dearsky/HandWritten-Analisys
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

笔迹鉴别程序

Browse Source 
考试的笔迹鉴别程序，分辨出不同人写的笔迹
This commit is contained in:
yanshui177
2017-05-17 16:50:37 +08:00
parent abe00d2e02
commit 962de04ffb
205 changed files with 17672 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

225
测试/本机测试/1-我-算法/HWCV-exe/segmentation.cpp Normal file
View File

@@ -0,0 +1,225 @@
/*
主函数文件：segmentation.cpp 主函数的实现文件
*/
#include "segmentation.h"
//#pragma comment( linker, "/subsystem:windows /entry:mainCRTStartup" )//无界面运行
/*主函数*/
int ComputeImage(vector<string> files, double bzckesa[50][50], double *wcd, int conti)
{
int i, ii, jj, k, size;
double bzcu[50][50] = { 0 }; //标准差中的u
double featurep[50][50][30] = { 0 }; //所有图像的轮廓方向特征初始化//干什么 //30
int feature[50][50][30] = { 0 }; //所有图像的特征值初始化 //所有图像指的什么意思 //30找出30的位置或者50的位置限制。。。。带入num_dir==49的情况进行类比
int featx[50][50] = { 0 }; //循环赋值的feature
int featureall; //图像特征值和 //做什么用
size = files.size();
for (i = 0; i < size; i++)
{
memset(featx, 0, sizeof(featx));
// strcpy(str,files[i].c_str());
singlefeature((char*)files[i].c_str(), featx); //featx[][50]
featureall = 0; //图像特征值和的初始化
for (ii = 0; ii < 48; ii++) //将featx存起来,回头看能不能用函数换掉
for (jj = ii + 1; jj < 47; jj++)
{
feature[ii][jj][i] = featx[ii][jj];
featureall = featureall + featx[ii][jj];
}
//求轮廓方向特征featurep，式(5) 与标准差中的u的和
for (ii = 0; ii < 48; ii++)
for (jj = ii + 1; jj < 47; jj++)
{
featurep[ii][jj][i] = (double)featx[ii][jj] / featureall;
bzcu[ii][jj] += (double)featx[ii][jj] / featureall * 1000; //标准差的值过小,进行放大（1）
}
}
//处理完一个人的每一张图片后
for (ii = 0; ii < 48; ii++)//求标准差中的u
for (jj = ii + 1; jj < 47; jj++)
bzcu[ii][jj] = bzcu[ii][jj] / size;
//求相似性就是带权卡方wcd
for (i = 0; i < size; i++)
for (ii = 0; ii < 48; ii++)
for (jj = ii + 1; jj < 47; jj++)
if (featurep[ii][jj][i] * featurep[ii][jj][conti] != 0 && bzckesa[ii][jj] != -1)
wcd[i] += pow((featurep[ii][jj][i] - featurep[ii][jj][conti]), 2) / ((featurep[ii][jj][i] + featurep[ii][jj][conti])*bzckesa[ii][jj]);
memset(feature, 0, sizeof(feature));
memset(featurep, 0, sizeof(featurep));
return 1;
}
int main(int argc, char* argv[])
{
/*变量定义*/
//string dir;
//if (argc < 2)
// return -1;
//else
// dir = argv[1];
// 调试
// string dir = "E:/xiangmu/Img/imgjiaobiao/010211100518"; //存储目录
// 定义变量
vector<string> dir; //存储目录
int conti = 1; //对比图像的标号
int size_dir, num_dir;
char record[2400] = { 0 };
// 获取待检测文件夹到size
char path[100] = "E:/xiangmu/Img/imgjiaobiao/";//D:/xiangmu/Img/imgjiaobiao/
searchDir(path, dir);//获取filePath下的所有一级目录并存储到dir中
// dir.push_back("E:/xiangmu/Img/imgjiaobiao/010211100518"); //存储目录
size_dir = dir.size(); //dir的大小就是学生的数量
stuAll = size_dir;
cout << "学生总数为" << stuAll << endl;
// 开始检测每个文件夹下的
for (num_dir = 0; num_dir < size_dir; num_dir++)//对每一个学生目录进行循环
{
cout <<(char*)dir[num_dir].c_str();
char record[2400] = { 0 };
FILE* fpzz = NULL;//需要注意
int i, ii, jj, feature[50][50][30] = { 0 }, featureall = 0;
double featurep[50][50][30] = { 0 };
double bzcu[50][50] = { 0 };
double bzckesa[50][50] = { 0 };
double wcd[30] = { 0 };
int featx[50][50] = { 0 };
int featdif[30] = { 0 };
float maxx = 0; //最大特征值的标号与值
int xyimgnum = 0; //嫌疑图片的数目
char str[80]; //存储地址
vector<string> suspict; //记录嫌疑图片地址
vector<float> suspict_wcd; //嫌疑图片的wcd值
vector<string> files; //存储该生所有考试文件路径
/*读取配置文件，并配置各项参数*/
if (!ReadConfig("D:/HWCV/config/configure.cfg"))
{
// SaveLog("\t配置文件读取失败\n", g_log_adr, "a");
SaveLog("\n0\n", g_log_adr, "a");
return 0;
}
char path[100] = "E:/xiangmu/Img/imgjiaobiao/";
/*步骤：读取标准差文件*/
int bzccolumns = 47;//txt文件中有47列
vector<double *> output_bzc;
if (!ReadScanf("D:\\HWCV\\config\\stdfile.db", bzccolumns, output_bzc))
{
memset(g_log_rec, 0, sizeof(g_log_rec));
// SaveLog("\n读取路径：D:\\HWCV\\config\\stdfile.db 的标准差文件失败！\n", g_log_adr, "a");
SaveLog("\n0\n", g_log_adr, "a");
return 0;
}
for (ii = 0; ii < 48; ii++)//output_vector可视为二维数组;输出数组元素：
for (jj = ii + 1; jj < 47; jj++)
bzckesa[ii][jj] = output_bzc[ii][jj];
//开始检测
//-------------------------------------------------------------//
getFiles(dir[num_dir].c_str(), files); //遍历当前文件夹下的所有文件
int size = files.size();
cout << " 文件数：" << size;
//开始对每一张图片进行处理
for (int r = 0; r < size; r++)
{
memset(wcd, 0, sizeof(wcd));
ComputeImage(files, bzckesa, wcd, r);
xyimgnum = 0;
//求卡方距离的最大值
for (i = 0; i < size; i++)
{
// cout << files[i].c_str() << " " << wcd[i] << endl;
if (wcd[i]>0.12)
{
xyimgnum++;
suspict.push_back(files[i].c_str());
suspict_wcd.push_back(wcd[i]);
}
}
if (xyimgnum < 3) break;
}
/*将结果存入log文件*/
memset(g_log_rec, 0, sizeof(g_log_rec));
// strcat(g_log_rec, GetTime());
// strcat(g_log_rec, "\t考生考号：");
strcat(g_log_rec, dir[num_dir].substr(27, 22).c_str());//学号
strcat(g_log_rec, ",");//图片总数为：
char pic_num[20];
_itoa(size, pic_num, 10);
strcat(g_log_rec, pic_num);
if (xyimgnum > 0)
{
stuSus++;
char b[20];
sprintf(b, ",%d", xyimgnum);
strcat(g_log_rec, b);
strcat(g_log_rec, "\n");
for (i = 0; i < xyimgnum; i++)
{
// cout << "嫌疑图像：" << files[i].c_str() << endl;
strcat(g_log_rec, ",,,");//\t嫌疑图像：
strcat(g_log_rec, suspict[i].c_str());
strcat(g_log_rec, ",");//相似度：
float sim = (1.0 - suspict_wcd[i]) * 100;
char a[20];
sprintf(a, "%g", sim);
strcat(g_log_rec, a);
strcat(g_log_rec, "%%\n");
}
}
else
{
// strcat(g_log_rec, "\t该考生没有嫌疑图像！\n");
strcat(g_log_rec, ",0,");
// cout << "该考生无嫌疑图像！" << endl;
}
SaveLog(g_log_rec, g_log_adr, "a");
printf("嫌疑数量：%d, ", xyimgnum);
picAll += size;
picSus += xyimgnum;
printf("全部:%d嫌疑：%d比例为：%g\n", picAll, picSus, ((float)picSus) / ((float)picAll));
xyimgnum = 0;
/*善后*/
suspict.clear();
suspict_wcd.clear();
memset(g_log_rec, 0, sizeof(g_log_rec));
memset(feature, 0, sizeof(feature));
memset(featurep, 0, sizeof(featurep));
memset(bzckesa, 0, sizeof(bzckesa));
memset(wcd, 0, sizeof(wcd));
memset(featdif, 0, sizeof(featdif));
files.clear();
memset(g_log_rec, 0, sizeof(g_log_rec));
/*返回值*/
}
dir.clear();
cout << "学生总数:" << stuAll << " 作弊人数：" << stuSus << endl;
printf("已经打印到txt中！");
string open = "start " + (string)g_log_adr;
system(open.c_str());
system("pause");
return 0; //(1-wcd[maxi])*100
}