goldroses's profile - activity

I would like to reduce the 3 matrixes to just 2 matrixes by converting and flipping in one command, or flipping in place. Is there any way how to increase efficiency of the code?

cv::Mat dImg; // assume its initalized with TYPE_16UC1
cv::Mat dImg_grey;
cv::Mat dImg_grey_flip;
dImg.convertTo(dImg_grey, CV_8UC1, 255. / (2500 - 0), 0);//  to grayscale cv::Mat, (max - min)
cv::flip(dImg_grey, dImg_grey_flip, 1);// cannot flip in place

Is it somehow possible to stuff the flip into the convert.

here is the code, which does not compile also its so basic.

     cv::Mat dImgnotmirr;
     cv::Mat dImg;
    dImg.create(dImgnotmirr.size(),dImgnotmirr.type());
    cv::flip(&dImgnotmirr, &dImg, 1);

error:

error: invalid initialization of reference of type ‘cv::InputArray {aka const cv::_InputArray&}’ from expression of type ‘cv::Mat*’ cv::flip(&dImgmirr, &dImg, 1); ^

Thank you good idea. Also this helped. Furthermore I could not calibrate because the ir light source of emitter is to weak. Would have to use additional halogen which too much . So i took parameters from here

I want to calibrate my kinect camera. The RGB-Images work fine with ./cpp-example-calibration but what do I do with the depth information. How do I calibrate these pictures or can i at all? Because the cv::mat is filled with depth information and not with RGB-Pixels. What sample to use with it? This is an example:

I want to set a pixel which meets certain criteria to 4096 (2^12), but what I end up with is 14; Its an cv::Mat dImg in the format 16UC1

        for(int j=0;j<dImg.rows;j++) 
        {
        uchar* pixrow = dImg.ptr<uchar>(j); 

        for (int i=0;i<dImg.cols;i++)
        {
            if(j == 240 && pixrow[i] ==0)  dImg.at<uchar>(j,i) = 2^12;

also this results in cout=14:

                if(j == 240 && pixrow[i] ==0)  pixrow[i]  = 2^12; //= 4096

I check the value with

 cout<<"\nmodified: "<<endl;
for(int i = 0; i<dImg.cols;i++){
  cout<<" "<<unsigned(dImg.at<uchar>(240,i));
}

SOLUTION (thank your for answering)

 for(int j=0;j<dImg.rows;j++) 
    {
        ushort* pixrow = dImg.ptr<ushort>(j);  //not uchar

        for (int i=0;i<dImg.cols;i++)
        {
            if(j == 239 && pixrow[i] ==0)  pixrow[i]  = 4096;

just to add: I print every 3 seconds one line of my cv::Mat encoded as 16UC1 as my video-feed has @30fps

    int count; cv::Mat dImg;
    if(count == 90){
        for(int i = 0; i<dImg.cols;i++){
            cout<<" "<<unsigned(dImg.at<uchar>(240,i));
        }
        count = 0; 
    }   
    count++;

I tried to get the value underneath an cv::Mat dImg in the format 16UC1 but only get rubbish when obtaining one pixel value. I expect an int ranging inbetween 0 and 2^16-1 (65535) but this is what i get:

135530515
583506521733990419
4.45466e-34
1.18455e-269

with

       cout<<dImg.at<int>(240,320)<<endl;
       cout<<dImg.at<long>(240,320)<<endl;
       cout<<dImg.at<float>(240,320)<<endl;
       cout<<dImg.at<double>(240,320)<<endl;

EDIT: The image resolution is 640x480. I choose the position randomly, for others i get the same. (c++). EDIT2: Using ansers solution i get:

    125
    11
    157
    8
    0
    0

This looks fine yay :)