ringholm's profile - activity

The question is simple to ask, but the anseer is probably not that simple.

I am rather curious as to how OpenCV determines the constants for the camera matrix.

Does it work by detecting the edges of the checkerboard, and then set the x and y axis to align with the edges of the squares? Then when calculating on multiple images from different angles, the fx,fy and cx,cy can be calculated?

But then what about the distortion coefficients? How are they determined?

I am trying to perform measurements on a depth image, by selecting a region of interest in the RGB image. However there seem to be an offset between the RGB/depth image, not to mention there are some distortion on the depth image. I have looked into different methods for calibrating, all of which employs use of matrices. The thing is I need to identify some the intrinsic parameters, for instance focal length on both axises, fx and fy and k1, k2, k3 and p1,p2. It sees that I can't make my camera work with OpenCV (only the RGB camera), so here goes: How do I obtain the parameters required? And how do I implement them? Btw. I am using LabVIEW for measuring the depth data.

I am trying to perform measurements on a depth image, by selecting a region of interest in the RGB image. However there seem to be an offset between the RGB/depth image, not to mention there are some distortion on the depth image. I have looked into different methods for calibrating, all of which employs use of matrices. The thing is I need to identify some the intrinsic parameters, for instance focal length on both axises, fx and fy and k1, k2, k3 and p1,p2. It sees that I can't make my camera work with OpenCV (only the RGB camera), so here goes: How do I obtain the parameters required? And how do I implement them? Btw. I am using LabVIEW for measuring the depth data.

I am trying to perform measurements on a depth image, by selecting a region of interest in the RGB image. However there seem to be an offset between the RGB/depth image, not to mention there are some distortion on the depth image. I have looked into different methods for calibrating, all of which employs use of matrices. The thing is I need to identify some the intrinsic parameters, for instance focal length on both axises, fx and fy and k1, k2, k3 and p1,p2. It sees that I can't make my camera work with OpenCV (only the RGB camera), so here goes: How do I obtain the parameters required? And how do I implement them? Btw. I am using LabVIEW for measuring the depth data.

Cheers

Thank you for the help guys. i actually figured it out. It turned out some other functions in the sample program I was modifying was trying to access the camera as well. It only did not output anything. Weird, but it works now. Thanks a lot!

I now changed it, I didn't realize I had it mixed up :/

VideoCapture cap(0);

Mat frame;


    cap >> frame;

printf("Image captured \n");
imshow("test", frame);
waitKey(0);
printf("Image Saved \n");

However, the problem is the same, it only shows a black image. Would you like to see the entire code? I added it here in a .txt file http://www.myupload.dk/showfile/c4l3m0.txt

OpenCV 2.4.9 :)

I am trying to save a color image from a SoftKinetic DS311 camera, using OpenCV.

However, the imagefile that is saved is black.

CvCapture *capture = cvCreateCameraCapture(0);
cvSetCaptureProperty(capture, CV_CAP_PROP_FRAME_WIDTH, 640);
cvSetCaptureProperty(capture, CV_CAP_PROP_FRAME_HEIGHT, 480);


// Get one frame
IplImage* frame;

for (int i = 0; i < 25; i++)
{
    frame = cvQueryFrame(capture);
}
printf("Image captured \n");
cvSaveImage("test.jpg", frame);


printf("Image Saved \n");

This is the part of the code that is supposed to save the .jpeg The program is in a function that is called continously in the main loop, so it should not be because the camera is not initialized?

I did reconsider, however you never know if I would need it again in the future. And also the method of writing a .txt file for LabVIEW interpretation is a crude method - it works, but just barely. I actually have Visual Studio 2013. Do I just download and install or how does it work? Btw, I really appreciate you guys taking time to help me though!

Okay, I decided to try OpenCV, However, do there exist an installing guide, where alle the links to all the additional packages and stuff that is required to work, are updated?

I know, this is a weird place to post, but I figured that since you guys know a lot about computer vision, I thought the answer might be here. Yes, I am new to C++. For some reason I have to use Visual Studio 2008, otherwise I get errors when I build my solution. And it seems that OpenCV runs best on VS2010 and forth (or am I mistaken?) Regarding fileformat: again, I am new, and this was a way to make LabVIEW read the depth data :)

Hello. I know some might find this weird. I have a SoftKinetic DS311 camera, and we managed to get a LabVIEW interface somewhat working. It works by having modified a sample code to writing .txt files containing the depth info, LabVIEW then scans these files as strings.

The sample code is written in C++, and I would like to also access the color image. Usually I would use LabVIEW to access the camera, but I can't when the sample code is running (the USB is probably busy). So I would like to modify the C++ code to save an image while simoultaneously writing depth info. I would prefer to do so without having to install OpenCV, since this seems quite extensive for my needs. Cheers