Speckle noise reduction

Holography related topics.
Martin

Speckle noise reduction

Post by Martin »

Dinesh wrote:Speckle is largely a reflective effect; speckle in transmissive media are caused by non-uniform index variations of the order of lambda
This is from an abstract of a paper by D. Gabor, Laser speckle and its elimination (IBM Journal of Research and Development, Volume 14, Issue 5, September 1970, Pages 509-514), http://dl.acm.org/citation.cfm?id=16629 ... ayout=flat:
ABSTRACT

"Objective" speckle arises from the uneven illumination of an object with a multiplicity of waves that interfere at its surface. "Subjective" speckle arises at rough objects even if they are illuminated evenly by a single wave. The noise in the image is caused by the interference of the point-figures, which have random phases. Subjective speckle cannot be reduced by except by extending the aperture. On the other hand the "objective" speckle in a plane, for instance in the plane of a transparency, can be reduced, and in the limit made invisible, by a special type of wide-angle illumination. This consists of a one-parameter family of plane waves, which can be produced by diffraction at a special grating, or two crossed gratings, close to the object plane. This makes it possible to produce multiple holograms, with the same insensitivity to dust or scratches as diffused holograms, but without any visible speckle in the reconstruction.
Dinesh

Speckle noise reduction

Post by Dinesh »

Right. "The noise in the image is caused by the interference of the point-figures, which have random phases." and the random phases can be caused by either surface roughness or random variations of index, in the transmissive case. If a multiple of frequencies were to hit (or transmit through) such a structure, the phase relations of each component of the multiple frequencies would average out and reduce the speckle, dependent on the number of frequencies and also, presumably, the cross correlation of their coherence. Presumable by "...and in the limit made invisible", he means that the limit is white light in which all frequencies are present and there is no cross correlation of coherence between the various frequencies. In this case, the random variations of phase would be altering too fast to be seen or recorded.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

I was reviewing some comments by Bob, and he had mentioned that "more object beams, less speckle." Also, I remember that Dave used an interesting technique for object beams: he passed the beam through a diffraction grating, in front of the diffuser. The various orders diffracted by the grating would form multiple points (in a line) which would strike the diffusion screen. By tilting the grating, the separation between the points could be changed arbitrarily. So each point (diffracted order) would generate its own speckle pattern, which would overlap the others. I want to try this as soon as possible, maybe I was barking up the wrong tree.
walschuler

Speckle noise reduction

Post by walschuler »

Dinesh said (» Wed Feb 05, 2014 7:06 am)

There seems to be an opinion that reconstructing a hologram with incoherent ("white") light will result in speckle. Not so. The reason that incoherent light will not create speckle is that the phase changes are so rapid, that the the speckle "blurs out".]

I normally shoot single beam Denisyuks, often shooting 2 colors and sometimes 3, using TEA pre-exposure color baths. In such exposures speckle is a minor issue, because 1) the plate is in a different location for each shot, as I do not try to exactly re-register it or the objects, so the speckle shifts location for each exposure; 2) the emulsion swelling creates different spot sizes for the different colors; and 3) the viewing is in white light and Dinesh's (and others') remarks above apply. White lights incoherence rules out stable speckle.

I usually use the speckle pattern visible when a spread laser beam strikes a surface to show students what it is and remark that it is a hall mark of coherent light.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Speckle noise reduction

Post by Joe Farina »

I saw this interesting technique for reducing speckle:

http://arxiv.org/ftp/arxiv/papers/1212/1212.5176.pdf

Basically, the laser beam goes through a colloidal solution of titanium dioxide, and the Brownian motion of the particles has the effect of cancelling (or averaging) the speckle pattern. Don't know if this can be used in holography, but should be easy enough to test (TiO2 = the pigment titanium white).
User avatar
jsfisher
Posts: 200
Joined: Sun Jan 04, 2015 12:30 am

Speckle noise reduction

Post by jsfisher »

Seems like the brownian motion would destroy coherence.
World's worst holographer
walschuler

Speckle noise reduction

Post by walschuler »

In addition to what Ed and others and I said (see my post above), I would add that you can see that if you move a laser beam across a surface the speckle pattern moves, and if the beam is stable and you move your head your eyes sweep the pattern. Also, the pattern will tend to be less visible if you shoot multicolor images, either with various color lasers or various pseudocolors. In either case the speckle of different colors will be of different scales proportional to wavelength, and tend to blur out.
Dinesh

Speckle noise reduction

Post by Dinesh »

walschuler wrote:I would add that you can see that if you move a laser beam across a surface the speckle pattern moves, and if the beam is stable and you move your head your eyes sweep the pattern
It'll be a slightly different pattern, since the phase of the beam is different in each case.
jsfisher wrote:Seems like the brownian motion would destroy coherence.
I don't think it'd "destroy coherence", since the coherence of the beam is determined by the beam source. This is actually an interesting question that I have been mulling over every now and again. I think it will cause a phase shift due to Doppler effects. The Doppler shift causes, in turn, a frequency shift, which in turn affects the phase. In terms of the frequency, the wave vector is 2*pi*f*n/c. But, due to Doppler, the new wave vector is (2*pi*f*n)/(v*c), where v is the velocity of the Brownian particles. The size of the particles that cause Brownian motion is of the order of a few microns, let's say 10*lambda. It is possible using the Einstein/Smoluchowski equations, to determine the velocity of the Brownian motion. So, you now have the momentum of the Brownian particles. I'm assuming that the medium in which the Doppler particles are embedded does not change its index due to Brownian motion. Thus, it is possible to determine whether the phase shift is sufficient to deviate the fringe structure by lambda/4. If so, then, since Brownian motion is random, then the fringe structure will be altering too fast to take a recording. It's not coherence that's lost, it's the phase relationship between the two beams. Included in this, although slightly different, is of course, the effect on phase of a turbulent medium. In the case of a turbulent medium, the "Brownian particles" are a fluid and so vary continuously.
Joe Farina
Posts: 804
Joined: Wed Jan 07, 2015 2:10 pm

Re: Speckle noise reduction

Post by Joe Farina »

Thanks to all the forum members who contributed to this post. I've learned quite a bit.

Having run some tests to reduce speckle in my holograms, I've found that a moving piece of clear acrylic (or glass) placed after the spread beam, and before the diffusion screen, will reduce speckle to a large extent. This applies to the object beam only, of course, in a split beam, assisted Denisyuk, or bypass arrangement. In my case, this is a lot easier than moving the diffusion screen itself. (The movement method should be isolated from the table.)

To test this, I set a clear piece of acrylic on my cell phone, which is on vibrate only. (The intensity was on high.) I vibrated it by calling from another phone, and the speckle essentially disappeared.

I bought some vibration motors on Amazon, and will try attaching one to the acrylic:

https://www.amazon.com/Vibration-Button ... =8-14&th=1
Post Reply