opencv 基于knn的手写数字字符识别

#include "basicOCR.h" void basicOCR::preprocessing(Mat &srcimage, int new_width, int new_height) { Mat src = srcimage.clone(); int row_min = src.rows - 1; int row_max = 0; int col_min = src.cols - 1; int col_max = 0; int nl = src.rows; //行数 int nc = src.cols; //列数 for (int j = 0; j<nl; j++) { uchar* data = src.ptr<uchar>(j); for (int i = 0; i<nc; i++) { if (data[i] == 0) { if (j < row_min) row_min = j; if (j > row_max) row_max = j; if (i < col_min) col_min = i; if (i > col_max) col_max = i; } } } int ROI_width = col_max - col_min; int ROI_height = row_max - row_min; int size = (ROI_width>ROI_height) ? ROI_width : ROI_height;//学习这种用法 Mat imageROI = src(Rect(col_min, row_min, ROI_width, ROI_height)); Mat tem; resize(imageROI, tem, Size(new_width, new_height), CV_INTER_LINEAR); threshold(tem, tem, 100, 255, THRESH_OTSU); srcimage.release(); srcimage.create(1, new_width*new_height, CV_32FC1); float* data1 = srcimage.ptr<float>(0); nl = tem.rows; //行数 nc = tem.cols; //列数 for (int j = 0; j<nl; j++) { uchar* data = tem.ptr<uchar>(j); for (int i = 0; i<nc; i++) { if (data[i] == 255) data1[j*new_height + i] = 1; else data1[j*new_height + i] = 0; } } } basicOCR::basicOCR()//构造函数 { //initial sprintf(file_path, "OCR/"); train_samples = 50;//训练样本，总共100个，50个训练，50个测试 classes = 10;//暂时识别十个数字 size = 128;// trainData.create(train_samples*classes, size*size, CV_32FC1);//训练数据的矩阵 trainClasses.create(train_samples*classes, 1, CV_32FC1); //Get data (get images and process it) getData(); //train train(); //Test test(); printf(" ------------------------------------------------------------------------\n"); printf("|\t识别结果\t|\t 测试精度\t|\t 准确率\t|\n"); printf(" ------------------------------------------------------------------------\n"); } void basicOCR::getData() { Mat src_image; char file[255]; int i, j; for (i = 0; i<classes; i++) { for (j = 0; j< train_samples; j++) { //加载pbm格式图像，作为训练 if (j<10) sprintf(file, "%s%d/%d0%d.pbm", file_path, i, i, j); else sprintf(file, "%s%d/%d%d.pbm", file_path, i, i, j); src_image = imread(file, 0); if (src_image.empty()) { printf("Error: Cant load image %s\n", file); //exit(-1); } //process file preprocessing(src_image, size, size); //Set data float* data1 = trainData.ptr<float>(i*train_samples+j); float* data2 = src_image.ptr<float>(0); for (int k = 0; k < src_image.cols; k++) { data1[k] = data2[k]; } //Set class label trainClasses.at<float>(i*train_samples + j, 0) = i; } } } void basicOCR::train() { knn = new KNearest(trainData, trainClasses, Mat(), false, K); } float basicOCR::classify(Mat img, int showResult)//第二个参数主要用来控制是测试训练样本还是手写识别 { Mat nearest; nearest.create(1, K, CV_32FC1); float result; //处理输入的图像 preprocessing(img, size, size); Mat tem1, tem2; result = knn->find_nearest(img, K, tem1, nearest, tem2); int accuracy = 0; for (int i = 0; i<K; i++) { if (nearest.at<float>(0, i) == result) accuracy++; } float pre = 100 * ((float)accuracy / (float)K); /*if(showResult==1) { printf("|\t %.0f \t| \t %.2f%% \t| \t %d of %d \t| \n",result,pre,accuracy,K); printf(" ------------------------------------------------------------------------\n"); }*/ printf("|\t %.0f \t| \t %.2f%% \t| \t %d of %d \t| \n", result, pre, accuracy, K); printf(" ------------------------------------------------------------------------\n"); return result; } void basicOCR::test() { Mat src_image; IplImage prs_image; CvMat row, data; char file[255]; int i, j; int error = 0; int testCount = 0; for (i = 0; i<classes; i++) { for (j = 50; j< 50 + train_samples; j++)//五十个测试样本，计算一下错误率 { sprintf(file, "%s%d/%d%d.pbm", file_path, i, i, j); src_image = imread(file, 0); if (src_image.empty()) { printf("Error: Cant load image %s\n", file); exit(-1); } //process file //preprocessing(src_image, size, size);//归一化图片 float r = classify(src_image, 0);//检验图片 if ((int)r != i) error++; testCount++; } } float totalerror = 100 * (float)error / (float)testCount; printf("测试系统误识率: %.2f%%\n", totalerror); }