Tim Wescott wrote:
> For 2 & 3 I'd check to make sure you're not having a vibration problem 
> -- does it do this on the ground, or only when you're shaking the h*** 
> out of it with the helicopter?  Remember that the notion that a 
> helicopter flies aerodynamically is a popular misconception -- 
> helicopters fly by vibrating so hard that some components exceed the 
> speed of light.  This causes an anti-gravity effect that makes the 
> helicopter lift of the ground.  The blades are just there for control. 
> This high vibration environment can be bad on electronics, particularly 
> if some of their sensitive components are the ones exceeding the speed 
> of light.
> 
Helicopter vibration is bad even at sub-light speeds. If you use direct 
edge connector PCBs, or IC sockets with flat pressed contacts, the 
contacts can wear completely away over a few month's use in a 
helicopter. Interestingly, while the metal surfaces are clearly in 
motion against each other to cause this, the system may function 
perfectly up to the point where the last of the metal has worn away. 
I've experienced that a few times, and I've never figured out how it 
happens. Shouldn't there be a lot of crackly connections if physical 
wear is occurring?

Steve

On Mon, 19 Nov 2007 22:08:37 -0500, Jerry Avins wrote:

> Tim Wescott wrote:
>> Adam Chapman wrote:
>>> Hi,
>>>
>>> I'm working on a robot vision system for a helicopter (http://
>>> www.rctoys.com/rc-toys-and-parts/DF-TSU/INDUSTRIAL.html) at
>>> university.
>>>
>>> I've just seen for the first time some video captured from the
>>> aircraft's onboard camera. The camera itself is not stabilised and
>>> rather small, the resolution is about 500 lines.
>>>
>>> The recorded video was sent from the aircraft to an analogue receiver
>>> via an analogue  radio data link.
>>>
>>> In this video there are a few types of noise.
>>>
>>> 1. "snowy" noise, i.e. little white dots at fairly high spatial
>>> frequency.
>>> 2. horizontal straight white lines that appear in one video frame,
>>> then dissapear again.
>>> 3. a colour channel in one frame appears higher then the others, e.g.
>>> a white object might flash green in one video frame.
>>>
>>>
>>> Could anybody tell me what causes these types of noise and if these
>>> phenomena have proper technical names?
>>>
>>> My visuall system has worked so far in simulation, but the videos i
>>> have used have not been noisy.
>>>
>>> I would like to be able to treat the cause rather than the effect, but
>>> if this is not possible i will have to try and cancel the noise with
>>> further processing.
>>>
>>> I think the best thing for the snowy noise is a median filter, but i
>>> also know that would be very expensive in computational terms.
>>>
>>> The minimum loss of image quality is essential for this image-
>>> recognition application.
>>>
>>> Thankyou very much for any advice offered
>>>
>>> Adam
>> 
>> The "snow" is most likely from Gaussian noise, as a consequence of a low 
>> path gain from the transmitter to the receiver.  If it is, a better 
>> receiver, more powerful transmitter, or better antennas will fix it.
>> 
>> The single white horizontal line sounds like an intermittent fault in 
>> the camera, although if the receiver is digital it could be getting 
>> faked out by a weak horizontal sync pulse.  Ditto the flashing color 
>> change in the video.
> 
> Loss of color-burst lock could also cause the color flash. The details 
> depend on PAL, NTSC, SECAM, or maybe some other transmission standard.
> 
>> For 2 & 3 I'd check to make sure you're not having a vibration problem 
>> -- does it do this on the ground, or only when you're shaking the h*** 
>> out of it with the helicopter?  Remember that the notion that a 
>> helicopter flies aerodynamically is a popular misconception -- 
>> helicopters fly by vibrating so hard that some components exceed the 
>> speed of light.  This causes an anti-gravity effect that makes the 
>> helicopter lift of the ground.  The blades are just there for control. 
>> This high vibration environment can be bad on electronics, particularly 
>> if some of their sensitive components are the ones exceeding the speed 
>> of light.
> 
> May I quote you on that? If yes, would you prefer attribution, or anonymity?
> 
> jerry

Yes you may, and I think I want attribution.

-- 
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html

Tim Wescott wrote:
> Adam Chapman wrote:
>> Hi,
>>
>> I'm working on a robot vision system for a helicopter (http://
>> www.rctoys.com/rc-toys-and-parts/DF-TSU/INDUSTRIAL.html) at
>> university.
>>
>> I've just seen for the first time some video captured from the
>> aircraft's onboard camera. The camera itself is not stabilised and
>> rather small, the resolution is about 500 lines.
>>
>> The recorded video was sent from the aircraft to an analogue receiver
>> via an analogue  radio data link.
>>
>> In this video there are a few types of noise.
>>
>> 1. "snowy" noise, i.e. little white dots at fairly high spatial
>> frequency.
>> 2. horizontal straight white lines that appear in one video frame,
>> then dissapear again.
>> 3. a colour channel in one frame appears higher then the others, e.g.
>> a white object might flash green in one video frame.
>>
>>
>> Could anybody tell me what causes these types of noise and if these
>> phenomena have proper technical names?
>>
>> My visuall system has worked so far in simulation, but the videos i
>> have used have not been noisy.
>>
>> I would like to be able to treat the cause rather than the effect, but
>> if this is not possible i will have to try and cancel the noise with
>> further processing.
>>
>> I think the best thing for the snowy noise is a median filter, but i
>> also know that would be very expensive in computational terms.
>>
>> The minimum loss of image quality is essential for this image-
>> recognition application.
>>
>> Thankyou very much for any advice offered
>>
>> Adam
> 
> The "snow" is most likely from Gaussian noise, as a consequence of a low 
> path gain from the transmitter to the receiver.  If it is, a better 
> receiver, more powerful transmitter, or better antennas will fix it.
> 
> The single white horizontal line sounds like an intermittent fault in 
> the camera, although if the receiver is digital it could be getting 
> faked out by a weak horizontal sync pulse.  Ditto the flashing color 
> change in the video.

Loss of color-burst lock could also cause the color flash. The details 
depend on PAL, NTSC, SECAM, or maybe some other transmission standard.

> For 2 & 3 I'd check to make sure you're not having a vibration problem 
> -- does it do this on the ground, or only when you're shaking the h*** 
> out of it with the helicopter?  Remember that the notion that a 
> helicopter flies aerodynamically is a popular misconception -- 
> helicopters fly by vibrating so hard that some components exceed the 
> speed of light.  This causes an anti-gravity effect that makes the 
> helicopter lift of the ground.  The blades are just there for control. 
> This high vibration environment can be bad on electronics, particularly 
> if some of their sensitive components are the ones exceeding the speed 
> of light.

May I quote you on that? If yes, would you prefer attribution, or anonymity?

jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Adam Chapman schrieb:
> In this video there are a few types of noise.
> 
> 1. "snowy" noise, i.e. little white dots at fairly high spatial
> frequency.

High frequency noise. Bad SNR.

> 2. horizontal straight white lines that appear in one video frame,
> then dissapear again.

LF noise.

> 3. a colour channel in one frame appears higher then the others, e.g.
> a white object might flash green in one video frame.

NTSC?

NTSC can't deal well with phase errors. That's why it's called Never The 
Same Color. PAL performs much better with respect to this as long as you 
take the average of 2 consecutive lines of the picture. However, this 
won't help you.


> I think the best thing for the snowy noise is a median filter, but i
> also know that would be very expensive in computational terms.
> 
> The minimum loss of image quality is essential for this image-
> recognition application.

Most likely the SNR and phase noise of your RF link is not suitable for 
your needs. But first of all you should check whether all kinds of noise 
come from the RF transmission. (Record to some media in the airplaine.)
In this case you should consider a digital transfer (with well known 
compression atrefacts). You should enrich your data with some ECC data 
before the modulation too. Normally it is no problem to fit high quality 
compressed video data of one video channel to the bandwith of an analog 
video channel.
You further should use some kind of standard for this purpose to get 
cheap and available A/D and D/A conversion units. Something like an MPEG 
video stream with COFDM and QPSK. (Don't know what digital standards are 
common outside europe.)


Marcel

Adam Chapman wrote:
> Hi,
> 
> I'm working on a robot vision system for a helicopter (http://
> www.rctoys.com/rc-toys-and-parts/DF-TSU/INDUSTRIAL.html) at
> university.
> 
> I've just seen for the first time some video captured from the
> aircraft's onboard camera. The camera itself is not stabilised and
> rather small, the resolution is about 500 lines.
> 
> The recorded video was sent from the aircraft to an analogue receiver
> via an analogue  radio data link.
> 
> In this video there are a few types of noise.
> 
> 1. "snowy" noise, i.e. little white dots at fairly high spatial
> frequency.
> 2. horizontal straight white lines that appear in one video frame,
> then dissapear again.
> 3. a colour channel in one frame appears higher then the others, e.g.
> a white object might flash green in one video frame.
> 
> 
> Could anybody tell me what causes these types of noise and if these
> phenomena have proper technical names?
> 
> My visuall system has worked so far in simulation, but the videos i
> have used have not been noisy.
> 
> I would like to be able to treat the cause rather than the effect, but
> if this is not possible i will have to try and cancel the noise with
> further processing.
> 
> I think the best thing for the snowy noise is a median filter, but i
> also know that would be very expensive in computational terms.
> 
> The minimum loss of image quality is essential for this image-
> recognition application.
> 
> Thankyou very much for any advice offered
> 
> Adam

The "snow" is most likely from Gaussian noise, as a consequence of a low 
path gain from the transmitter to the receiver.  If it is, a better 
receiver, more powerful transmitter, or better antennas will fix it.

The single white horizontal line sounds like an intermittent fault in 
the camera, although if the receiver is digital it could be getting 
faked out by a weak horizontal sync pulse.  Ditto the flashing color 
change in the video.

For 2 & 3 I'd check to make sure you're not having a vibration problem 
-- does it do this on the ground, or only when you're shaking the h*** 
out of it with the helicopter?  Remember that the notion that a 
helicopter flies aerodynamically is a popular misconception -- 
helicopters fly by vibrating so hard that some components exceed the 
speed of light.  This causes an anti-gravity effect that makes the 
helicopter lift of the ground.  The blades are just there for control. 
This high vibration environment can be bad on electronics, particularly 
if some of their sensitive components are the ones exceeding the speed 
of light.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

Hi,

I'm working on a robot vision system for a helicopter (http://
www.rctoys.com/rc-toys-and-parts/DF-TSU/INDUSTRIAL.html) at
university.

I've just seen for the first time some video captured from the
aircraft's onboard camera. The camera itself is not stabilised and
rather small, the resolution is about 500 lines.

The recorded video was sent from the aircraft to an analogue receiver
via an analogue  radio data link.

In this video there are a few types of noise.

1. "snowy" noise, i.e. little white dots at fairly high spatial
frequency.
2. horizontal straight white lines that appear in one video frame,
then dissapear again.
3. a colour channel in one frame appears higher then the others, e.g.
a white object might flash green in one video frame.


Could anybody tell me what causes these types of noise and if these
phenomena have proper technical names?

My visuall system has worked so far in simulation, but the videos i
have used have not been noisy.

I would like to be able to treat the cause rather than the effect, but
if this is not possible i will have to try and cancel the noise with
further processing.

I think the best thing for the snowy noise is a median filter, but i
also know that would be very expensive in computational terms.

The minimum loss of image quality is essential for this image-
recognition application.

Thankyou very much for any advice offered

Adam