Speckle noise reduction

Holography related topics.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

As my DCG holograms improve, the various defects and noise sources have become easier to isolate -- but not necessarily easier to solve ;)

Dinesh brought up the subject of a new technique to reduce or eliminate speckle in digital holography, and I was wondering if speckle can be reduced in traditional "analog" DCG. In practically all of the reflection holograms I've ever seen or done, there is some very fine (hardly noticeable unless one looks at the film very closely) graininess, which looks like very tiny black holes.

I assume this is speckle. I am starting to read up on the subject, and trying to reduce it (if possible) is a new subject to me.

The 3 following approaches come to mind:

1) Modify the object to make it less apparent. For example, I've seen that speckle is especially noticeable in larger areas of solid, uninterrupted, light colors (for example white) on the object. On the other hand, metallic objects may not show speckle (maybe it's still there, but not noticeable for some reason).

2) Modify the method of holographic recording (?)

3) Modify or change the reconstructing light (?)

If anyone has thoughts on this (or can suggest any papers), I would appreciate it.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

I tried to take some photos of test-strip DCG holograms done last weekend, to show the speckle. It's difficult to photograph, but it seems to be noticeable in the marble with the white swirl in the middle. Different types of objects (different colors/surfaces) seem to have a big influence on how apparent the speckle is.
Attachments
P5010054.JPG
P5010054.JPG (40.62 KiB) Viewed 19627 times
P5010051.JPG
P5010051.JPG (38.25 KiB) Viewed 19627 times
Dinesh

Speckle noise reduction

Post by Dinesh »

I've actually been thinking about this for a while now. I think that speckle does reduce the "pictorial smoothness" of an image. It's hard to state this in a clear way, but the goal of holography has always been to create a three dimensional image that appears to be "really there". Yet, a holographic image is clearly not "really there", because of mismatches in hue and tonal value and speckle.

At any rate, let's start by examining the cause of speckle. It occurs because any surface that appears to be smooth actually has variations in smoothness at a scale that's too small for the eye to detect; perhaps a "smooth" surface may have an rms roughness of a few microns. Imagine that you're a bacterium crawling across a surface that seems smooth to the human eye, then you'll see a terrain with hills, cliffs etc. When light scatters off such a surface, there are "random" phase variations caused by the random roughness and the orientation of the roughness. These random phase variations cause random "interference dots" which record on the medium.

So, how to overcome this? Well, three ways occurred to me. One way is, of course, to polish the shinola out of the surface. In other words, use maybe cerium oxide (like the amateur astronomers who make their own mirrors use), or something similar. This may be done if the surface is relatively flat to begin with, but it's not practical for a surface with lots of contours. Another way is to pinhole the object. If a large lens is placed in front of the object, ie between object and plate, and a small pinhole were placed at the focal point of the lens, then the "noise" (speckle) may be filtered out. The diverging beam after the pinhole should be clean. The disadvantage is the the object will then be "backwards" and, depending on the lens used, will be magnified (or minified). The advantage is, of course, that this is a one-step setup - you can place the H2 right inside the image field. Still another method - much more speculative - is to create a hologram whose phase profile is the exact opposite of the random phase profile of the speckle; a sort of "Holographic Schmidt Camera" for speckle correction. This latter solution is probably a PhD project for someone!
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

Thank you Dinesh, that is very helpful. The "microscopic roughness" causing speckle has been confirmed in my tests, but I didn't realize it until you described this effect. I have a particular kind of paint which produces a beautiful, matte, velvety surface, yet it also produces a lot of speckle in a hologram. I believe the nice flat, velvet-like effect in this paint (Pelikan Plaka from Germany) comes from the microscopic roughness of the surface.
kaveh1000
Posts: 56
Joined: Wed Jan 14, 2015 5:04 pm

Speckle noise reduction

Post by kaveh1000 »

From what I recall this is a fundamental problem that has no solution. The speckle will be there whatever you do, as it is caused by the random interference of laser light, and a consequence of the high degree of coherence.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

Thanks Kaveh. It does seem like a fundamental problem. But now that I'm taking a close look at a variety of surfaces in holograms that I've done, I'm seeing a lot of variation in the perceived speckle. For example, on a shiny, silver object, I cannot detect any speckle. On a pure flat white surface, the speckle is very noticeable. The darker a surface gets, the less noticeable it is (probably because speckle is like "poking tiny black holes" in the surface of an object -- as seen through the film -- and the darker the object, the less apparent the black holes).

Maybe other forum members could check this on their holograms, comparing the apparent speckle on different kinds of objects, e.g., shiny metallic, dull metallic, paints of different colors, glossy colors, matte colors, etc.
Ed Wesly
Posts: 513
Joined: Wed Jan 07, 2015 2:16 pm

Speckle noise reduction

Post by Ed Wesly »

Speckle is in the eye of the beholder.

Or the camera. It is well-known that when photographing laser-lit holograms there is a trade off between aperture openings (f/stops) for adequate depth of field without too much speckle. The smaller the opening, the better the depth of field, but the speckle gets bigger.

It is hard to say from your photos is whether the dots are coherent speckle or coating issues. It is difficult to generate speckle with white light but I have seen it off the dashboard of my old VW GTI, with the sun at grazing incidence, off its textured dashboard surface. There was a great Scientific American Amateur Scientist article on the phenomena.

To check if the artifacts in your hologram are truly speckle, you could use your eyes or a single lens reflex camera, film or digital, as you don’t necessarily need to take pictures, just to observe what happens to the “speckle” as you view through the lens as it is stopped down. Hopefully your SLR has a depth of field preview button.

You could also stop down your eyes by making a sphincter with your thumb and forefinger to act as an iris in front of your eye. Practice looking at a laser lit object to see the effect, then apply it to the hologram. If the pattern enlarges, then it’s speckle.

Another test for speckle is observing which way it moves when the observer moves. Without correcting my myopia, (glasses off) speckle moves contrary to my direction. With the cheaters on, it moves slowly with me. Moving the SLR with the lens in and out of focus will also show the same effects.

So take a look at the holograms with a different eye and see what you shall see! And you laser lit surfaces!
"We're the flowers in the dustbin" Sex Pistols
Dinesh

Speckle noise reduction

Post by Dinesh »

It's not completely random. It appears random because of the random nature of the surface roughness. Thus, reflection from a pool of liquid not subject to no vibration does not show speckle. Similarly, a pool of mercury will show no speckle. The reason that Joe does not see any speckle on a metallic surface is also because metallic surfaces tend to be smoother, since any manufactured metallic surface was cooled from the liquid state, at least today. I suspect a horses shoe may not be so speckle-free, since it's hammered. The pure white surface was probably painted, so would have considerable surface roughness, but at a low scale of about 2 to 3 lambda. Most good mirrors are polished to an rms of roughly lambda/5 to lambda/10, which is too low to cause speckle; the rms roughness should be at least lambda.

The problem can be solved with equipment available today. At the hobbyist level, just polishing the surface can reduce speckle quite a bit, assuming a relative flat surface. Shadowgram techniques may also mitigate speckle, since there's no reflection from any surface. At a more advanced level, it is possible to scan a surface with an AFM, map the surface profile down to the sub-nanometre scale, and so create a phase mask. There also exist new digital techniques, such as the one I mentioned on the "off topic" page by the Japanese team. New optical computing techniques may also be used, such as, for instance, superimposing an optical neural network on the surface.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

Thank you Ed, I will be giving that some thought. Maybe it will help to describe how I currently "see" speckle. If I take almost any clean reflection hologram (good in every respect) and focus on it as closely as possible, and ask myself "why doesn't this look exactly like a real object?", the first objection that usually comes to mind is a very fine, almost (but not quite) imperceptible graininess. Depending on the hologram (and now I'm thinking it has to do with the surface of the object more than anything else) this graininess may be totally absent (I have never in my life seen this over the entire image of any reflection hologram, but I have seen it in certain sections of holograms), or it is present in varying levels, and sometimes it gets obtrusive.

Just to throw something speculative out there, I wonder, if the input beam is modulated during exposure, would it have any effect on the speckle distribution? By modulation, I was thinking of slightly changing or wiggling the angle of the beam going into the spatial filter (in a Denisyuk setup). If the speckle pattern could be changed many times during the recording, maybe it would tend to cancel out?

Thanks for you comments, Dinesh, I will be mulling over them.
Ed Wesly
Posts: 513
Joined: Wed Jan 07, 2015 2:16 pm

Speckle noise reduction

Post by Ed Wesly »

I have done speckle averaging when photographing holograms in two ways. By replaying a laser transmission hologram with a beam diverged or more properly scattered by a piece of ground glass. Of course if you do this, you will see a very coarse speckle pattern, but if the ground glass is moving, like spinning on an electric motor shaft, the speckle will average out.

Another trick was utilized in photographing some bubble chamber holograms when I worked at Fermilab. The particle tracks were pretty small, and could get obscured by speckle. I had a 500 millimeter focal length Reflex-Nikkor lens on my trusty Nikon F, and used a rotating off-center iris on it.

This might be a bit hard to comprehend unless you know what these lenses look like. The Reflex-Nikkors are catadioptric designs, combinations of lenses and mirrors. Attached is a slice of one of my classroom handouts, although the diagram is courtesy of the Nikon F-Nikkormat Handbook of Photography.
ReflexJoeF.jpg
I cut an off-center hole in a circle of cardboard, similar to the second illustration, and jigged it so that the off center circle swept through the doughnut shaped clear aperture of the front of the lens during the photographic exposure. This averaged out the speckle by having a variety of apertures to look through.

Back to your original question, about moving something during exposure to cancel out speckle, you would probably average out the holographic interference fringes also, like the photographic case in the first paragraph, yielding a low or no-contrast holographic pattern.
"We're the flowers in the dustbin" Sex Pistols
Post Reply