Looks like I am answering my question myself! Here is a solution I have found and I feel it is good. Please do take a look and give your suggestions and also feel free to make changes in the file and do tell us about it to make it even faster.
How to triangulate multiple objects in stereo images?
2 visualizzazioni (ultimi 30 giorni)
Mostra commenti meno recenti
Stalin Sabu Thomas
il 16 Ott 2018
Risposto: Stalin Sabu Thomas
il 24 Ott 2018
I used the example in https://in.mathworks.com/help/vision/ref/triangulate.html and it works. My question is, if I detect multiple objects (here multiple faces) in a stereo pair (left and right camera image), how do I triangulate all the detected objects?
In the given example image, there are 3 faces. If I detect the three faces in both the stereo images, how do I triangulate each of them? Because if I blindly just give the points to triangulate it gives wrong results as its triangulating the wrong pixels or one object might not be detected in another image. I think I have to find SURFRegions or Harris or something to match features but how do I do it with just the points I detected? I am just a beginner in matlab and especially in stereo images. This is also my first question in MATLAB Answers. Please do help me. I am using Matlab R2018a in Windows 10.
The image is an example of what I am trying to explain. In short the face(s) found in both left and right image must match and only those respective faces must be triangulated to find its distance/depth.
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/198034/image.png)
--EDIT--
I have been searching for an answer for a while. I tried to use ssim, psnr, immse, nromxcorr2, etc. to all the bboxes but its not easy as the all the bboxes images have different dimension and not to mention how heavy the computation is! Please do help
0 Commenti
Risposta accettata
Più risposte (0)
Vedere anche
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!