What is a good way to recognize an object from two different images in android system?

asked 2018-12-29 12:52:56 -0500

ONIGIRI gravatar image

updated 2018-12-29 12:56:07 -0500

I am trying compare two images containing a same object and show a result of similarity between them. At first I tried histogram comparison and matchTemplate() function using "TM_CCOEFF_NORMED". But that did not seem to be a good idea and the result was not satisfying. My second approach was to use ORB to extract features and find key points. But then I came to know that, ORB feature matching is to find the position of a known object. My app wasn't going anywhere with feature extraction since it keeps detecting wrong features and match wrong key points.

Sorry for making the question messy and confusing. I am kind of new to all this. It'd be really helpful if I could get some suggestions on what actually to use to detect an object using a photo of that object, from another image.

edit retag flag offensive close merge delete

Comments

It indeed is true that I found those ideas on the internet. I did my research on them and saw feature extraction to be a common way of applying image processing to recognize objects. I may have done my research wrong but i do need some idea here to move on to do some progress. I don't want to use deep learning methods here and got a very little idea of everything.

ONIGIRI gravatar imageONIGIRI ( 2018-12-29 13:19:48 -0500 )edit

apologies for being somewhat harsh on you before.

imho, what you want is not doable, simply. maybe you need to change the idea, so it's "less broad" ?

(e.g. restrict it to certain subset of objects)

berak gravatar imageberak ( 2018-12-30 09:37:44 -0500 )edit

Sorry for being so late to reply.. Thanks for the suggestion really. I am trying to make it less broad by adding some android features like cropping the image so that the object is focused on more before comparing the images.

ONIGIRI gravatar imageONIGIRI ( 2019-01-12 08:34:30 -0500 )edit