TimeManx's profile - activity

@JAyThaRevo The camera does move. The images could be taken at a farther or nearer distance like the images that I've posted.

I'm trying to find if a new object has been added to a scene. The initial scene would most likely be lacking any kind of objects but might contain some patterns. The image for the scene containing the new object could also be taken from farther distance.

Would feature detection be a good option to achieve this? Or maybe something else? I'm basically looking for pointers to what I should start reading about.

This is how the two scenes might be

You're post looks all mumbo-jumbo to me (only because I'm a computer vision noob, you've explained it very nicely). I don't even know what Hough voting is. What I'm doing right now is matching the FAST features of the two frames using the FLANNBASED descriptor matcher and discarding feature movements which are below a certain threshold. But I don't get very good results with this. Are there any improvements/flaws that you see here? Or should I dig deeper into your post?

I need to stabilize the shakes that a moving camera gets, for example, when a person is walking with the camera in his hand.

Till now, I've been able to detect the features of both frames using the FAST feature detector and calculate the displacement of the points.

What should I do next? Other posts suggest calculating the homography matrix and then using it to warp the new frame but will that not distort the linear motion of the camera as well? I just need to remove the shakes.

@Marde Yes. I'm using 2.4.6 and it's still slow and blurry. Any way to fix this? I've had to write a lot of extra code just to get the Mat of the current frame.

@AndreyPavlenko Is there a way to create a Mat backed by a NIO ByteBuffer allocated by us. Alternatively, is there a way to get the Buffer of a Mat ?

@Quattro Were you able to find a way?

Also I'm a little confused about SIFT/SURF/ORB. In another post, someone suggested that feature detection techniques are more useful when we need to detect some specific images against a large database and that they aren't really useful in these kinds of scenarios.

You mean background subtraction followed by blob detection?

You're probably not running it in fullscreen Add this to your onCreate

getWindow().setFlags(WindowManager.LayoutParams.FLAG_FULLSCREEN,
            WindowManager.LayoutParams.FLAG_FULLSCREEN);

I've implemented the latentSVM detector in my android application acoording to how it was done in the latentsvm_multidetect.cpp sample but I just get a black screen with the log saying "Preview frame received. Frame size:152064" Any ideas?

Don't bother. Found it here.

@GilLevi I read in a post that for latentSVM detector, openCV provides classifiers for people, cars, etc. But I could only find a classifier for cats in the OpenCV folders.

Where can I find the classifiers for people, cars, etc. I could only find for cats in the OpenCV folders.

I started out with detecting people using the samples. HOG has good accuracy but a very slow for use on an android phone. Haar cascade was faster than HOG but accuracy was quite low. Also, I remember reading that Haar isn't suitable for human detection. For vehicle detection, Haar is working fine.

So could you guys suggest what would be best for detection of these objects

• People
• Vehicles
• Trees
• Buildings