2018-06-08

• ROI操作

    Mat coin = imread("coin.jpg");
    Mat im = imread("1.jpg");
    //提取出图片特定区域
    //Mat imROI(im, Rect(im.cols - coin.cols, im.rows - coin.rows, coin.cols, coin.rows));
    Mat imROI = im(Range(im.rows - coin.rows, im.rows), Range(im.cols - coin.cols, im.cols));
    coin.copyTo(imROI);
    imshow("", im);

• 访问像素点

Parallel Pixel Access in OpenCV using forEach

at方法实现椒盐噪声

随机选择一些像素，将其替换为白色或者黑色

void salt(cv::Mat image, int n) {
    int i, j;
    for (int k = 0; k<n; k++) {
        // rand()是随机数生成器
        i = std::rand() % image.cols;
        j = std::rand() % image.rows;
        if (image.type() == CV_8UC1) { // 灰度图像
            image.at<uchar>(j, i) = 255;
        }
        else if (image.type() == CV_8UC3) { // 彩色图像
            image.at<cv::Vec3b>(j, i)[0] = 255;
            image.at<cv::Vec3b>(j, i)[1] = 255;
            image.at<cv::Vec3b>(j, i)[2] = 255;
        }
    }
}

迭代器

void colorReduce(cv::Mat image, int div = 64) {
    auto it = image.begin<Vec3b>();

    while(it < image.end<Vec3b>()){
        // 处理每个像素 ---------------------
        (*it)[0] = (*it)[0] / div*div + div / 2;
        (*it)[1] = (*it)[1] / div*div + div / 2;
        (*it)[2] = (*it)[2] / div*div + div / 2;

        ++it;
    }
}

int main() {
    Mat im = imread("1.jpg");
    colorReduce(im);
    imshow("", im);


END:
    waitKey(0);
    system("pause");
    return 0;
}

• 反向投影（Back Projection）
  Back Projection
  给出ROI的直方图分布，主要要归一化，把每个柱子当作对应强度出现的概率，然后在测试图像中，每个像素点的强度去查询前面的概率（0~1），然后作为新图像的强度（也可以乘以255得到整形强度值）。
  程序流程，BGR->HSV->Hue
  然后取出ROI进行生成直方图数据
  然后再backproject在整个Hue上面。
  程序运行后，框一个框，然后回车或者是空格。
  对魔方的识别还不是很好，每种颜色都要自定义一些参数，然后会有稍微一些混淆。肤色识别的很好。
  
  
  
  
  
  
  代码：

string windowName = "back project";
const int maxSlideValue = 255;
int slideValue;
Mat src, hsv, hue, ROI, hist, imBackProject;
int threashold = 10;
int histSize = 10;

void on_trackbar1(int slideValue, void* imBackProject){
    threashold = slideValue;

    //分析ROI的直方图分布
    float hue_range[] = { 0, 255 };
    const float* ranges = { hue_range };
    calcHist(&ROI, 1, 0, Mat(), hist, 1, &histSize, &ranges);
    normalize(hist, hist, 0, 255, NORM_MINMAX);

    calcBackProject(&hue, 1, 0, hist, *(Mat *)imBackProject, &ranges, 1, true);

    Mat imBackProject2;
    threshold(*(Mat *)imBackProject, imBackProject2, threashold, 255, THRESH_BINARY);
    imshow(windowName, imBackProject2);
}


void on_trackbar2(int slideValue, void* imBackProject) {
    histSize = max(slideValue,2);

    //分析ROI的直方图分布
    float hue_range[] = { 0, 255 };
    const float* ranges = { hue_range };
    calcHist(&ROI, 1, 0, Mat(), hist, 1, &histSize, &ranges);
    normalize(hist, hist, 0, 255, NORM_MINMAX);

    calcBackProject(&hue, 1, 0, hist, *(Mat *)imBackProject, &ranges, 1, true);

    Mat imBackProject2;
    threshold(*(Mat *)imBackProject, imBackProject2, threashold, 255, THRESH_BINARY);
    imshow(windowName, imBackProject2);
}


int main() {


    //转化为HSV
    //src = imread("4cube.png");
    src = imread("hands.jpg");
    cvtColor(src, hsv, COLOR_BGR2HSV);
    hue.create(hsv.size(), hsv.depth());
    int ch[] = { 0, 0 };
    mixChannels(&hsv, 1, &hue, 1, ch, 1);

    //while (1) {
        //选出ROI
        Rect r = selectROI(src);
        ROI = hue(r);



        namedWindow(windowName);
        createTrackbar("阈值化的阈值", windowName, &slideValue, maxSlideValue, on_trackbar1, &imBackProject);
        createTrackbar("直方图纵轴个数", windowName, &slideValue, maxSlideValue, on_trackbar2, &imBackProject);
        on_trackbar1(threashold, &imBackProject);
        on_trackbar2(histSize, &imBackProject);

        //threshold(imBackProject, imBackProject, 30, 255, THRESH_BINARY);
        //imshow(windowName, imBackProject);
    //}

END:
    waitKey(0);
    system("pause");
    return 0;
}