opencv-图像二值化、阈值获取、直方图、轮廓提取

Code:

#include <cv.h>

#include <highgui.h>

#include <windows.h>

// 获取直方图

// 1. pImageData 图像数据

// 2. nWidth 图像宽度

// 3. nHeight 图像高度

// 4. nWidthStep 图像行大小

// 5. pHistogram 直方图

BOOL GetHistogram(unsigned char *pImageData, int nWidth, int nHeight, int nWidthStep,

int *pHistogram)

{

int i = 0;

int j = 0;

unsigned char *pLine = NULL;

// 清空直方图

memset(pHistogram, 0, sizeof(int) * 256);

for (pLine = pImageData, j = 0; j < nHeight; j++, pLine += nWidthStep)

{

for (i = 0; i < nWidth; i++)

{

pHistogram[pLine[i]]++;

}

}

return TRUE;

}

// 大津法取阈值

// 1. pImageData 图像数据

// 2. nWidth 图像宽度

// 3. nHeight 图像高度

// 4. nWidthStep 图像行大小

// 函数返回阈值

int Otsu(unsigned char *pImageData, int nWidth, int nHeight, int nWidthStep)

{

int i = 0;

int j = 0;

int nTotal = 0;

int nSum = 0;

int A = 0;

int B = 0;

double u = 0;

double v = 0;

double dVariance = 0;

double dMaximum = 0;

int nThreshold = 0;

int nHistogram[256];

// 获取直方图

GetHistogram(pImageData, nWidth, nHeight, nWidthStep, nHistogram);

for (i = 0; i < 256; i++)

{

nTotal += nHistogram[i];

nSum += (nHistogram[i] * i);

}

for (j = 0; j < 256; j++)

{

A = 0;

B = 0;

for (i = 0; i < j; i++)

{

A += nHistogram[i];

B += (nHistogram[i] * i);

}

if (A > 0)

{

u = B / A;

}

else

{

u = 0;

}

if (nTotal - A > 0)

{

v = (nSum - B) / (nTotal - A);

}

else

{

v = 0;

}

dVariance = A * (nTotal - A) * (u - v) * (u - v);

if (dVariance > dMaximum)

{

dMaximum = dVariance;

nThreshold = j;

}

}

return nThreshold;

}

// 二值化

// 1. pImageData 图像数据

// 2. nWidth 图像宽度

// 3. nHeight 图像高度

// 4. nWidthStep 图像行大小

// 5. nThreshold 阈值

BOOL Threshold(unsigned char *pImageData, int nWidth, int nHeight, int nWidthStep,

unsigned int nThreshold)

{

int i = 0;

int j = 0;

unsigned char *pLine = NULL;

for (pLine = pImageData, j = 0; j < nHeight; j++, pLine += nWidthStep)

{

for (i = 0; i < nWidth; i++)

{

if (pLine[i] >= nThreshold)

{

pLine[i] = 0x00;

}

else

{

pLine[i] = 0xff;

}

}

}

return TRUE;

}

// 二值化图像取反

void Not(IplImage *edge)

{

int i;

int j;

for(i=0;i<edge->height;i++)

for(j=0;j<edge->width;j++)

{

if(edge->imageData[edge->height*i+j]==0)

edge->imageData[edge->height*i+j]=1;

else

edge->imageData[edge->height*i+j]=0;

}

}

BOOL FindContours(unsigned char *pImageData, int nWidth, int nHeight, int nWidthStep)

{

int i = 0;

int j = 0;

unsigned char *pLine[3] = { NULL, NULL, NULL };

for (j = 1; j < nHeight - 1; j++)

{

pLine[0] = pImageData + nWidthStep * (j - 1);

pLine[1] = pImageData + nWidthStep * j;

pLine[2] = pImageData + nWidthStep * (j + 1);

for (i = 1; i < nWidth - 1; i++)

{

if (pLine[0][i-1] == 0xFF && pLine[0][i] == 0xFF && pLine[0][i+1] == 0xFF &&

pLine[1][i-1] == 0xFF && pLine[1][i] == 0xFF && pLine[1][i+1] == 0xFF &&

pLine[2][i-1] == 0xFF && pLine[2][i] == 0xFF && pLine[2][i+1] == 0xFF)

{

pLine[0][i-1] = 0;

}

else

{

pLine[0][i-1] = pLine[1][i];

}

}

}

return TRUE;

}

int main(int argc, char* argv[])

{

IplImage *src = cvLoadImage("e://five.jpg", CV_LOAD_IMAGE_GRAYSCALE);

IplImage *dsc = cvCreateImage(cvSize(src->width*0.5,src->height*0.5), src->depth, src->nChannels);

// 转换图像大小

cvResize(src, dsc, CV_INTER_LINEAR);

src->width*=0.5;

src->height*=0.5;

// 复制图像

// 两个图像的大小类型通道数必须相同

cvCopy(dsc, src );

cvNamedWindow("src", CV_WINDOW_AUTOSIZE);

cvShowImage("src", src);

// 取图像阀值

unsigned int threshold = Otsu((unsigned char *)src->imageData, src->width, src->height, src->widthStep);

// 图像二值化

if (TRUE == Threshold((unsigned char *)src->imageData, src->width, src->height, src->widthStep,

threshold))

{

cvNamedWindow("dsc", CV_WINDOW_AUTOSIZE);

cvShowImage("dsc", src);

// 图像取反

// cvNot(src, src);

FindContours((unsigned char *)src->imageData,src->width , src->height, src->widthStep);

cvNamedWindow("not", CV_WINDOW_AUTOSIZE);

cvShowImage("not", src);

cvWaitKey(0);

}

cvDestroyAllWindows();

cvReleaseImage(&src);

cvReleaseImage(&dsc);

printf("Hello World!/n");

return 0;

}