1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Gradient problem

  1. Jun 17, 2006 #1
    I was wondering if anyone could help me with this rather arbitrary math problem.

    I need a description of what a gradient is in mostly non-math terms. I know a gradient is a sort of slope, but I don't understand where you would draw the line of what is and what is not a gradient. Meaning which slopes are and are not a gradient. Is this even calculable? I know http://en.wikipedia.org/wiki/Gradient" [Broken] on it has some equations but I don't understand any of that.

    I am using this for an image processing project where the maximum pixel intensity value is 765 and minimum is 0. So for example is a sequence of pixels with the values 10,15,20,25 a gradient? I'd think so, but is 9,16,21,24 also a gradient? Where do you draw the line between gradient and non-gradient?

    Thanks a lot :)
     
    Last edited by a moderator: May 2, 2017
  2. jcsd
  3. Jun 17, 2006 #2
    A gradient, in the mathematical sense, is a measure of how much a surface is changing. If you could imagine a small stretch of wavy ocean, you could plug in the x and y of any point on that section of ocean and you'd "get out" an arrow (a vector). That arrow will point in the steepest direction -- for example, for a climber standing at the foot of Everest, the gradient of the land at the climbers position is a vector that points UP at Everest. because from where he's standing, that's the steepest way he could go. It's hard to describe, but it's a vector Calculus topic. Loosely, though, a gradient is how something is changing from one place to another, over space. That means a lot of things can have gradients -- height, color, temperature, anything that can change from place to place.

    In your sense, a gradient is used in the color sense -- one color fading to another. It's arbitrary, and I don't see any reason why you should need to "define" a gradient. Indeed, if I interpret your numbers right, both of your examples are gradients, because the color changes over space. The former gradient might look smoother though.
     
    Last edited: Jun 17, 2006
  4. Jun 17, 2006 #3
    Perhaps try

    http://amath.colorado.edu/outreach/demos/hshi/2001Spr/snake/snake.html [Broken]
    "Visualizing calculus: The use of the gradient in image processing"

    A gradient is not a sequence of numbers, or a path. It is a vector at a single point, that describes the how the scalar function slopes there - in which direction it slopes, and how steeply it slopes. That's what a vector describes - a direction and a magnitude. This is of course multivariable calculus, and it will be of no use to you if you avoid the math. What is your background in calculus?
     
    Last edited by a moderator: May 2, 2017
  5. Jun 17, 2006 #4
    Ok, that clears it up. It didn't make sense that gradients could not be arbitrarily defined, although it would make things a bit simpler :|

    Thanks a lot :-)
     
  6. Jun 17, 2006 #5
    None at all. However I don't think I need it. I am simply using gradients as one method of defining an edge within an image. The problem was figuring out where to draw the line of a single gradient and an edge. Since a gradient can be completely arbitrary, it looks like that cant be the only clue in finding edges.

    Thanks though, the link is helpful
     
    Last edited by a moderator: May 2, 2017
  7. Jun 17, 2006 #6
    Well no. A mathematical gradient is different than the very specific type of gradient that you're talking about. In an image, a gradient is linear, so you could determine the beginning and end of it. It's an adopted use of the word "gradient." But since you were asking on a Math forum (and not a Photoshop forum), I assumed you wanted a definition of a gradient in general, and every image has a mathematical gradient.

    One warning: watch out for round-off errors. That could be a big pain.
     
    Last edited: Jun 17, 2006
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook




Similar Discussions: Gradient problem
  1. What are gradients? (Replies: 3)

  2. Stochastic gradient (Replies: 1)

  3. Gradient conversions (Replies: 3)

Loading...