 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Medical Forum / General / Vision / November 2009contrast sensitivity, visual acuity, and Photoshop
Argh! I've been reading about "contrast sensitivity", "visual acuity", and wonder what the difference is. When I try to find this out, I am immersed in "cycles per degree". When I try to find out what that is, I get math. It's so abstract. :-( Are these two things different? It seems that both are measured in "cycles per degree". Is there a way of explaining how CS differs from VA (if it does), using terms used in photography? If you have poor CS or VA, what do you not see? Resolution, shadow detail, sharpness? ...... help! thanks, Liz Indianapolis IN USA thanks, Liz In article <00b8ae25-844f-4164-9c2b-e8aa6f5e891d@p35g2000yqh.googlegroups.com>, > Argh! > [quoted text clipped - 19 lines] > thanks, > Liz Sorry Liz. If you are trying to quantify these terms to an objective standard, I do not see how you can avoid math. Otherwise, all you have is English teach arty-fartiness. Just take the Snellen chart. Not all letters require the same acuity to be read. The cycles per degree or spatial frequency gets around that. Even so, seeing is not what is called a linear physical phenomenon. There are all kinds of things that upsets visual measurement, even those based upon spatial frequency. Bill  SignatureAs the years go by, dying just before having to fill out a tax return has merit. http://www.google.co.uk/search?hl=en&source=hp&q=define%3A+contrast+sensitivity& btnG=Google+Search&meta=&aq=f&oq= http://www.google.co.uk/search?hl=en&q=define%3A+visual+acuity&btnG=Search&meta= Easy. > Sorry Liz. If you are trying to quantify these terms to an objective > standard, I do not see how you can avoid math. I'm not complaining about the measurement aspect. Of course you have to use math to measure it. I'm saying that I can't figure out WHAT is being measured. I mean in terms of anything concrete that I know what it is. Unfortunately I don't know what "spatial frequency", "transfer modulation function", are. Sorry! Yes, I looked them up and read several times. Liz In article <da6362e5-b77e-40f4-87eb-101ce2d9df39@c3g2000yqd.googlegroups.com>, > > Sorry Liz. If you are trying to quantify these terms to an objective > > standard, I do not see how you can avoid math. [quoted text clipped - 9 lines] > > Liz Do you understand audio or temporal frequency? If you do, you understand that your hearing system has limited high frequency response. That response limits the sharpness of clicks you might be able to hear. Similarly, your visual system has limited capability for things changing rapidly as a function of position. Spatial frequency corresponds to the rate at which a picture can "wave" as a function of position in the picture. If that spatial frequency is high enough, say in black and white, it becomes harder to tell whether you have gray rather than a fluctuation from black to white and back. Does that make any sense? Bill SignatureAs the years go by, dying just before having to fill out a tax return has merit. Liz <fraternobombus@yahoo.com> wrote in part: > Argh! > > I've been reading about "contrast sensitivity", "visual > acuity", and wonder what the difference is. When I try to find > this out, I am immersed in "cycles per degree". When I try > to find out what that is, I get math. It's so abstract. :-( Some of us like that stuff ! > Are these two things different? It seems that both are > measured in "cycles per degree". People often use terms imprecisely. I would say "visual acuity" presumes some sort of standard contrast, while "contrast sensitivity" is somewhat more general. > Is there a way of explaining how CS differs from VA (if it > does), using terms used in photography? Oh, a photog -- look up "Modular Transfer Function" ... maybe not if you dislike abstraction more than ignorance. "cycles per degree" is similar to "line pairs per mm" > If you have poor CS or VA, what do you not see? > Resolution, shadow detail, sharpness? ...... help! It is all the same thing! You can see wires against sky with very high resolution but cannot see something big camouflaged. Contrast is what drives vision, and the more of it there is, the smaller details you can see. -- Robert > I've been reading about "contrast sensitivity", "visual acuity", and > wonder what the difference is. When I try to find this out, I am > immersed in "cycles per degree". When I try to find out what that is, > I get math. It's so abstract. Pretend your eye is the center of a circle. If you look at a letter "E" several feet away, you could measure the height of the letter in degrees, couldn't you? The nice thing about degrees is we don't care how far away something is, as long as it covers the same angle. Since an "E" has three "arms" we can say, going vertically, it's black, then white, then black, then white again, then black. That's 2.5 black/white "cycles", or a 2.5-cycle "square wave grating." If the E is one degree tall, it's 2.5 cycles/degree. If the letter E is very blurry, you have something that isn't solid black-white-black, but rather lighter-darker-lighter etc, plotting a "sine wave" pattern if you graph the brightness. Patterns of darker-lighter-darker are called "sine wave gratings." "Contrast sensitivity" explores how well you see the pattern when it's made up of fainter and fainter shades of gray. It's typically tested with sine-wave patterns but you can also gray out a pattern of solid bars without blurring it, or use a gray snellen chart, for a similar effect. -MT Mike, your explanation is SO much better than those articles! > > I've been reading about "contrast sensitivity", "visual acuity", and > > wonder what the difference is. > > Pretend your eye is the center of a circle. If you look at a letter "E" > several feet away, you could measure the height of the letter in degrees, > couldn't you? Degrees as in 0 to 360? The way astronomers say something is "15 degrees above the horizon"? I think so. In other words, straight ahead is 0, straight overhead is 90, and the space in between is divided into 90 equal pieces... > The nice thing about degrees is we don't care how far away something is, as > long as it covers the same angle. Yes, I see how the distance wouldn't matter. OK. > Since an "E" has three "arms" we can say, going vertically, it's black, then > white, then black, then white again, then black. That's 2.5 black/white > "cycles", or a 2.5-cycle "square wave grating." If the E is one degree tall, > it's 2.5 cycles/degree. So a "cycle" is about changes in appearance. From black to white, or silver to pink, or any other change that's visibly different from what's next to it? And the number of "cycles" is the number of pairs of distinguishable objects? Would black green white pink white grey orange blue be 4 cycles? > If the letter E is very blurry, you have something that isn't solid > black-white-black, but rather lighter-darker-lighter etc, plotting a "sine > wave" pattern if you graph the brightness. Patterns of darker-lighter-darker > are called "sine wave gratings." OK. Are they described using cycles too? > "Contrast sensitivity" explores how well you see the pattern when it's made > up of fainter and fainter shades of gray. It's typically tested with > sine-wave patterns but you can also gray out a pattern of solid bars without > blurring it, or use a gray snellen chart, for a similar effect. OK. Try this: You have a white piece of paper with horizontal black bars on it. Someone walks further and further away with this piece of paper. Finally they are so far away that you cannot see the bars; you only see the whole paper as grey. That is an example of a limit of your visual acuity. Right? (In Photoshop I would call this "resolution".) Suppose instead of black and white, you have black and dark grey. You can tell them apart when the paper is near you. The person walks away. They haven't walked very far - nowhere near as far as they went with the black and white paper - when you can no longer see any barring. The whole paper looks one dark color. This is a failure in contrast sensitivity? (In Photoshop I would call this "tonal contrast".) But..... hmmmm. I feel as though there is a lot more but I have to think of how the situations differ. I think there must be interactions between being able to discern a sharp edge and being able to see a tonal difference. Liz > Degrees as in 0 to 360? The way astronomers say something is "15 > degrees above the horizon"? I think so. In other words, straight > ahead is 0, straight overhead is 90, and the space in between is > divided into 90 equal pieces... BinGO gurlie. > So a "cycle" is about changes in appearance. From black to white, or > silver to pink, or any other change that's visibly different from > what's next to it? That's the principle, absolutely. > Would black green white pink white grey orange blue be 4 cycles? Generally a "cycle" would be a repetition, any repetition. But if we limit our analysis to "cycles" we can plot as waves, then the math becomes MUCH simpler, because we already have mathematics for waves. The same trigonometry works for all waves, in electricity, water, and music, as well as light. >> wave" pattern if you graph the brightness. Patterns of >> darker-lighter-darker >> are called "sine wave gratings." > > OK. Are they described using cycles too? Absolutely. "Cycles per degree" is a unit of spatial frequency, the frequency of a pattern repeating across space. > Try this: > You have a white piece of paper with horizontal black bars on it. [quoted text clipped - 4 lines] > Right? > (In Photoshop I would call this "resolution".) That's precisely correct- resolution in cycles per degree. Then you also realize resolution is not contrast. Low contrast targets can be sharp. Blurry targets can have 100% contrast (blackest at 0, whitest at 1). In photography, the emulsion or the CCD will record pixel-for-pixel what it sees. The retina is different. It distills "pixel" information into signals about borders with orientation and direction of motion. Contrast plays a role, as does movement, dark adaptation, retinal eccentricity, and age. Sorry I wander, but no borders=no vision. > Suppose instead of black and white, you have black and dark grey. You > can tell them apart when the paper is near you. That's "low contrast." But still square-wave, right, because it's well-focused letters, right? > The person walks away. > They haven't walked very far - nowhere near as far as they went with > the black and white paper - when you can no longer see any barring. > The whole paper looks one dark color. > This is a failure in contrast sensitivity? Well, it's a failure in *sensitivity." In this case you're confounding two variables and you don't know which had more effect, contrast or resolution. Put that letter on a new photoshop layer, reduce the contrast until you can barely-barely see the letter. That's testing _contrast_ specifically. > (In Photoshop I would call this "tonal contrast".) But you wouldn't confuse it with resolution. In this particular situation, you have three or more variables and you must control them. Have a glimpse at http://www.contrastsensitivity.net/csc.html. Those are all clinical CS tests that are done in controlled circumstances. CS would be tedious and difficult for creating eyeglass prescriptions. CS testing is done with corrected vision, because CS studies are more neurological, functional testing for the retina and optic nerve. Refractive blur (resolution effects) would be artifact. > I think there must be interactions between being able to discern a > sharp edge and being able to see a tonal difference. You are precisely right. And we're talking about the same thing. Spatial frequency is how many cycles per degree. Sharp edges and snellen charts are considered square waves in the laboratory. Round edges, blurry images, behave like sine waves and trigonometry applies like everywhere else there are sine waves. But it's the "spatial" domain and not "temporal" - the x-axis is distance, not time. Treating space like time, we get some interesting conclusions. Like, mathematically, a true 90-degree square wave has infinite frequencies at its corners. In music, you hear this in the "buzz" differentiating a square or triangular waveform from a sinusoidal one. Fun, huh? -MT, terminally geek >Treating space like time, we get some interesting conclusions...... No no - time (and sounds) confuse me more! Please, let's stick with only what the eyeball can detect..... >> So a "cycle" is about changes in appearance. From black to white, or silver to pink, or any other change that's visibly different from what's next to it? > That's the principle, absolutely. >> Would black green white pink white grey orange blue be 4 cycles? > Generally a "cycle" would be a repetition. ...... "Cycles per degree" is a unit of spatial frequency, the frequency of a pattern repeating across space. OK. But what if it never repeats? Say that you look in front of you, left to right. You see a green leaf next to a brown leaf next to open sky next to grey bark next to a yellow leaf. Nothing has repeated, so how many cycles do you call it? (Or do we not want to go there?) >>Try this: >>You have a white piece of paper with horizontal black bars on it. Someone walks further and further away with this piece of paper. Finally they are so far away that you cannot see the bars; you only see the whole paper as grey. That is an example of a limit of your visual acuity. Right? (In Photoshop I would call this "resolution".) > That's precisely correct- resolution in cycles per degree. >Then you also realize resolution is not contrast. Of course not. >Low contrast targets can be sharp. Yes. >Blurry targets can have 100% contrast (blackest at 0, whitest at 1). Yes. (Except, I tend to call this "full spread of tonal values", because I picture "contrast" as having to do mainly with edges. But "contrast" also means differences in light/dark. "Contrast" is a word with too many meanings.) >In photography, the emulsion or the CCD will record pixel-for-pixel what it sees. The retina is different. It distills "pixel" information into signals about borders...... Yes, I have heard that our brain "notices" edges. That's why Photoshop has "sharpening". (Bumblebees see edges well, too.) > Contrast plays a role, as does movement...... but no borders=no vision. I didn't know it was that extreme, but OK. >> Suppose instead of black and white, you have black and dark grey. You can tell them apart when the paper is near you. >That's "low contrast." But still square-wave, right, because it's well-focused letters, right? Right. Crisp, clean edges. >> The person walks away. They haven't walked very far - nowhere near as far as they went with the black and white paper - when you can no longer see any barring. The whole paper looks one dark color. This is a failure in contrast sensitivity? >Well, it's a failure in *sensitivity." In this case you're confounding two variables and you don't know which had more effect, contrast or resolution. Oh no. That's what I was afraid of. >> I think there must be interactions between being able to discern a sharp edge and being able to see a tonal difference. >You are precisely right. .... OK. >In this particular situation, you have three or more variables and you must control them. What's the third variable here? > www.contrastsensitivity.net/csc.html >Those are all clinical CS tests that are done in controlled circumstances. OH. Yes, I see what they mean. (OMG - they want $1600 for a chart!!!) I see visually what the charts are trying to do. Howeer, when I look at an MTF graph, I still don't quite follow what is being measured. The horizontal axis is in spatial cycles per degree. We know what that is. The vertical axis is always 1 to 100 percent, but... percent of WHAT? :-( Liz (sorry, I still can't get Google to do hard returns properly) L. > Howeer, when I look at an MTF graph, I still don't quite follow what > is being measured. The horizontal axis is in spatial cycles per > degree. We know what that is. > The vertical axis is always 1 to 100 percent, but... > percent of WHAT? Contrast. 100% contrast is black on white. 1% is very light gray on very light gray. In simple terms, It is a graph of acuity vs contrast. Judy > > Howeer, when I look at an MTF graph, I still don't quite follow what > > is being measured. The horizontal axis is in spatial cycles per [quoted text clipped - 6 lines] > > In simple terms, It is a graph of acuity vs contrast. And each point on the graph represents the best the person can see? So the bigger the percentage contrast, the more cycles per degree they should be able to see.... So the dots on the graph should go from the lower left to the upper right....? That is, if there is a dot at 10% and 4 cycles, that suggests that the person can also see 20% at 4 cycles, 30% at 4 cycles, 40% at 4 cycles, etc. And it suggests that they should be able to see 10% at 3 cycles, 2 cycles. True? But the lines don't usually look that way; they're humpier. Liz It would be helpful to look at http://en.wikipedia.org/wiki/Test_card. In particular, look at the gallery. There are a number of examples of wedged lines in which the spatial frequency varies along the wedge. After looking at that, more meaningful quesgtions can be formulated. Bill SignatureAs the years go by, dying just before having to fill out a tax return has merit. > > In simple terms, It is a graph of acuity vs contrast. > > And each point on the graph represents the best the person can see? A particular point will represent the lowest contrast at which that particular acuity (cycles per degree) can still be detected. > So the bigger the percentage contrast, the more cycles per degree they > should be able to see.... You would think so but it turns out the human visual system doesn't work that way. Both very high and very low contrast results in fewer cycles per degree. Some vision disorders impact certain cycles more than others. > So the dots on the graph should go from the lower left to the upper > right....? Not always. It's called the contrast sensitivity function because it is a complex curve, not a simple line. > That is, if there is a dot at 10% and 4 cycles, that suggests that the > person can also see 20% at 4 cycles, 30% at 4 cycles, 40% at 4 cycles, Not always. If you want to know if the person can see 4 cycles at 40%, you need to measure it. > And it suggests that they should be able to see 10% at 3 cycles, 2 > cycles. No always. If you want to know if the person can see 3 cycles with 10% contrast, you need to measure it. > But the lines don't usually look that way; they're humpier. Yup. It's a complex curve. Judy In article <42ed73b7-73db-4f0a-8825-4e48bfcad815@k4g2000yqb.googlegroups.com>, > In simple terms, It is a graph of acuity vs contrast. What is the (preferably quantitative) definition of acuity? Bill SignatureAs the years go by, dying just before having to fill out a tax return has merit. > OK. But what if it never repeats? Say that you look in front of > you, left to right. You see a green leaf next to a brown leaf next to > open sky next to grey bark next to a yellow leaf. Nothing has > repeated, so how many cycles do you call it? (Or do we not want to > go there?) You can go there, it's just hard to put into numbers. Vision is made up of borders and edges of all sorts. But you asked about spatial frequency and that's a mathematical term and explaining it requires a narrow definition for "cycles" as simple, uniform and repetitive. > Yes. (Except, I tend to call this "full spread of tonal values", In the darkroom we called it "the zone." >because I picture "contrast" as having to do mainly with edges. Bzzzt. Contrast has nothing to do with edges. Contrast is the range, the difference between the blackest black and the whitest white in your image. Mathematically it's (max-min)/max IIRC. Boost the contrast to max, and you only black and white with no grays. Reduce contrast to minimum and you get 18% gray fog. Right up until the image fades to gray, if there are edges detectable they are not necessarily _blurred._ > But "contrast" also means differences in light/dark. > "Contrast" is a word with too many meanings. I think you'll find the dictionary definitions are less ambiguous. >> Contrast plays a role, as does movement...... but no borders=no vision. > I didn't know it was that extreme, but OK. Yes. Vision fades out in about 15 seconds if you can completely immobilize the eye. It keeps moving because it has to. > What's the third variable here? Contrast, resolution, and blur related to distance (optical focus.) It has the effect of reducing both resolution and contrast. Other variables are motion, intensity, and eccentricity off the macula. > Howeer, when I look at an MTF graph, I still don't quite follow what > is being measured. The horizontal axis is in spatial cycles per > degree. We know what that is. > The vertical axis is always 1 to 100 percent, but... > percent of WHAT? To get a graph of sensitivity, you measure threshold (minimum visible contrast) then turn the graph over, so it's y=1/threshold. Then there's an upper limit (100%) because it's where contrast is zero. > :-( :-) > > OK. But what if it never repeats? Say that you look in front of > > you, left to right. You see a green leaf next to a brown leaf next to [quoted text clipped - 3 lines] > > You can go there, it's just hard to put into numbers. I can imagine that a whole landscape (as when you're hunting birds) would be a mess to describe. > > Yes. (Except, I tend to call this "full spread of tonal values", > [quoted text clipped - 3 lines] > > Bzzzt. Contrast has nothing to do with edges. Contrast is the range, THAT'S it. The *range* of tonal values. (What you fix with Levels.) > Mathematically it's (max-min)/max IIRC. ACK!! Mike, stop! > >> Contrast plays a role, as does movement...... but no borders=no vision. > > I didn't know it was that extreme, but OK. > > Yes. Vision fades out in about 15 seconds if you can completely immobilize > the eye. It keeps moving because it has to. Ah... I've heard that people see stationary lights in the dark as moving around, and think they're UFOs because they seem to be moving, when in fact it's their eye moving with no other visual reference points because everything else is dark. (I won't comment on the other reasons they think it's a UFO.) > > What's the third variable here? > > Contrast, resolution, and blur related to distance (optical focus.) It has > the effect of reducing both resolution and contrast. Other variables are > motion, intensity, and eccentricity off the macula. Leave those last three out for the moment. Blur related to distance? You don't mean light scattering around in the air between the person and the object...? I can't Google optical focus without abandoning this message. > > Howeer, when I look at an MTF graph, I still don't quite follow what > > is being measured. The horizontal axis is in spatial cycles per [quoted text clipped - 4 lines] > To get a graph of sensitivity, you measure threshold (minimum visible > contrast) then turn the graph over, so it's y=1/threshold. I'm confused, I need to think about this. Liz > > > when I look at an MTF graph, I still don't quite follow what is being measured. The horizontal axis is in spatial cycles per degree. > > > The vertical axis is always 1 to 100 percent, but percent of WHAT? > > To get a graph of sensitivity, you measure threshold (minimum visible contrast) then turn the graph over, so it's y=1/threshold. I'm not sure how minimum visible contrast is measured. If the person sees a 40% contrast, is that .40? Can you give me one or two points on the graph as an example? That is, what's X, what's Y, and how did you get them? (Make up a number, it need not represent a real case of anyone's vision.) I feel like I'm almost there but not quite. thanks, Liz >I'm not sure how minimum visible contrast is measured. If the person >sees a 40% contrast, is that .40? >Can you give me one or two points on the graph as an example? On the x axis is frequency, cycles per degree, big gratings at the low end and small gratings at the high end. On the y axis is sensitivity. Think of it as "threshhold" where a certain minimal amount of contrast is necessary to see the grating at that frequency. Across the x axis you'll get a slightly concave curve, tails upward at each side becase more contrast is necessary to see at higher and lower frequencies. Now just invert the graph and call the y axis "sensitivity" which is the inverse of threshhold. -MT > >I'm not sure how minimum visible contrast is measured. If the person > >sees a 40% contrast, is that .40? > >Can you give me one or two points on the graph as an example? > > On the x axis is frequency, cycles per degree, big gratings at the low end > and small gratings at the high end. Right. > On the y axis is sensitivity. Think of it as "threshhold" where a certain > minimal amount of contrast is necessary to see the grating at that > frequency. Oh. > Across the x axis you'll get a slightly concave curve, tails upward at each > side becase more contrast is necessary to see at higher and lower > frequencies. Yes. > Now just invert the graph and call the y axis "sensitivity" which is the > inverse of threshhold. OH... so that's what you meant by "turn it upside down". Yes, most of these graphs seem to have "inverted smiley" curves. The graphs seem to use a variety of different units to express "sensitivity". (Or whatever they are calling whatever is on their Y axis.) Some use the log of something. A lot don't even tell what units Y is in (my 5th-grade math teacher would flunk them instantly). I'm not sure what units are typical. Will look at more graphs and see if there is some sort of standard pattern. thanks, Liz > Can you give me one or two points on the graph as an example? > > That is, what's X, what's Y, and how did you get them? Have you looked at the Wikipedia on CSF, it includes a diagram that might help you: http://en.wikipedia.org/wiki/Contrast_(vision)#Contrast_sensitivity Judy > Have you looked at the Wikipedia on CSF, it includes a diagram that > might help you: > > http://en.wikipedia.org/wiki/Contrast_(vision)#Contrast_sensitivity Yes, this is one of the best visual explanations I've seen - worth a thousand words!... It's interesting to put a piece of cardboard over your screen, covering the bottom half of the diagram, and move it towards the top and back towards the bottom to find out where you can start to see contrast. cheers, Liz |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.