在医疗图像检测领域，为了能够使得采集到效果较好的X光或超声图像，会经常需要进行图像增强，如下图所示。图像增强的方法有很多，本文主要针对直方图均衡化相关的方法进行C++实现。共分为以下几个部分：
- 1 直方图均衡化（HE）
- 2 自适应直方图均衡化（AHE）
- 3 限制对比度的自适应直方图均衡化
- 4 局部直方图统计量
直方图均衡化（HE）
直方图均衡化（histogram equlization）的主要思想是将一副图像的直方图分布变成近似均匀分布，从而增强图像的对比度。直方图均衡化虽然只是数字图像处理(Digital Image Processing)里面的基本方法，但是其作用很强大，是一种很经典的算法。
关于直方图均衡化原理介绍可以看这篇文章
OpenCV代码：
Mat srcMat = imread("./1.bmp", 0);
Mat histMat;
equalizeHist(srcMat, histMat);
自己手写代码如下：
//HE(直方图均衡化)
bool HE_MY(Mat gray, Mat result) {
//统计0~255像素值的个数
map<int, int>mp;
for (int i = 0; i < gray.rows; i++) {
uchar* ptr = gray.data + i * gray.cols;
for (int j = 0; j < gray.cols; j++) {
int value = ptr[j];
mp[value]++;
}
}
//统计0~255像素值的频率，并计算累计频率
map<int, double> valuePro;
int sum = gray.cols * gray.rows;
double sumPro = 0;
for (int i = 0; i < 256; i++) {
sumPro += 1.0 * mp[i] / sum;
valuePro[i] = sumPro;
}
//根据累计频率进行转换
for (int i = 0; i < gray.rows; i++) {
uchar* ptr1 = gray.data + i * gray.cols;
for (int j = 0; j < gray.cols; j++) {
int value = ptr1[j];
double p = valuePro[value];
result.at<uchar>(i, j) = p * 255;
}
}
return true;
}
自适应直方图均衡化（AHE）
自适应直方图均衡化（adaptive histogram equalization）其实是将图像切分为许多不同的区域, 对每个区域分别进行直方图均衡化，即局部直方图均衡化。通常会产生块状的效果，需要对其进行插值运算，使得块状区域之间过度较为自然。常用的插值算法有：最邻近插值、线性插值、双线性插值、双三次插值，opencv中的 resize() 函数就是使用的双线性插值的方法。具体介绍可以参考这篇文章：图像处理之-----插值算法
几种对比度拉伸算法的效果对比如下图所示。原图的阴影区域中存在一些对比度不明显的字符，使用全局拉伸的直方图均衡化后，阴影区域中的字符对比度拉伸强度较小，而且白色区域的一些噪声被放大了。使用AHE拉伸后，阴影区域的字符对比度明显，但对比度拉伸范围过大，使得阴影区域的亮度提高了很多，且放大了很多噪声点。使用CLAHE算法则能够得到较好的效果，将在下面进行介绍。
AHE（插值）算法代码如下：
Mat AHE(Mat originImage, int blockNumber = 16) {
Mat src = originImage.clone();
Mat src1 = originImage.clone();
Mat dst;
//const int blockNumber = 8;//把图像分成block数量
int width = src.cols;
int height = src.rows;
int singleBlockWidth = src.cols / blockNumber;//每个block大小
int singleBlockHeight = src.rows / blockNumber;
int (*pixelNumber)[256] = new int[blockNumber * blockNumber][256]{ 0 };//存储不同block中各个不同灰度像素数量
float (*total)[256] = new float[blockNumber * blockNumber][256]{ 0.0 };//累积分布函数
for (int i = 0; i < blockNumber; i++)
{
for (int j = 0; j < blockNumber; j++)
{
int startPixelW = (i)*singleBlockWidth;
int endPixelW = (i + 1) * singleBlockWidth;
int startPixelH = (j)*singleBlockHeight;
int endPixelH = (j + 1) * singleBlockHeight;
int number = i + blockNumber * j;//统计运算到哪一个block了
int singleBlockPixelNumber = singleBlockWidth * singleBlockHeight;
for (int x = startPixelW; x < endPixelW; x++)//统计不同block中各个不同灰度像素数量
for (int y = startPixelH; y < endPixelH; y++)
{
int pixelValue = src.at<uchar>(y, x);
pixelNumber[number][pixelValue]++;
}
for (int k = 0; k < 256; k++)//计算累积分布函数
{
if (k == 0)
total[number][k] = 1.0 * pixelNumber[number][k] / singleBlockPixelNumber;
else
total[number][k] = total[number][k - 1] + 1.0 * pixelNumber[number][k] / singleBlockPixelNumber;
}
}
}
//利用累积分布函数对于原像素灰度在各自block中进行映射
for (int i = 0; i < blockNumber; i++)
{
for (int j = 0; j < blockNumber; j++)
{
int startPixelW = (i)*singleBlockWidth;
int endPixelW = (i + 1) * singleBlockWidth;
int startPixelH = (j)*singleBlockHeight;
int endPixelH = (j + 1) * singleBlockHeight;
int number = i + blockNumber * j;
int singleBlockPixelNumber = singleBlockWidth * singleBlockHeight;
for (int x = startPixelW; x < endPixelW; x++)
for (int y = startPixelH; y < endPixelH; y++)
{
int pixelValue = src1.at<uchar>(y, x);
src1.at<uchar>(y, x) = total[number][pixelValue] * 255;
}
}
}
//插值
for (int i = 0; i < width; i++)
{
for (int j = 0; j < height; j++)
{
//four coners
if (i <= singleBlockWidth / 2 && j <= singleBlockHeight / 2)
{
int num = 0;
src.at<uchar>(j, i) = (int)(total[num][src.at<uchar>(j, i)] * 255);
}
else if (i <= singleBlockWidth / 2 && (j >= ((blockNumber - 1) * singleBlockHeight + singleBlockHeight / 2))) {
int num = blockNumber * (blockNumber - 1);
src.at<uchar>(j, i) = (int)(total[num][src.at<uchar>(j, i)] * 255);
}
else if (i >= ((blockNumber - 1) * singleBlockWidth + singleBlockWidth / 2) && j <= singleBlockHeight / 2) {
int num = blockNumber - 1;
src.at<uchar>(j, i) = (int)(total[num][src.at<uchar>(j, i)] * 255);
}
else if (i >= ((blockNumber - 1) * singleBlockWidth + singleBlockWidth / 2) && j >= ((blockNumber - 1) * singleBlockHeight + singleBlockHeight / 2)) {
int num = blockNumber * blockNumber - 1;
src.at<uchar>(j, i) = (int)(total[num][src.at<uchar>(j, i)] * 255);
}
//four edges except coners
else if (i <= singleBlockWidth / 2){
//线性插值
int num_i = 0;
int num_j = (j - singleBlockHeight / 2) / singleBlockHeight;
int num1 = num_j * blockNumber + num_i;
int num2 = num1 + blockNumber;
float p = (j - (num_j * singleBlockHeight + singleBlockHeight / 2)) / (1.0f * singleBlockHeight);
float q = 1 - p;
src.at<uchar>(j, i) = (int)((q * total[num1][src.at<uchar>(j, i)] + p * total[num2][src.at<uchar>(j, i)]) * 255);
}
else if (i >= ((blockNumber - 1) * singleBlockWidth + singleBlockWidth / 2)) {
//线性插值
int num_i = blockNumber - 1;
int num_j = (j - singleBlockHeight / 2) / singleBlockHeight;
int num1 = num_j * blockNumber + num_i;
int num2 = num1 + blockNumber;
float p = (j - (num_j * singleBlockHeight + singleBlockHeight / 2)) / (1.0f * singleBlockHeight);
float q = 1 - p;
src.at<uchar>(j, i) = (int)((q * total[num1][src.at<uchar>(j, i)] + p * total[num2][src.at<uchar>(j, i)]) * 255);
}
else if (j <= singleBlockHeight / 2) {
//线性插值
int num_i = (i - singleBlockWidth / 2) / singleBlockWidth;
int num_j = 0;
int num1 = num_j * blockNumber + num_i;
int num2 = num1 + 1;
float p = (i - (num_i * singleBlockWidth + singleBlockWidth / 2)) / (1.0f * singleBlockWidth);
float q = 1 - p;
src.at<uchar>(j, i) = (int)((q * total[num1][src.at<uchar>(j, i)] + p * total[num2][src.at<uchar>(j, i)]) * 255);
}
else if (j >= ((blockNumber - 1) * singleBlockHeight + singleBlockHeight / 2)) {
//线性插值
int num_i = (i - singleBlockWidth / 2) / singleBlockWidth;
int num_j = blockNumber - 1;
int num1 = num_j * blockNumber + num_i;
int num2 = num1 + 1;
float p = (i - (num_i * singleBlockWidth + singleBlockWidth / 2)) / (1.0f * singleBlockWidth);
float q = 1 - p;
src.at<uchar>(j, i) = (int)((q * total[num1][src.at<uchar>(j, i)] + p * total[num2][src.at<uchar>(j, i)]) * 255);
}
//双线性插值
else {
int num_i = (i - singleBlockWidth / 2) / singleBlockWidth;
int num_j = (j - singleBlockHeight / 2) / singleBlockHeight;
int num1 = num_j * blockNumber + num_i;
int num2 = num1 + 1;
int num3 = num1 + blockNumber;
int num4 = num2 + blockNumber;
float u = (i - (num_i * singleBlockWidth + singleBlockWidth / 2)) / (1.0f * singleBlockWidth);
float v = (j - (num_j * singleBlockHeight + singleBlockHeight / 2)) / (1.0f * singleBlockHeight);
src.at<uchar>(j, i) = (int)((u * v * total[num4][src.at<uchar>(j, i)] +
(1 - v) * (1 - u) * total[num1][src.at<uchar>(j, i)] +
u * (1 - v) * total[num2][src.at<uchar>(j, i)] +
v * (1 - u) * total[num3][src.at<uchar>(j, i)]) * 255);
}
}
}
return src;
}
限制对比度的自适应直方图均衡化（CLAHE）
CLAHE是1993年提出的方法，通过对局部直方图均衡化的对比度进行限制，从而避免所有局部区域的像素灰度值都映射到（0-255）的范围内，能够较好地保持原图的自然度，同时也能起到抑制噪声的作用。
上图a代表原图的直方图，b是其累积概率密度函数；c和d则为进行对比度限制后的直方图和累积概率分布函数。对比度限制原理如上图右部分所示，通过设置一个阈值（通常情况下该值表示为图像像素个数的百分比），将像素数较多的区域切除掉，平均分配到所有其他的灰度值区域。未使用限制对比度时，原图中灰度值为60的像素映射后变为217，灰度值为160的像素映射后变为242，限制对比度后，则分别变为了115和209。可见能够起到限制对比度拉伸和抑制噪声的作用。
具体细节部分可以参考原论文：Contrast Limited Adaptive Histogram Equalization
作者为什么可以有这种灵感我们不得而知，但对于这种局部对比度限制的方法博主也在许多其他论文中看到过，我认为应该没有比较科学的解释，只是因为这种方法确实能够限制对比度的拉伸，且得到的效果也挺不错的。
几种不同的图像增强效果如下：
OpenCV 代码如下
Mat claheMat;
Ptr<CLAHE> clahe = createCLAHE(40.0, Size(8, 8));
clahe->apply(blurMat, claheMat);
自己手写 代码如下：
//CLAHE
Mat CLAHE_MY(Mat src, int block = 8, int limit = 4)
{
Mat CLAHE_GO = src.clone();
int width = src.cols;
int height = src.rows;
int width_block = width / block; //每个小格子的长和宽
int height_block = height / block;
//存储各个直方图
int(*tmp2)[256] = new int[block * block][256]{ 0 };
float(*C2)[256] = new float[block * block][256]{ 0.0 };
//分块
int total = width_block * height_block;
for (int i = 0; i < block; i++)
{
for (int j = 0; j < block; j++)
{
int start_x = i * width_block;
int end_x = start_x + width_block;
int start_y = j * height_block;
int end_y = start_y + height_block;
int num = i + block * j;
//遍历小块,计算直方图
for (int ii = start_x; ii < end_x; ii++)
{
for (int jj = start_y; jj < end_y; jj++)
{
int index = src.at<uchar>(jj, ii);
tmp2[num][index]++;
}
}
//裁剪和增加操作
int LIMIT = limit * width_block * height_block / 255;
int steal = 0;
for (int k = 0; k < 256; k++)
{
if (tmp2[num][k] > LIMIT) {
steal += tmp2[num][k] - LIMIT;
tmp2[num][k] = LIMIT;
}
}
int bonus = steal / 256;
//hand out the steals averagely
for (int k = 0; k < 256; k++)
{
tmp2[num][k] += bonus;
}
//计算累积分布直方图
for (int k = 0; k < 256; k++)
{
if (k == 0)
C2[num][k] = 1.0f * tmp2[num][k] / total;
else
C2[num][k] = C2[num][k - 1] + 1.0f * tmp2[num][k] / total;
}
}
}
//计算变换后的像素值
//根据像素点的位置，选择不同的计算方法
for (int i = 0; i < width; i++)
{
for (int j = 0; j < height; j++)
{
//four coners
if (i <= width_block / 2 && j <= height_block / 2)
{
int num = 0;
CLAHE_GO.at<uchar>(j, i) = (int)(C2[num][CLAHE_GO.at<uchar>(j, i)] * 255);
}
else if (i <= width_block / 2 && j >= ((block - 1) * height_block + height_block / 2)) {
int num = block * (block - 1);
CLAHE_GO.at<uchar>(j, i) = (int)(C2[num][CLAHE_GO.at<uchar>(j, i)] * 255);
}
else if (i >= ((block - 1) * width_block + width_block / 2) && j <= height_block / 2) {
int num = block - 1;
CLAHE_GO.at<uchar>(j, i) = (int)(C2[num][CLAHE_GO.at<uchar>(j, i)] * 255);
}
else if (i >= ((block - 1) * width_block + width_block / 2) && j >= ((block - 1) * height_block + height_block / 2)) {
int num = block * block - 1;
CLAHE_GO.at<uchar>(j, i) = (int)(C2[num][CLAHE_GO.at<uchar>(j, i)] * 255);
}
//four edges except coners
else if (i <= width_block / 2)
{
//线性插值
int num_i = 0;
int num_j = (j - height_block / 2) / height_block;
int num1 = num_j * block + num_i;
int num2 = num1 + block;
float p = (j - (num_j * height_block + height_block / 2)) / (1.0f * height_block);
float q = 1 - p;
CLAHE_GO.at<uchar>(j, i) = (int)((q * C2[num1][CLAHE_GO.at<uchar>(j, i)] + p * C2[num2][CLAHE_GO.at<uchar>(j, i)]) * 255);
}
else if (i >= ((block - 1) * width_block + width_block / 2)) {
//线性插值
int num_i = block - 1;
int num_j = (j - height_block / 2) / height_block;
int num1 = num_j * block + num_i;
int num2 = num1 + block;
float p = (j - (num_j * height_block + height_block / 2)) / (1.0f * height_block);
float q = 1 - p;
CLAHE_GO.at<uchar>(j, i) = (int)((q * C2[num1][CLAHE_GO.at<uchar>(j, i)] + p * C2[num2][CLAHE_GO.at<uchar>(j, i)]) * 255);
}
else if (j <= height_block / 2) {
//线性插值
int num_i = (i - width_block / 2) / width_block;
int num_j = 0;
int num1 = num_j * block + num_i;
int num2 = num1 + 1;
float p = (i - (num_i * width_block + width_block / 2)) / (1.0f * width_block);
float q = 1 - p;
CLAHE_GO.at<uchar>(j, i) = (int)((q * C2[num1][CLAHE_GO.at<uchar>(j, i)] + p * C2[num2][CLAHE_GO.at<uchar>(j, i)]) * 255);
}
else if (j >= ((block - 1) * height_block + height_block / 2)) {
//线性插值
int num_i = (i - width_block / 2) / width_block;
int num_j = block - 1;
int num1 = num_j * block + num_i;
int num2 = num1 + 1;
float p = (i - (num_i * width_block + width_block / 2)) / (1.0f * width_block);
float q = 1 - p;
CLAHE_GO.at<uchar>(j, i) = (int)((q * C2[num1][CLAHE_GO.at<uchar>(j, i)] + p * C2[num2][CLAHE_GO.at<uchar>(j, i)]) * 255);
}
//双线性插值
else {
int num_i = (i - width_block / 2) / width_block;
int num_j = (j - height_block / 2) / height_block;
int num1 = num_j * block + num_i;
int num2 = num1 + 1;
int num3 = num1 + block;
int num4 = num2 + block;
float u = (i - (num_i * width_block + width_block / 2)) / (1.0f * width_block);
float v = (j - (num_j * height_block + height_block / 2)) / (1.0f * height_block);
CLAHE_GO.at<uchar>(j, i) = (int)((u * v * C2[num4][src.at<uchar>(j, i)] +
(1 - v) * (1 - u) * C2[num1][src.at<uchar>(j, i)] +
u * (1 - v) * C2[num2][src.at<uchar>(j, i)] +
v * (1 - u) * C2[num3][src.at<uchar>(j, i)]) * 255);
}
}
}
return CLAHE_GO;
}
局部直方图统计量
局部直方图统计量的方法在冈萨雷斯的《数字图像处理》第四版中有介绍，比较简单好用，对于特定的场景往往能起到更好的效果。
代码如下：
//局部直方图统计量
/*
功能：对于图像中的每个像素点，截取其7*7的邻域框，计算邻域框的灰度均值mxy和方差deltaxy1，与整幅图像的邻域mean和方差delta1进行比较，
if (mxy <= mean * k0 && deltaxy1 <= k2 * delta1 && deltaxy1 >= k1 * delta1){ 将该像素灰度值*10 };
*/
double getMean(double* r_pdf) {
double m = 0;
for (int i = 0; i < 256; i++)
m += i * r_pdf[i];
return m;
}
double getVariance(double* r_pdf, double m)
{
double delta = 0;
for (int i = 0; i < 256; i++)
if (r_pdf[i] != 0)
delta += (pow((i - m), 2) * r_pdf[i]);
return delta;
}
void init(double* r_pdf) {
for (int i = 0; i < 256; i++) {
r_pdf[i] = 0;
}
}
int histStatistic(cv::Mat& src, int dim, float k0, float k1, float k2, float E) {
if (dim % 2 == 0) {
return -1;
}
int width = src.rows;
int height = src.cols;
int mn = width * height;
double r_pdf[256] = {};
for (int row = 0; row < src.rows; row++) {
for (int col = 0; col < src.cols; col++) {
int g = src.at<uchar>(row, col);
r_pdf[g] += 1.0 / mn;
}
}
double mean = getMean(r_pdf);
double delta = getVariance(r_pdf, mean);
double delta1 = std::sqrt(delta);
double mxy = 0;
double deltaxy = 0;
double local = dim * dim;
for (int i = dim / 2; i < height - dim / 2 - 1; i++)
{
for (int j = dim / 2; j < width - dim / 2; j++) {
init(r_pdf);
//统计局部直方图
for (int p = j - dim / 2; p < j + dim / 2 + 1; p++) {
for (int q = i - dim / 2; q < i + dim / 2 + 1; q++) {
int g = src.at<uchar>(p, q);
r_pdf[g] += 1.0 / local;
}
}
mxy = getMean(r_pdf);
deltaxy = getVariance(r_pdf, mxy);
double deltaxy1 = sqrt(deltaxy);
if (mxy <= mean * k0 && deltaxy1 <= k2 * delta1 && deltaxy1 >= k1 * delta1) {
src.at<uchar>(j, i) = src.at<uchar>(j, i) * E;
}
}
}
}
998
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
