){
if (i != j && digitsOnField.contains(i) && digitsOnField.contains(j)) {
float response = Skeleton::machines.m[min(i,j)][max(i,j)]->predict(sampleMat);
if ((response == 0.0 && i == min(i,j)) || (response == 1.0 && i == max(i,j))){
vote++;
}
}
}
//qDebug() << i << ": " << vote;
if (vote > maxVote){
maxVote = vote;
intMaxVote = i;
}
}
if (intMaxVote == -1){
qDebug() << "Error : No vote is greater than 0";
}
else {
list.push_back(intMaxVote);
}
}
return list;
}
int main()
{
const int sampleTypeCount = 7; //共有几种字体
const int sampleCount = 50; //每种字体的样本数
const int sampleAllCount = sampleCount*sampleTypeCount;
const int featureCount = 256; //特征维数
CvANN_MLP bp;// = CvANN_MLP(layerSizes,CvANN_MLP::SIGMOID_SYM,1,1);
string str_dir[sampleTypeCount];
str_dir[0] = "A水滴渍";
str_dir[1] = "B水纹";
str_dir[2] = "C指纹";
str_dir[3] = "D釉面凹凸";
str_dir[4] = "X凹点";
str_dir[5] = "Y杂质";
str_dir[6] = "Z划痕";
float trainingData[sampleAllCount][featureCount] = { 0 };
float outputData[sampleAllCount][sampleTypeCount] = { 0 };
int itemIndex = 0;
for (int index = 0; index < 7; index++)
{
for (int i = 1; i <= 50; i++)
{
outputData[itemIndex][index] = 1;
cout << str_dir[index] << "_" << i << endl;
stringstream ss;
char num[4];
sprintf(num, "%03d", i);
ss << "特征样本库\\" << str_dir[index] << "\\" << num << ".jpg";
string path;
ss >> path;
//读取灰度图像以便计算灰度直方图
cv::Mat f = cv::imread(path, 0);
cv::Mat grayHist;
// 设定bin数目，也就是灰度级别，这里选择的是0-255灰度
int histSize = 63;
//cv::equalizeHist(f, f);
cv::normalize(f, f, histSize, 0, cv::NORM_MINMAX);
//cv::bitwise_xor(f, cv::Scalar(255), f);//反相
// 设定取值范围，设定每级灰度的范围。
float range[] = { 0, 255 };
const float* histRange = { range };
bool uniform = true; bool accumulate = false;
cv::calcHist(&f, 1, 0, cv::Mat(), grayHist, 1, &histSize, &histRange, uniform, accumulate);
for (int j = 0; j < 256; j++)
{
trainingData[itemIndex][j] = grayHist.ptr<float>(0)[0];
}
itemIndex++;
/*
// 创建直方图画布
int hist_w = 400; int hist_h = 400;
int bin_w = cvRound((double)hist_w / histSize);
cv::Mat histImage(hist_w, hist_h, CV_8UC3, cv::Scalar(0, 0, 0));
/// 将直方图归一化到范围 [ 0, histImage.rows ]
cv::normalize(grayHist, grayHist, 0, histImage.rows, cv::NORM_MINMAX, -1, cv::Mat());
/// 在直方图画布上画出直方图
for (int i = 1; i < histSize; i++)
{
line(histImage, cv::Point(bin_w*(i - 1), hist_h - cvRound(grayHist.at<float>(i - 1))),
cv::Point(bin_w*(i), hist_h - cvRound(grayHist.at<float>(i))),
cv::Scalar(0, 0, 255), 2, 8, 0);
}
stringstream s;
s << "samples\\反相正规化直方图\\" << str_dir[index] << "\\";
//s << "samples\\正规化直方图\\" << str_dir[index] << "\\";
//s << "samples\\均衡化直方图\\" << str_dir[index] << "\\";
//s << "samples\\直方图\\" << str_dir[index] << "\\";
//string dir = s.str();
//char* c;
//int len = dir.length();
//c = new char[len + 1];
//strcpy(c, dir.c_str());
//CheckDir(c);
s << "" << num << ".jpg";
s >> path;
cv::imwrite(path, histImage);
s.clear();
s << "samples\\反相正规化直方图\\" << str_dir[index] << "\\" << "Hist_" << num << ".jpg";
//s << "samples\\正规化直方图\\" << str_dir[index] << "\\" << "Hist_" << num << ".jpg";
//s << "samples\\均衡化直方图\\" << str_dir[index] << "\\" << "Hist_" << num << ".jpg";
//s << "samples\\直方图\\" << str_dir[index] << "\\" << "Hist_" << num << ".jpg";
s >> path;
cv::imwrite(path, grayHist);
/// 显示直方图
//cv::namedWindow("calcHist Demo", CV_WINDOW_AUTOSIZE);
//cv::imshow("calcHist Demo", histImage);
//cv::waitKey(0);
*/
}
}
//创建一个网络
cv::Mat layerSizes = (cv::Mat_<int>(1, 3) << featureCount, 25, sampleTypeCount);//创建一个featureCount输入 IDC_EDIT_YinCangCount隐藏 sampleTypeCount输出的三层网络
CvANN_MLP_TrainParams param;
param.term_crit = cvTermCriteria(CV_TERMCRIT_ITER + CV_TERMCRIT_EPS, 5000, 0.01);
param.train_method = CvANN_MLP_TrainParams::BACKPROP;
param.bp_dw_scale = 0.2;
param.bp_moment_scale = 0.1;
cv::Mat inputs(sampleAllCount, featureCount, CV_32FC1, trainingData);//样品总数，特征维数，储存的数据类型
cv::Mat outputs(sampleAllCount, sampleTypeCount, CV_32FC1, outputData);
bp.create(layerSizes, CvANN_MLP::SIGMOID_SYM);
bp.train(inputs, outputs, cv::Mat(), cv::Mat(), param);
bp.save("ANN_mlp.xml");
itemIndex = 0;
for (int index = 0; index < 7; index++)
{
for (int i = 1; i <= 50; i++)
{
cv::Mat sampleMat(1, featureCount, CV_32FC1, trainingData[itemIndex]);//样品总数，特征维数，储存的数据类型
cv::Mat nearest(1, sampleTypeCount, CV_32FC1, cv::Scalar(0));
bp.predict(sampleMat, nearest);
float possibility = -1;
int outindex = 0;
for (int i = 0; i < nearest.size().width; i++){
float x = nearest.at<float>(0, i);
if (x>possibility){
possibility = x;
outindex = i;
}
}
cout << str_dir[index] << "_" << i << "：" << outindex << "->" << possibility << "->" << str_dir[outindex] << endl;
itemIndex++;
}
}
return 0;
}
