Lighting in a identification tasks

[max_i]

Hello, All!

Please, anybody can help me?

Our task is to identification the position of the needle of a analog needle meters.

It is exploited for the experimental work.
Now we use the web-camera (Genius iSlim 1300) with halogen lighting for this.

However, we come into collision with some disadvantages:

- are not sure durability of the system;
- lamps consume much electricity;
- impossible for a man a long look at the scale;

- light does not adequately identification of the quality, as needle is very thin and the metal, which results in some cases, disappearance of read an image due to reflection off.

What hardware would you recommend (camera, lighting. arrangement) and etc?

Need:

- small size (like used camera),

- around-the-clock (may be even in dark conditions),

- sampling rate in 4 times per second is enough,

- preferably USB connection.

I've heard about the IR illumination (satisfying our requirements).

But it is, as I read, can only be used with black-and-white cameras (in principle to us does not matter).

What solution you could recommend for the reliable result (how it can see and read)?

Looking forward to a response!!!

Thanks!

Max O. 

Engineer and software developer,

TeSLa.

[BruceAmmons]

Max,

I would probably select an LED ring light.  I would probably use one that is larger than the dial so that the light hits it at about a 45 to 60 degree angle.  This would hopefully eliminate the reflection of the lights in the image.  It will also eliminate the shadows.

If you can't afford machine vision lighting, I have seen some very inexpensive ring lights that are designed for accent lighting for automobiles.  They were popular for FIRST robots last year.

Bruce

[max_i]

Thank you for the help, Bruce!

What the wavelength of the LED ring light is better to use?

There are references to the specific models?

Ring lights should be placed coaxially of the camera?

Max O. 

Engineer and software developer,

TeSLa.

[BruceAmmons]

I would expect any color to work well.  Red is usually the cheapest.  White is good too, but usually more expensive and not really necessary.

If you search the internet for "LED ring light", you should find all sorts of information.

Yes, the ring light usually is located around the lens of the camera, but can be closer than the camera if desired.

Is the glass in front of the meter parallel to the background, or angled?  Angled could have problems with reflections.

Bruce

[AnalogKid2DigitalMan]

Bruce:

Would a ring or area (pancake) light with a diffuser be a better choice?

I think the Vision Assitant software package comes with a shipping example for reading analog needle meters.

-AK2DM

[max_i]

Hello,

Thanks for the idea, Bruce!
We will attempt to.

That is, from the point of view of the identification quality IR LED does not have a great advantages compared with White backlight (comparing only LED Backlights)?

The glass in the front of the meter is parallel to the background.

Thank you, AnalogKid2DigitalMan!

We just use NI Vision Development Module and NI-IMAQ for reading analog needle meter.

Max O.
Engineer and software developer,
TeSLa.

[BruceAmmons]

A flat or pancake light wouldn't really improve the lighting significantly, and you would probably see a reflection of the light in the glass.

IR lighting is usually only used when regular lighting will not work.  Since regular lighting will work well, there is no reason to use it.

As long as the ring light is about twice the diameter of the gage, you should be fine.

Bruce

[max_i]

Hello, Bruce!

Sorry for molestation.

Ie a simple Web-camera and a White LED ring lights completely meet our needs?

Thanks!

Best regards,

Max O. 

Engineer and software developer,

TeSLa.

[BruceAmmons]

I would think so, but I can't make any promises.

Bruce

[max_i]

Hello, Bruce!

While working on the lighting we faced still having some problems.
Using the LED ring of the 16 white LEDs disappeared shadows that were on the previous image (left one shade under the arrow, which does not prevent the identification). The light is distributed unevenly, but it also is not prevent identification in principle. On the picture can see the reflection of light (bright areas) from the glass. Perhaps this is due to the insufficient of scattering. We use transparent plexiglass 2mm backed with both sides light scattering film. May be here need used completely diffusing plexiglass or clear plexiglass with matte film?
What solution would you recommend for this problem?

And if take 2 lamps like this http://ru.farnell.com/pro-elec/mr16wt15/led-lamp-12v-mr16-white-15-led/dp/1466118

with a scattering film and secure the sides of the camera, the shadow should again appear?
Also, the backlight is not enough if an identifiable area exposed to direct sunlight.
Though, system is more stable in a closed room with constant lighting.
In this regard, a few questions:
Is there any material, literature, references to the calculation of LED lighting?
What should be the illumination value of the area (identification area is about 15 * 15 cm), the brightness of the LEDs for reliable identification at all times (but not in the full dark)?

For example, the lighting in the afternoon sun is 100000 lux.

Thank you!

Best regards,

Max O. 

Engineer and software developer,

TeSLa.

[max_i]

Hello, Bruce!

While working on the lighting we faced still having some problems.
Using the LED ring of the 16 white LEDs disappeared shadows that were on the previous image (left one shade under the arrow, which does not prevent the identification). The light is distributed unevenly, but it also is not prevent identification in principle. On the picture can see the reflection of light (bright areas) from the glass. Perhaps this is due to the insufficient of scattering. We use transparent plexiglass 2mm backed with both sides light scattering film. May be here need used completely diffusing plexiglass or clear plexiglass with matte film?
What solution would you recommend for this problem?

And if take 2 lamps like this http://ru.farnell.com/pro-elec/mr16wt15/led-lamp-12v-mr16-white-15-led/dp/1466118

with a scattering film and secure the sides of the camera, the shadow should again appear?
Also, the backlight is not enough if an identifiable area exposed to direct sunlight.
Though, system is more stable in a closed room with constant lighting.
In this regard, a few questions:
Is there any material, literature, references to the calculation of LED lighting?
What should be the illumination value of the area (identification area is about 15 * 15 cm), the brightness of the LEDs for reliable identification at all times (but not in the full dark)?

For example, the lighting in the afternoon sun is 100000 lux.

Thank you!

Best regards,

Max O. 

Engineer and software developer,

TeSLa.