图像增强——Roberts算子、Sobel算子、Prewitt算子、Laplace算子

#include<stdio.h>
#include"bmp.h"
#include "stdlib.h"
#include "math.h"
#include "time.h"
#include <iostream>
#include<string>
using namespace std;#define PI 3.14159
#define LENGTH_NAME_BMP 30//文件名长度#define U 0//高斯分布的均值
#define D 0//高斯分布的均方差
FILE *fpi, *fpw;
BITMAPFILEHEADER strHead;
RGBQUAD strPla[256];
BITMAPINFOHEADER strInfo;int Roberts1[4] = { -1,0,0,1 };
int Roberts2[4] = { 0,-1,1,0 };
int Sobel1[9] = { -1,-2,-1,0,0,0,1,2,1 };
int Sobel2[9] = { -1,0,1,-2,0,2,-1,0,1 };
int Prewitt1[9] = { -1,-1,-1,0,0,0,1,1,1 };
int Prewitt2[9] = { -1,0,1,-1,0,1,-1,0,1 };
int Laplace1[9] = { 0,1,0,1,-4,1,0,1,0 };
int Laplace2[9] = { 1,1,1,1,-8,1,1,1,1 };//显示位图文件头信息
void showBmpHead(BITMAPFILEHEADER pBmpHead)
{cout << "位图文件头" << endl;cout << "文件大小" << pBmpHead.bfSize << endl;cout << "保留字_1" << pBmpHead.bfReserved1 << endl;cout << "保留字_2" << pBmpHead.bfReserved2 << endl;cout << "实际位图数据的偏移字节数:" << pBmpHead.bf0ffBits << endl;
}void showBmpInforHead(tagBITMAPINFOHEADER pBmpInforHead)
{cout << "位图信息头" << endl;cout << "结构体的长度" << pBmpInforHead.biSize << endl;cout << "位图宽:" << pBmpInforHead.biWidth << endl;cout << "位图高：" << pBmpInforHead.biHeight << endl;cout << "biPlanes平面数:" << pBmpInforHead.biPlanes << endl;cout << "biBitCount采用颜色位数" << pBmpInforHead.biBitCount << endl;cout << "压缩方式" << pBmpInforHead.biCompression << endl;cout << "biSizeImage实际位图数据占用的字节数" << pBmpInforHead.biSizeImage << endl;cout << "X方向分辨率:" << pBmpInforHead.biXPelsPerMeter << endl;cout << "Y方向分辨率:" << pBmpInforHead.biYPelsPerMeter << endl;cout << "使用的颜色数:" << pBmpInforHead.biClrUsed << endl;cout << "重要颜色数:" << pBmpInforHead.biClrImportant << endl;
}int main()
{int Gaus_S();void conv2(IMAGEDATA *filter, int arr1[], int arr2[], int res[], int filterW, int filterH, int arrW, int arrH);void RobertsMethod(IMAGEDATA *filter, int filterW, int filterH);void Laplace(IMAGEDATA *filter, int filterW, int filterH);char strFile[LENGTH_NAME_BMP];//bmp文件名IMAGEDATA *imagedata = NULL;//动态分配存储原图片的像素信息的二维数组IMAGEDATA *imagedataRot = NULL;//动态分配处理后的图片像素信息int width, height;//图片的宽度和高度cout << "请输入所要读取的文件名（末尾要带有.bmp）:" << endl;cin >> strFile;fopen_s(&fpi,strFile,"rb");//读取bmp图片信息、初始化数组if (fpi != NULL){WORD bfType;fread(&bfType, 1, sizeof(WORD), fpi);if (0x4d42 != bfType){cout << "the file is not a bmp file!" << endl;return NULL;}//读取bmp文件的文件头和信息头fread(&strHead, 1, sizeof(tagBITMAPFILEHEADER), fpi);showBmpHead(strHead);fread(&strInfo, 1, sizeof(tagBITMAPINFOHEADER), fpi);showBmpInforHead(strInfo);for (unsigned int nCounti = 0; nCounti < strInfo.biClrUsed; nCounti++){fread((char *)&strPla[nCounti].rgbBlue, 1, sizeof(BYTE), fpi);fread((char *)&strPla[nCounti].rgbGreen, 1, sizeof(BYTE), fpi);fread((char *)&strPla[nCounti].rgbRed, 1, sizeof(BYTE), fpi);cout << "strPla[nCounti].rgbBlue" << strPla[nCounti].rgbBlue << endl;cout << "strPla[nCounti].rgbGreen" << strPla[nCounti].rgbGreen << endl;cout << "strPla[nCounti].rgbRed" << strPla[nCounti].rgbRed << endl;}width = strInfo.biWidth;height = strInfo.biHeight;imagedata = (IMAGEDATA *)malloc(width*height*sizeof(IMAGEDATA));//初始化原始图片的像素数组for (int i = 0; i < height; ++i){for (int j = 0; j < width; ++j){(*(imagedata + i*width + j)).blue = 0;(*(imagedata + i*width + j)).green = 0;(*(imagedata + i*width + j)).red = 0;}}fread(imagedata, sizeof(struct tagIMAGEDATA)*width, height, fpi);fclose(fpi);//加入高斯噪声int x, y, p;srand((unsigned)time(NULL));for (y = 0; y < height; y++){for (x = 0; x < width; x++){(*(imagedata + y*width + x)).blue += Gaus_S();(*(imagedata + y*width + x)).green += Gaus_S();(*(imagedata + y*width + x)).red += Gaus_S();}}//Roberts算子int method;cout << "选择Roberts算子请输入1,Sobel算子请输入2，Prewitt算子请输入3,Laplace算子请输入4：" << endl;cin >> method;switch (method){case 1:RobertsMethod(imagedata, width, height);break;case 2:int res1[25];for (y = 1; y < height - 2; y++){for (x = 1; x < width - 2; x++){conv2(imagedata + y*width + x, Sobel1, Sobel2, res1, 3, 3, 3, 3);}}break;case 3:int res2[25];for (y = 1; y < height - 2; y++){for (x = 1; x < width - 2; x++){conv2(imagedata + y*width + x, Prewitt1, Prewitt2, res2, 3, 3, 3, 3);}}break;case 4:int res3[25];for (y = 1; y < height - 2; y++){for (x = 1; x < width - 2; x++){conv2(imagedata + y*width + x, Laplace1, Laplace2, res3, 3, 3, 3, 3);}}break;default:cout << "输入错误" << endl;}}else{cout << "file open error" << endl;system("pause");return NULL;}//保存bmp图片cout << "保存的文件名(必须在文件名后输入.bmp):" << endl;char s[LENGTH_NAME_BMP];cin >> s;if (fopen_s(&fpw, s, "wb") != NULL){cout << "create the bmp file error!" << endl;system("pause");return NULL;}WORD bfType_w = 0x4d42;fwrite(&bfType_w, 1, sizeof(WORD), fpw);fwrite(&strHead, 1, sizeof(tagBITMAPFILEHEADER), fpw);strInfo.biWidth =  width;strInfo.biHeight =  height;fwrite(&strInfo, 1, sizeof(tagBITMAPINFOHEADER), fpw);//保存调色板数据for (unsigned int nCounti = 0; nCounti < strInfo.biClrUsed; nCounti++){fwrite(&strPla[nCounti].rgbBlue, 1, sizeof(BYTE), fpw);fwrite(&strPla[nCounti].rgbGreen, 1, sizeof(BYTE), fpw);fwrite(&strPla[nCounti].rgbRed, 1, sizeof(BYTE), fpw);}//保存像素数据for (int i = 0; i < height; ++i){for (int j = 0; j <  width; j++){fwrite(&(*(imagedata + i * width + j)).red, 1, sizeof(BYTE), fpw);fwrite(&(*(imagedata + i * width + j)).green, 1, sizeof(BYTE), fpw);fwrite(&(*(imagedata + i * width + j)).blue, 1, sizeof(BYTE), fpw);}}fclose(fpw);//释放内存delete[] imagedata;system("pause");return 0;}//产生高斯样本，以U为均值，D为均方差
int Gaus_S() {double sum = 0;for (int i = 0; i < 12; i++)sum += rand() / 32767.00;return int(U + D*(sum - 6));
}//卷积计算
void conv2(IMAGEDATA *filter, int arr1[],int arr2[],int res[],int filterW,int filterH, int arrW, int arrH)
{int temp1;int temp2;int i = 0, j = 0;//bluefor (; i < filterH+arrH-1; i++){for (; j <filterW+ arrW-1; j++){temp1 = 0;temp2 = 0;for (int m = 0; m < filterH; m++){for (int n = 0; n < filterW; n++){if ((i - m) >= 0 && (i - m) < arrH && (j - n) >= 0 && (j - n) < arrW){temp1 += (*(filter+m*filterW+n)).blue*arr1[(i - m)*arrW+j-n];temp2 += (*(filter + m*filterW + n)).blue*arr2[(i - m)*arrW+j-n];}}}res[i*filterW+j]= (temp1^2+temp2^2)^(1/2);}}(*(filter)).blue = res[((i-1)/2)*filterW+(j-1)/2];//redfor (i=0; i < filterH + arrH - 1; i++){for (j=0; j <filterW + arrW - 1; j++){temp1 = 0;temp2 = 0;for (int m = 0; m < filterH; m++){for (int n = 0; n < filterW; n++){if ((i - m) >= 0 && (i - m) < arrH && (j - n) >= 0 && (j - n) < arrW){temp1 += (*(filter + m*filterW + n)).red*arr1[(i - m)*arrW + j - n];temp2 += (*(filter + m*filterW + n)).red*arr2[(i - m)*arrW + j - n];}}}res[i*filterW + j] = (temp1 ^ 2 + temp2 ^ 2)^(1/2);}}(*(filter)).red = res[((i - 1) / 2)*filterW + (j - 1) / 2];//greenfor (i = 0; i < filterH + arrH - 1; i++){for (j = 0; j <filterW + arrW - 1; j++){temp1 = 0;temp2 = 0;for (int m = 0; m < filterH; m++){for (int n = 0; n < filterW; n++){if ((i - m) >= 0 && (i - m) < arrH && (j - n) >= 0 && (j - n) < arrW){temp1 += (*(filter + m*filterW + n)).green*arr1[(i - m)*arrW + j - n];temp2 += (*(filter + m*filterW + n)).green*arr2[(i - m)*arrW + j - n];}}}res[i*filterW + j] = (temp1 ^ 2 + temp2 ^ 2)^(1/2);}}(*(filter)).green = res[((i - 1) / 2)*filterW + (j - 1) / 2];
}
//Roberts算子计算函数
void RobertsMethod(IMAGEDATA *filter,int filterW,int filterH)
{for (int m = 1; m < filterH-2; m++)for (int n = 1; n < filterW-2; n++){(*(filter + m*filterW + n)).blue = (((*(filter + m*filterW + n)).blue - (*(filter + (m + 1)*filterW + n + 1)).blue) ^ 2 + ((*(filter + m*filterW + n + 1)).blue - (*(filter + (m + 1)*filterW + n)).blue)) ^ (1 / 2);(*(filter + m*filterW + n)).green = (((*(filter + m*filterW + n)).green - (*(filter + (m + 1)*filterW + n + 1)).green) ^ 2 + ((*(filter + m*filterW + n + 1)).green - (*(filter + (m + 1)*filterW + n)).green)) ^ (1 / 2);(*(filter + m*filterW + n)).red = (((*(filter + m*filterW + n)).red - (*(filter + (m + 1)*filterW + n + 1)).red) ^ 2 + ((*(filter + m*filterW + n + 1)).red - (*(filter + (m + 1)*filterW + n)).red)) ^ (1 / 2);}
}

图像增强——Roberts算子、Sobel算子、Prewitt算子、Laplace算子相关推荐

几种边缘检测算子的比较Roberts，Sobel，Prewitt，LOG，Canny
from:https://blog.csdn.net/gdut2015go/article/details/46779251 边缘检测是图像处理和计算机视觉中的基本问题,边缘检测的目的是标识数字图像中 ...
sobel、prewitt、Isotropic算子处理加噪图片
#Isotropic算子 import cv2 import numpy as np from PIL import Image import math import matplotlib.pyplo ...
图像边缘检测——一阶微分算子 Roberts、Sobel、Prewitt、Kirsch、Robinson
图像为什么会有边缘? 图像边缘一般指图像的灰度变化率最大的位置.成因主要如下: 1.图像灰度在表面方向变化不连续: 2.图像中物体在空间上的深度不一致: 3.在光滑的表面上颜色不一致: 4.图像中物体 ...
图像边缘检测——一阶微分算子 Roberts、Sobel、Prewitt、Kirsch、Robinson（Matlab实现）
图像边缘一般指图像的灰度变化率最大的位置.成因主要如下: 1.图像灰度在表面法向变化不连续: 2.图像中物体在空间上的深度不一致: 3.在光滑的表面上颜色不一致: 4.图像中物体的光影边缘检测指的是 ...
OpenCV3学习（6.1）——边缘检测---Canny,Sobel,Prewitt,Robert,Laplace,LOG,DOG算子
图像边缘信息主要集中在高频段,通常说图像锐化或检测边缘,实质就是高频滤波.我们知道微分运算是求信号的变化率,具有加强高频分量的作用.在空域运算中来说,对图像的锐化就是计算微分.由于数字图像的离散信号, ...
【图像处理】——Python图像分割边缘检测算法之一阶梯度算子（Roberts、Prewitt、Sobel、 Kirsch、Canny算子）
目录前言一.边缘检测算法 1.一阶算子 2.二阶算子二.一阶算子原图像lena 1.Roberts算子不同方向的算子模板梯度的计算系统代码: 自定义函数代码结果 2.Prewitt 不 ...
图像处理中的梯度、导数如何求?(Robert算子,Sobel算子,Prewitt算子,Laplace算子)
梯度的求法是多种多样的,根据不同的处理需要选择合适的算子(模版). 1.水平垂直差分法 2.Robert 梯度算子 3.Sobel算子垂直方向 ...
【OpenCV 】Sobel 导数/Laplace 算子/Canny 边缘检测
canny边缘检测见OpenCV [七]----边缘提取算子(图像边缘提取)--canny算法的原理及实现 1 Sobel 导数 1.1.1 原因上面两节我们已经学习了卷积操作.一个最重要的卷积运算 ...
python图像边缘提取_python通过robert、sobel、Laplace算子实现图像边缘提取详解
实现思路: 1,将传进来的图片矩阵用算子进行卷积求和(卷积和取绝对值) 2,用新的矩阵(与原图一样大小)去接收每次的卷积和的值 3,卷积图片所有的像素点后,把新的矩阵数据类型转化为uint8 注意: ...
GDAL 遥感图像处理锐化（Laplace算子、Sobel算子）
GDAL 图像锐化简介拉普拉斯(Laplace)算子部分代码: 索贝尔(Sobel)算子部分代码: 处理效果原图 (Laplace) (Sobel) 结尾参考文章简介图像锐化(imag ...

图像增强——Roberts算子、Sobel算子、Prewitt算子、Laplace算子

图像增强——Roberts算子、Sobel算子、Prewitt算子、Laplace算子相关推荐

最新文章

热门文章