Content-Type: text/plain;
charset="iso-8859-1"
X-MIME-Autoconverted: from 8bit to quoted-printable by
pimout7-int.prodigy.net id LAA73280
Jim,
I've done a lot of work with CCD cameras and, in my experience, the lenses
are almost always the determining elements in the size of the circle of
confusion. This is especially true as the CCDs get smaller and the lenses
get cheaper. I have generally found that even with low-cost CCDs, the CCD
makes a good lens tester. This argues for the applicability of the formulas
as they were developed for film.
If the lens is better than the CCD, another problem arises - the image
contains a lot of aliasing. In broadcast, the MTF of the lens is purposely
reduced to avoid this phenomenon. I remember a particularly striking
example of this problem where brushed aluminum was being imaged. The image
essentially disappeared behind intense moir� patterns. In these cases,
depth of focus seems not to have any meaning - at best focus, nothing useful
can be seen. Only if the lens is defocused enough to remove all image
energy above the CCD cutoff frequency does anything useful show up and this
information is then limited, of course, to spatial frequencies lower than
that of the CCD array.
Sampling theory says that about 2.8 samples per cycle are needed to
reproduce both the frequency and phase of a signal (remember the Kell
factor?) so this argues that for practical use, the circle of confusion
should be 1.4 times the pixel pitch. Certainly situations can be contrived
where subpixel measurements can be made but these always require a clean,
isolated edge to image. In such cases, the depth of field will be
determined by the s/n of the signal rather than by optical considerations
alone.
I guess what I'm trying to convey is that using the pixel pitch alone for
imaging calculations almost always obscures the real limits to image
quality.
Regards, Dave Gilblom
+ + + + + + + + + + + + + + + + + + +
+ David Gilblom, President +
+ Pacific Photonics, Inc. +
+ P.O. Box 3039 +
+ Los Altos, CA 94024-3039 +
+ Phone: 650-428-0409 +
+ Fax: 650-240-4005 +
+ E-mail: [EMAIL PROTECTED] +
+ + + + + + + + + + + + + + + + + + +
> -----Original Message-----
> From: by way of Jeanne Anderson <[EMAIL PROTECTED]>
> [mailto:[EMAIL PROTECTED]]
> Sent: Monday, October 04, 1999 7:07 AM
> To: [EMAIL PROTECTED]
> Subject: Depth of Field calculation for area arrays ...
>
>
>
> Dear Colleagues in the Electronic Imaging Technical Group:
>
> I have been trying to calculate depth of field for certain
> close-up situations in my own work. I am quite familiar
> with the standard formulae for determining the bounds of
> the depth of field ...
>
> Near Limit :
>
> Dn = 1/(1 + Dcf/F**2)
>
> Far Limit :
>
> Dn = 1/(1 - Dcf/F**2)
>
> where
>
> D = focused distance, c = limiting circle of confusion,
> f = f-number of the lens apertue, F = focal length of the lens.
>
> My question relates to the "appropriate" value to use for
> the circle of confusion. Typically, for film, a value of 1/1000
> of the focal length of the lens is used (Focal Encycopedia of
> Photography). Alternatively, an absolute value of 1/30 mm is
> recommended (Kingslake, R. "Optical System Design"). However,
> these figures are based on the limits of human vision when
> examining FILM. (The basis for these values is ... "How big
> can a point-image become before it is no longer perceived as
> a point?")
>
> Clearly, video systems and CCD sensors introduce another element
> to this. I've made a brief examination of the literature, and
> the best discussion I can find is a technical note provided by EG&G
> Reticon in their camera handbook (Application Note #127). In this
> paper, the authors use the inter-pixel spacing as the "appropriate"
> value for the circle of confusion ("acceptable blur spot"). Can
> anyone on this list justify/dispute this for me? My intuition tells
> me this is probably a good "starting point", but my intuition and
> reality are often at odds with each other. The EG&G note simply
> states what is done, without justification. It does not attribute
> the note to a particular individual.
>
> Any thoughts, feedback, references, etc. would be helpful.
>
> Regards, and thanks,
>
> Jim WALTON
> Chairman, SPIE Technical Group on High-Speed Photography,
> Videography and Photonics
>
> *************************************************************
> *************************************************************
> * *
> * 4D VIDEO CELEBRATED ITS 11TH ANNIVERSARY ON OCTOBER 1 '99 *
> * *
> *************************************************************
> *************************************************************
> * * *
> * JAMES S. WALTON, Ph.D. * INTERNET: [EMAIL PROTECTED] *
> * President, 4D VIDEO * *
> * 825 Gravenstein Hwy North * PHONE: (707) 829-8883 *
> * Suite #4 * *
> * SEBASTOPOL, CA 95472 * FAX: (707) 829-3527 *
> * * *
> *************************************************************
> *************************************************************
>
