Is Image Deblurring Possible with Sequential CCD Scanning?

  • Thread starter Thread starter Andre
  • Start date Start date
  • Tags Tags
    Image
Click For Summary

Discussion Overview

The discussion centers around the possibility of image deblurring using sequential CCD scanning technology, particularly in the context of motion blur removal. Participants explore the effectiveness of various algorithms and the underlying mechanics of digital cameras, including CCD and CMOS sensors.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested

Main Points Raised

  • Some participants express skepticism about the effectiveness of deblurring for out-of-focus images, suggesting it may only work for motion blur due to less information loss.
  • There is a discussion about the innovation of calculating the trajectory of the CCD without prior knowledge, with some participants questioning whether this involves exhaustive search algorithms.
  • One participant notes that the algorithm appears to plot motion, implying that the camera captures images in a sequential manner rather than all at once.
  • Another participant mentions that each pixel integrates photon strikes while the shutter is open, contributing to the understanding of how motion blur can be analyzed.
  • Some participants reference standard techniques for removing blur caused by uniform linear motion, indicating that this is a recognized area in image processing.
  • There is mention of the differences between CCD and CMOS sensors, particularly regarding how they handle exposure and the implications for motion blur.
  • One participant emphasizes that removing motion blur is feasible, contrasting it with defocus, which they claim cannot be removed.

Areas of Agreement / Disagreement

Participants do not reach a consensus on the effectiveness of the proposed deblurring techniques, with multiple competing views on the capabilities of different camera technologies and algorithms remaining unresolved.

Contextual Notes

Participants highlight limitations in understanding the specific algorithms used for deblurring and the operational differences between CCD and CMOS sensors, which may affect the outcomes discussed.

Computer science news on Phys.org


Wow!
 


Question is when will it be out? I bought Photoshop for Spawn eonas ago, cost me a fortune. I'll bet she lost it, it was cool even back then.
 


Evo said:
Question is when will it be out? I bought Photoshop for Spawn eonas ago, cost me a fortune. I'll bet she lost it, it was cool even back then.
Every new release costs a fortune, and if you pay for upgrades, you'll still pay the same over time. Staying on the bleeding edge is quite expensive.
 


Right, here are the current prices. Some elementary photo editing can also be done with freeware like Gimp
 


Andre said:
Is it true?

i think so. it won't work for something out of focus, but with a motion blur, there is not so much loss of information. when i took an image processing course back in the mid-90s, we went over blur algorithms. the thing i don't remember from back then is this innovation they've added, which is to calculate the trajectory of the CCD without knowing it prior (say from a flyover, where it would be essentially a straight line). guessing the trajectory is pretty impressive, assuming that is what is being done. hopefully it's not an exhaustive search algorithm that simply computes the trajectory when it's done (like maybe using a sharpness-measuring algorithm to find an endpoint).
 


Funny, just about an hour ago Junior sent me this link: http://vimeo.com/28962540

Not from Adobe, but impressive picture related technology.
 
I noticed the algortihm plots motion. This would seem to imply that the camera involved didn't take the picture all at once, but scanned the ccd's to capture the image in a sequential but moderaly fast sequence, the equivalent of a moving shutter. Do all digital cameras work this way (effective moving shutter) or do some of them sample or at least capture light on all the ccd's at once?
 
I am not sure what you mean, but you can read the motion from the picture taken all at once:

marcin_curves.jpg


This was shot by manually moving camera with the open shutter.
 
  • #10
rcgldr said:
I noticed the algortihm plots motion. This would seem to imply that the camera involved didn't take the picture all at once, but scanned the ccd's to capture the image in a sequential but moderaly fast sequence, the equivalent of a moving shutter. Do all digital cameras work this way (effective moving shutter) or do some of them sample or at least capture light on all the ccd's at once?

not sure about the sequential part (don't think so), but each pixel is integrating photon strikes while the shutter is open.

http://www.google.com/search?q=remo...s=org.mozilla:en-US:official&client=firefox-a

"removal of blur caused by uniform linear motion" is a standard technique in Gonzalez/Woods'
 
Last edited by a moderator:
  • #11
None of this is "magic". Here's one way to do the math - at least, there are enough clues to figure out how to do it, if you know some general image processing theory:
http://www.mathworks.co.uk/help/toolbox/images/bqqhld4.html
 
Last edited by a moderator:
  • #13
rcgldr said:
This would seem to imply that the camera involved didn't take the picture all at once, but scanned the ccd's to capture the image in a sequential but moderaly fast sequence, the equivalent of a moving shutter. Do all digital cameras work this way (effective moving shutter) or do some of them sample or at least capture light on all the ccd's at once?

"Cheap" cameras use CMOS sensors where the electronics creates a "rolling shutter" scan, which is cheaper to implement. See http://en.wikipedia.org/wiki/Rolling_shutter for extreme examples of the distortion it can create.

It is possible to build a more complex CMOS sensor that scans multiple rows of pixels in parallel, and CCDs intrinsically work in parallel. CCDs need a physical shutter (as in a DSLR) to control the exposure, but for CMOS sensors the "exposure time" for each pixel is controlled electronically.
 

Similar threads

  • · Replies 2 ·
Replies
2
Views
2K
Replies
16
Views
3K
  • · Replies 2 ·
Replies
2
Views
7K
Replies
2
Views
2K
Replies
10
Views
7K
Replies
10
Views
5K
  • · Replies 14 ·
Replies
14
Views
8K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 31 ·
2
Replies
31
Views
6K
  • · Replies 4 ·
Replies
4
Views
2K