C++实现汉字识别

#include <stdlib.h> #include <stdio.h> #include <math.h> #include <cv.h> #include <highgui.h> //宏定义字数，最多150 #define WORDSNUM 105 //宏定义行数，最多7 #define ROWSNUM 3 //int END=0; int rows=0; //为了方便，存字的文件名列表定义为全局变量 const char *rowfilenamelist[7]= {"row0.jpg","row1.jpg","row2.jpg","row3.jpg","row4.jpg","row5.jpg","row6.jpg"}; const char *wordfilenamelist[180]= {"col0.jpg","col1.jpg","col2.jpg","col3.jpg","col4.jpg","col5.jpg","col6.jpg", "col7.jpg","col8.jpg","col9.jpg","col10.jpg","col11.jpg","col12.jpg","col13.jpg", "col14.jpg","col15.jpg","col16.jpg","col17.jpg","col18.jpg","col19.jpg","col20.jpg", "col21.jpg","col22.jpg","col23.jpg","col24.jpg","col25.jpg","col26.jpg","col27.jpg", "col28.jpg","col29.jpg","col30.jpg","col31.jpg","col32.jpg","col33.jpg","col34.jpg", "col35.jpg","col36.jpg","col37.jpg","col38.jpg","col39.jpg","col40.jpg","col41.jpg", "col42.jpg","col43.jpg","col44.jpg","col45.jpg","col46.jpg","col47.jpg","col48.jpg", "col49.jpg","col50.jpg","col51.jpg","col52.jpg","col53.jpg","col54.jpg","col55.jpg", "col56.jpg","col57.jpg","col58.jpg","col59.jpg","col60.jpg","col61.jpg","col62.jpg", "col63.jpg","col64.jpg","col65.jpg","col66.jpg","col67.jpg","col68.jpg","col69.jpg", "col70.jpg","col71.jpg","col72.jpg","col73.jpg","col74.jpg","col75.jpg","col76.jpg", "col77.jpg","col78.jpg","col79.jpg","col80.jpg","col81.jpg","col82.jpg","col83.jpg", "col84.jpg","col85.jpg","col86.jpg","col87.jpg","col88.jpg","col89.jpg","col90.jpg", "col91.jpg","col92.jpg","col93.jpg","col94.jpg","col95.jpg","col96.jpg","col97.jpg", "col98.jpg","col99.jpg","col100.jpg","col101.jpg","col102.jpg","col103.jpg","col104.jpg", "col105.jpg","col106.jpg","col107.jpg","col108.jpg","col109.jpg","col110.jpg","col111.jpg", "col112.jpg","col113.jpg","col114.jpg","col115.jpg","col116.jpg","col117.jpg","col118.jpg", "col119.jpg","col120.jpg","col121.jpg","col122.jpg","col123.jpg","col124.jpg","col125.jpg", "col126.jpg","col127.jpg","col128.jpg","col129.jpg","col130.jpg","col131.jpg","col132.jpg", "col133.jpg","col134.jpg","col135.jpg","col136.jpg","col137.jpg","col138.jpg","col139.jpg", "col140.jpg","col141.jpg","col142.jpg","col143.jpg","col144.jpg","col145.jpg","col146.jpg", "col147.jpg","col148.jpg","col149.jpg"};//文件名列表 //将图像img中的矩形Rect部分存到文件filename中去 void savetofile(const char* filename,CvRect Rect,IplImage* img); //找一行文字切分上顶点像素行 int findN1(IplImage *image); //找一行文字切分时行底像素行 int findN2(IplImage *image,int I); //找一列文字切分时左边像素列 int findColleft(IplImage *image); //找一列文字切分时右边像素列 int findColright(IplImage *image,int I); //生成矩形 CvRect makeRect(CvRect rect,int x,int y,int w,int h); //行切割时，将图像img中i,j像素行之间的文字存到文件filename中，其余的部分存放到tempfile文件中 void cutRowandSave(int i,int j,IplImage* img,const char* filename,const char* tempfile); ////////找给定图片的字的平均宽宽 int findEvenwidth(IplImage* img); ////清理白条 void Clear(const char* wordfilenamelist,double min); ///得到字图片最大高度和宽度 int getMaxwidth(); int getMaxheight(); ///重设字图片大小为统一 void resize(int maxWidth,int maxHeight); ///行切分 void rowDivide(IplImage *img); //列切分 void colDivide(); int findN1(IplImage *image)//找第一行文字上边沿行 { int i,j; int height = image->height; int width = image->width; int step = image->widthStep; int channels = image->nChannels; unsigned char* data = (uchar *)image->imageData;//取图像数据，数据经过二值化后是0-255之间 // printf("Processing a %dx%d image with %d channels\n",height,width,channels); //输出图像信息 int count=0;//用来记录一行中黑点数 for(i=0;i<height;i++) { for(j=0;j<width;j++) { if(data[i*step+j*channels+1]==0)//如果是黑点 { count=count+1;//黑点计数加1 } } if(count>=5)//如果一行中超过五个黑点，则说明是一行文字的顶部 { // printf("%d,%d\n",i,j); return i; } count=0;//换一行扫瞄时计数器清零 } return 0;//否则返回0，说明图中没有文字 } int findN2(IplImage *image,int I)//找一行文字切分的底部像素行 { int i,j; int height = image->height; int width = image->width; int step = image->widthStep; int channels = image->nChannels; unsigned char* data = (uchar *)image->imageData; int count=0; for(i=I;i<height;i++) { for(j=0;j<width;j++) { if(data[i*step+j*channels+1]==0)//如果是黑点 { count=count+1;//黑点计数加1 } } if(count<5)//如果某一行黑点小于5个则说明这一行文件到底了，可以根据需要改动这个参数。 { return i; } count=0;//否则下一行开始扫瞄时，计数器需要清零 } return 0; } int findColleft(IplImage *image)//找一列字左部开始处 { int i,j; int height = image->height; int width = image->width; int step = image->widthStep; int channels = image->nChannels; unsigned char* data = (uchar *)image->imageData; // printf("Processing a %dx%d image with %d channels\n",height,width,channels); //输出图像信息 int count=0; int pre=0;//用来记录是否连续几列像素都是黑点较多，用以确定是否真的开始了一列 for(i=0;i<width;i++) { for(j=0;j<height;j++) { if(data[j*step+i*channels+1]<=10)//由于有躁声，小于11的值都认为是黑点 { count=count+1;//如果是黑点则计数器加1 } } if(count>2)//如果一列多于两个黑点 { pre=pre+1;//记录下来 } if(pre>2) { // printf("\n找到了right:%d\n",i);//如果连续三列多于两个黑点则返回 return i-2; } count=0;//换列时候计数器清零 } return 0; } ////////////////// int findColright(IplImage *image,int I)//找一列字的结束处 { int i=I; int j; int height = image->height; int width = image->width; int step = image->widthStep; int channels = image->nChannels; unsigned char* data = (uchar *)image->imageData; int count=0; int pre=0;//用来记录是否连续几列都是空白，可有效区分部首偏旁 for(i=I+1;i<width;i++) { for(j=0;j<height;j++) { if(data[j*step+i*channels+1]<=10) { count=count+1; // printf("Count=:%d ",count); } // printf("%d,",data[j*step+i*channels+1]); } if(count<1)//某一列黑点个数极少 { pre++; } if(pre>1)//连续多列黑点个数极少则返回 { return i-1; } count=0; } // printf("已经没有汉字了\n"); // END=1; return -1; } ///为矩形框赋值，此函数用处不大 CvRect makeRect(CvRect rect,int x,int y,int w,int h) { rect.x=x; rect.y=y; rect.width=w; rect.height=h; return rect; } /////将i,j像素行之间的图像剪下并存入filename中，剩下部分存入tempfile中 //可以再扩展一个将i，j像素列之间图像剪下并存入filename中的函数 void cutRowandSave(int i,int j,IplImage* img,const char* filename,const char* tempfile) { IplImage* localImg=img; IplImage* pImgRect=0; IplImage* temp=0; CvRect Rect; Rect.x=0;Rect.y=0;Rect.width=0;Rect.height=0;//初始化 Rect=makeRect(Rect,0,i,localImg->width,j-i);//为矩形框赋值 savetofile(filename,Rect,img);//存入文件中 Rect.x=0; Rect.y=j; Rect.height=img->height-j; savetofile(tempfile,Rect,img);//存剩余部分入文件中 } //将图像img中的矩形Rect部分存入文件filename中 void savetofile(const char* filename,CvRect Rect,IplImage* img) { IplImage* localImg=img; IplImage* pImgRect=0; pImgRect = cvCreateImage(cvSize(Rect.width, Rect.height), IPL_DEPTH_8U,1); cvSetImageROI(localImg,Rect); cvCopy(localImg,pImgRect); cvSaveImage(filename,pImgRect); }// ////用一行字的图片img来估计字的平均宽度 int findEvenwidth(IplImage* img) { IplImage* localImg=img; printf("开始估计字的平均宽度\n"); int I=findColleft(img); int J=findColright(img,I); CvRect Rect; Rect.x=0;Rect.y=0;Rect.width=0;Rect.height=0;//初始化 Rect=makeRect(Rect,J,0,img->width-J,img->height); savetofile("temp.jpg",Rect,img); for(int k=1;k<50;k++) { img=cvLoadImage("temp.jpg",0); I=findColleft(img); J=findColright(img,I); if(I==0&&J==1) { printf("得到平均字宽的估计值为:%d\n",localImg->width/k); return localImg->width/k; } Rect=makeRect(Rect,J,0,img->width-J,im