4
$\begingroup$

I have already tried searching about subject, and the only one thing I could understand is that computer vision involves object recognition, but image processing does not involve object recognition.

What are other differences between image processing & computer vision??

Improve this question
$\endgroup$
1
  • $\begingroup$ Please accept Fat32’s answer. I’ll post additional google search comments to that answer. $\endgroup$
    – Rethunk
    Sep 23, 2019 at 16:07

1 Answer 1

Reset to default
9
$\begingroup$

Digital image processing is an extension of digital signal processing and linear system theory into two dimensional signals.

Image processing involves all low level tasks such as filter design and filtering, spatial scaling, sampling, intensity manipulations, geometry manipulations, Fourier analysis and spectrum analysis, motion estimation, noise reduction, edge detection, image enhancement and restorations etc.

Computer vision involves high levels tasks such as object recognition, scene recognition and image understanding; a more computer science perspective than a DSP type. Note that computer vision inevitably will use image processing at its preprocessing or post processing steps.

Improve this answer
$\endgroup$
5
  • 2
    $\begingroup$ Straightforward and clear answer. I would add that computer vision (or spinoffs) can also work on 3D data, which may or may not be generated in a form recognizable as an image. There are related search terms: machine vision, medical imaging, artificial intelligence (as it relates to vision), and so on. Terms like “machine vision” and “medical imaging” have industry-specific meanings, and may be used in place of “computer vision.” $\endgroup$
    – Rethunk
    Sep 23, 2019 at 16:18
  • $\begingroup$ @Rethunk Thank you for the comment. That's very right. Fields have far away boundaries that's hard to cover in a short answer. $\endgroup$
    – Fat32
    Sep 23, 2019 at 18:44
  • $\begingroup$ I'd say that Image Processing encompasses signals of arbitrary dimensionality (3D images as acquired in tomography or confocal microscopy, 2 or 3 spatial dimensions + time, or + frequency in a hyper-spectral system, etc.). Also, it is much more than an extension of linear system theory -- most of the advances in Image Processing involve highly non-linear algorithms. $\endgroup$
    – Cris Luengo
    Sep 26, 2019 at 21:53
  • $\begingroup$ @CrisLuengo Naming conventions may vary. Not big deal. I assumed that when it's contrasted with computer vision, the implied dimensionality is 2D for images. 3D spatio-temporal data is better termed as motion-picture or video. Same for multidimensional (Volumetric for 3D) data analysis. I consider that it's the linear part of the system theory that really creates the link between DSP and image processing, as non-linear algorithms (which are quite important and abundant as you state) do not permit much mathematical development of image signal processing theory. $\endgroup$
    – Fat32
    Sep 26, 2019 at 22:18
  • $\begingroup$ I'm sure many people have many different meanings for these terms. I routinely work with 3D images (three spatial dimensions), and I'm sure they are images. I also think that it is perfectly possible to apply a linear smoothing filter to a video sequence as a single 3D image, the linear system theory extends trivially to arbitrary dimensions. Such a filtering operation is obviously unrelated to any "high level task" which is your definition of Computer Vision (and which I agree with). -- Regarding mathematical development of theory: look into Mathematical Morphology, it's awesome math! $\endgroup$
    – Cris Luengo
    Sep 26, 2019 at 22:52

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.