8

I wonder if anyone can explain what is the relationship between brightness (or Intensity) and the blurriness of an image..

I want to filter rank the blurred images using intensity as a parameter.

Bill the Lizard
  • 398,270
  • 210
  • 566
  • 880
devan
  • 1,643
  • 8
  • 37
  • 62

3 Answers3

8

Examine the intensity levels in the frequency domain (FFT the image).

If there are high frequencies, image is not blurred. If all intensity is in low frequencies, image is blurred.

Alternatively, calculate Laplacian gradients of image, (for instance, current pixel * 4 - sum of 4 neighbors). If there are high gradients, image is not blurred, if there are none, image is blurred.

enter image description here

Mateen Ulhaq
  • 24,552
  • 19
  • 101
  • 135
Danny Varod
  • 17,324
  • 5
  • 69
  • 111
5

Edit: See Danny Varod's answer.

I don't think there is a relationship. (But believe your books/lecture notes over me.)

Imagine slowly increasing the intensity of the pixels (either straight up RGB or by another color scheme). The image will never get blurry, but it will continously "wash out" until it gets to the maximum intensity (say, 100% white).

Happy homeworking.

  • 1
    Increasing the values, while in the linear range does not cause blurring (and unlike blurring is reversible). If you increase the intensity too much, the values will reach saturation. Saturation will lead to white regions and eventually to a white image. A white image is immune to blurring (Blur(constant) == constant). Still, increasing or decreasing the general (entire image's) intensity will not cause or fix blurring. – Danny Varod Jul 10 '12 at 08:49
1

I probably have no idea what I'm talking about, but the amount of difference in brightness (contrast) can indicate sharpness or blurriness. Blurry images tend to have lower contrast than sharper images. Although, it isn't as simple as just knowing the highest and lowest brightness values, but also how quickly it changes between them. A large change over a small area indicates higher contrast than the same amount of change over a larger area.

Joel Beckham
  • 18,254
  • 3
  • 35
  • 58
  • so the standard deviation of the pixels in the image should give a measure of the contrast, hence the blurriness you mean? – AruniRC May 08 '11 at 12:47
  • @AruniRC - It seems like that could give you a rough idea of the bluriness. Like I said I have no idea what I'm talking about. :) You should probably listen to Danny's answer. – Joel Beckham May 10 '11 at 19:48
  • @AruniRC The STD of the entire image wouldn't help. You need to limit the statistics to space limited regions, to see if there are any sharp changes. The affect of the blur effect is a smooth image. So if the image is not smooth, it is blurred. – Danny Varod Jul 10 '12 at 08:45