| Question about forensics and/or archaeology
Posted: 7/8/2009 6:21:04 PM | I’ve noticed in the course of other discussion threads that there are several people here who have a background in forensics and/or archaeology. I have a question that I would like those with some experience in these areas to consider.
The background: I have a number of small artefacts that I need to photograph for an ongoing research project. Previously, I had been using a film camera - an Olympus OM-4 T - with an auto bellows and a collection of very nice bellows mount macro (flat field) lenses.
Last year, I decided to go digital; and despairing of Olympus ever releasing a full frame digital camera, I decided to switch to Nikon and invested in a D700. I really missed my Zuiko bellows mount macro lenses, though; so this spring I built an Olympus-to-Nikon adaptor to allow me to use those Olympus Zuiko macro lenses and auto bellows on the Nikon D700.
(There are no commercial adaptors available, since the distances from the lens mounts to film/sensor planes make Olympus lenses and Nikon bodies incompatible - in that Olympus lenses would never focus at infinity on Nikon bodies… which isn’t an issue with bellows mount macro lenses since they can never focus at infinity on any camera anyway).
All of which is working really well; but of course the depth of focus is invariably very shallow with such a set-up. I can increase the depth of focus by stopping down the lenses I am using; but then the images begin to soften due to the diffraction effect of the small apertures thus being used.
SO: The Question (finally).
I am using a deconvolution algorithm which makes use of a Fast Fourier Transform operation, but I am having to estimate the blur radius by comparing an image shot at an optimal aperture (say, f8) with a diffraction-softened one shot at the smallest aperture available to me (say, f32).
This works quite well and does a nice job of removing the diffraction blur, leaving me with a very sharp image that has the maximum depth of focus I can achieve; but I am wondering, does anyone know a more accurate method for determining the blur radius due to diffraction than zooming in to 1200 percent and counting the difference in pixel spreads for specular highlights? | |
|
| Question about forensics and/or archaeology
Posted: 7/8/2009 11:36:01 PM | | I don't think I can answer the question, but is diffraction blur the same as spherical or chromatic abberation ? My field is optics but from adifferent focal point. Forgive the pun. I dispense eye glasses. | |
|
| Question about forensics and/or archaeology
Posted: 7/9/2009 1:38:11 AM | Good question. No, chromatic aberration is due to the fact that different wavelengths of light are focused at slightly different points by the same lens. This problem is more or less corrected in modern camera lenses by specialized coatings and by special glass elements (low dispersion glass).
Spherical aberration is a different phenomenon altogether, whereby the curvature of the lens surface (necessary in order to gather light) causes the center of the lens to focus light at a different point than the edges of the lens. This is not a significant factor in most instances, since the distance from the lens to the subject is many orders of magnitude greater than the curvature of the lens; so a curvature of a few centimeters will have no noticeable effect on a subject which is tens or hundreds of meters away. When the subject is much, much closer - say, around ten centimeters - and the depth of focus drops to a few millimeters, then the curvature of the lens will be greater than the depth of focus; and then it will be impossible to focus both the center of the subject and the edges of the subject at the same time.
This is corrected by the specialized optics of macro lenses, such as the bellows mount macro lenses that I am using. This is why they are called "flat field" lenses - the center and edges of the lenses will focus on the same plane.
Diffraction blur is caused by closing the aperture down, usually to: 1) use the center of the lens, which is optically superior to the edges (the best setting is usually around f8) and; to 2) increase the depth of focus the lens produces, since smaller apertures produce greater depth of focus.
The aperture setting of a lens is a ratio between the lens length and the diameter of the lens opening. As the aperture gets smaller, this ratio increases: hence, the "f" number gets bigger. There is always a certain amount of bending to the light passing through the aperture; this is diffraction, and it occurs at the edge of the opening through the lens. This can be considered a constant; but as the aperture grows smaller, the percentage of light passing straight through the lens decreases in relation to the amount of light that is bending at the edge of the opening. This means that at the smallest opening of a lens aperture, a relatively large amount of the light reaching the focal plane is made up of light which has bent slightly at the edge of the lens aperture, which results in a 'softening' of the image or a loss in detail (resolution).
This can be corrected through a deconvolution algorithm, which runs through a Fast Fourier Transfer. This converts the pixel values of the image into frequencies which can be mathematically manipulated; the Fast Fourier Transfer is then reversed to reconstruct a standard pixelated image.
One needs to establish the amount of diffraction to be corrected, in terms of the blur that a point of light undergoes in passing through the lens; and that is what my question addresses. I can compare the results obtained with an optimal lens setting to that achieved by the smallest aperture available; but then I have to estimate the amount of blur that has occurred.
I'm just wondering if anyone knows of a better, more consistently quantifiable approach than the process of estimation I am using - since the fluid nature of the bellows extension, the variety of bellows mount macro lenses I am using, the variation in subject size and the wide variation in the actual size of the specular highlights I am estimating diffraction blur from make a formulaic approach to establishing the diffraction blur a little too involved for practicality's sake. | |
|
| Question about forensics and/or archaeology
Posted: 7/9/2009 7:44:47 AM | When I was making eyeglasses we had a guy come in to have a couple lenses made for him. And some coloured filters. Some eyeglass labs can do this work and will do it...for a fee of course.
We could make the lenses...but he had to mount them himself.
Maybe should have asked about cameras and such rather than archealogy...might have gotten more responses.
| |
|
| Question about forensics and/or archaeology
Posted: 7/9/2009 7:59:43 AM | Well, it is a bit of a specialized application; most people who use cameras would not be familiar with it, whereas many people who are involved with forensics and archaeology might be.
I have been thinking that I should have added "astronomy" in the thread header, since this process is often used to remove various forms of blurring from images of stars (which, as perfect points of light, are easy to define the blur radius for). | |
|
|
Home
login 
