c++ - 如何使用 Opencv 删除背景图像

Question

我是新的 opencv 。我写了一个删除背景。
我的输入图像

我将我的程序编码为以下步骤：
- 计算平均像素

//define roi of image
cv::Rect roi(0, 0, 20 , 20 );

//copies input image in roi
cv::Mat image_roi = imgGray( roi );

//imshow("roi", image_roi);
//computes mean over roi
cv::Scalar avgPixelIntensity = cv::mean( image_roi );
//prints out only .val[0] since image was grayscale
cout << "Pixel intensity over ROI = " << avgPixelIntensity.val[0] << endl;

- 根据平均像素值创建新的 Mat 图像：

//create new mat image base on avgPixelIntensity
cv::Mat areaSampleArv(imgGray.rows, imgGray.cols,imgGray.type(),avgPixelIntensity.val[0]);
imshow("areaSampleArv", areaSampleArv);

- 反转图像：

void image_invert(Mat& image){
int height, width, step, channels;
uchar *data;

height = image.cols;
width  = image.rows;
step   = (int)image.step;
channels = image.channels();
data = (uchar *)image.data;

for(int i = 0; i < height; i++){
    for(int j = 0; j < width; j++){
        for(int k = 0; k < channels; k++){
            data[i*step + j*channels + k] = 255 - data[i*step + j*channels + k];
        }
    }
}

//imwrite("/Users/thuydungle/Desktop/1234/inverted.png", image);
imshow("inverted", image);}

我的图像反转结果：

- 使用原始图像添加倒置图像：

 Mat dst;
 dst = areaSampleArv + im0;
 imshow("dst", dst);

我的任何图像结果：

看起来很糟糕，我可以使用阈值来提取数字吗？
那么，你能告诉我如何解决吗？
感谢 ！

Answer 1

您可以尝试cv:inRange()基于颜色的阈值。

cv::Mat image = cv::imread(argv[1]);
if (image.empty())
{
    std::cout << "!!! Failed imread()" << std::endl;
    return -1;
}

cv::Mat threshold_image;

// MIN B:77 G:0 R:30    MAX B:130 G:68 R:50
cv::inRange(image, cv::Scalar(77, 0, 30), 
                   cv::Scalar(130, 68, 50), 
                   threshold_image);

cv::bitwise_not(threshold_image, threshold_image); 

cv::imwrite("so_inrange.png", threshold_image);

int erode_sz = 4;
cv::Mat element = cv::getStructuringElement(cv::MORPH_ELLIPSE,
                                   cv::Size(2*erode_sz + 1, 2*erode_sz+1),
                                   cv::Point(erode_sz, erode_sz) );

cv::erode(threshold_image, threshold_image, element);
cv::imwrite("so_erode.png", threshold_image);

cv::dilate(threshold_image, threshold_image, element);
cv::imwrite("so_dilate.png", threshold_image);

cv::imshow("Color Threshold", threshold_image);
cv::waitKey();

您也可以执行cv::blur(threshold_image, threshold_image, cv::Size(3, 3));aftercv::bitwise_not()以获得稍好的结果。

玩得开心改变代码。