Object tracking of known objects.

edited August 2015 in How To...

Let me first explain the problem statement as clearly as possible.

I have a series of images with slight shift in camera angle (of say a room) (360 degree view in a series of images) I will select a particular area of the first image and then I need to find the same sub image in all the rest of the images and mark those areas. So this is not basically object detection but I need to track a pre defined image.

What I have in mind is select the pixels using get() and then run a script to find the same array in all the rest of the images.

I haven't tried this out yet.

I am just looking for some elegant solutions using processing if possible before beginning.

Help me out if there are some pre existing libaries which can ease this task.

EDIT

Attaching a few images to give a better understanding of my problem. There are around 80-90 such images. and there will be say horn in 25 of those. So I will select the horn area in the first image and I need to write a code that will mark the horn in all the rest of the images which have the horn in them. I hope I have made my issue a bit more clear.

0 1 2 3 4 5

Answers

  • Can you show some of the images. It will help.

  • are you going to make a panorama?

    are the overlapping areas all in the same height?

    Meaning rotating the cam while the height (look at y value / angle vertically) stays the same?

  • What I have in mind is select the pixels using get() and then run a script to find the same array in all the rest of the images.

    From the description you give I don't see how this can work: if each image is taken from a different angle the pixels won't match exactly and such an array comparison simply won't work. I would think that one of the computer vision libraries will serve you better: this sounds equivalent to OpenCV's MarkerDetection demo; though with a more complex 'marker'.

Sign In or Register to comment.