5
$\begingroup$

I am reading image processing book.while reading i am getting terms like image detection,image recognition and image segmentation. it seems that all above terms are one and the same.i am totally confused amongst them.so can anybody help me to understand or distinguish between the above?

$\endgroup$
7
$\begingroup$

It can be a little confusing at times, and the terms are not completely independent.

Detection:

In detection, you are simply detection the presence of something. For example, you might design an algorithm to detect beach-balls from pictures of a beach. You would feed an image into your algorithm, and it will spit out an answer 'yes', if there was a beach ball, and 'no' if there was no beach ball. This is very simply what it means to 'detect' something.

Segmentation:

Segmentation is a method - a technique - an algorithm. So for example, if someone says "How did you design your beach-ball detection algorithm?", you might say "I used a segmentation algorithm". Or you might say, "I used a template-matching algorithm". Or you might say "I used a scale and shift-invariant matching score algorithm", etc. Those are all methods of doing something, and segmentation is just a method.

In particular, segmentation simply means a thresholding of your image into various areas. In a simple case, say you want to separate background from foreground. Then, you segment the image into two regions: A region that corresponds to the background, and a region that corresponds to the foreground.

Image Recognition:

Lastly, there is image recognition, which is simply, a very broad goal. For example, you might say, "I achieved image recognition for beach scenes, by detecting beach balls using the method of segmentation, detecting water using the method of k-means, and detecting bikinis using the method of template matching.


Of course, those are just examples, but broadly speaking, this is how you should parse the different words and terminologies you come across in this field. Hope that helped!

$\endgroup$
2
$\begingroup$

If I could add something to @user4619's great answer (not yet allowed to comment so... ;) ), I would say that object detection is also often about providing the location of the object. While I think I have come through the expression "object localization" at least once in the literature, I have seen more often "object detection" for locating some object in a given image. Some methods are able to provide only the presence/absence of the object, others are able to give the coordinates of a "centroid-like" feature of the object and some methods provide even the boundaries of the object.

Though, what was said before is still true and a pretty relevant explanation (maybe I could suggest the OP to mark the answer as correct if it is satisfying to him/her?)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.