Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[dave w]

OK Let me put this another way. There ain't no "tools",  there ain't no magic formulas. It is all trial and error and that includes masking the detectors. You could use a portable IR source (there are battery video camera lights which are IR, not full spectrum light) to define the borders of the field, but the borders will be true only for the concentrated IR source,  a warm body would produce completely different borders).  And even with masks on the PR511s detector windows, the edge of field detection, and the sensitivity of the detector will not be consistant.

[beelocks]

It also sounds to me like "precision" that I am "requiring" is being over-estimated:  I really only need to segregate the detection areas into about 6-8 "quadrants" around the building, and be able to tell which quadrant motion has been detected in - without getting triggers on two or more sensors for any given motion.  For example, if motion is occuring to the northwest of the building, I need it to trigger only the northwest sensor, and not the north, northwest and west sensors. 

I'm sure that given enough time and motivation it's entirely possible to do almost what you need.

Imagine the purely hypothetical set-up...

All motion sensors are active until motion is detected in one zone. As soon as motion is detected in a single zone all other motion sensors become deactivated. When motion is no longer sensed in the first zone all sensors become active again. Scenario repeats.

This could possibly be achieved with clever use of macros and flags. Sensor one sets a flag so that motion on other sensors is ignored until sensor one no longer sees motion.

Obviously there must be 'blind' space between the sensors for this to occur - the blind space must be fairly significant to allow for heat, light and temperature variations that will naturally occur.


You can see the variance between two seemingly identical motion sensors by mounting them immediately next to each other and running a set of tests to see which one activates and when - the angle and range will differ even on high end motion sensors.


{Rhetorical Question}
What do you want to happen if your motion source tracks around your property (looking in different windows, trying doors, climbing downpipes)?
{/Rhetorical Question}

Intruder approaches from NorthWest (NW motion detected), intruder moves left (N motion ignored), intruder moves further and smashes window on NorthEast side of house as camera system happily records no motion to NorthWest.

Intruder approaches from NorthWest as intruder 2 approaches from North - since you only want motion detection in a single zone that must mean that only one intruder is detected.



If you need to detect movement within the view of the camera and nowhere else, you MUST use the camera to detect the movement.
If you need to vaguely detect motion somewhere near the camera, then you have all the stuff and you just need to set it out as best you can.

There's no polite way of saying it, but although you may think that a video detector is not practical for your application, I think the truth is that your application is not practical unless there's something vital that I'm missing - it makes absolutley no sense to focus motion on a single zone and ignore any and all motion in other zones.

I'll stand aside now and wait to see what happens

[MD Corie]

OK Let me put this another way. There ain't no "tools",  there ain't no magic formulas. It is all trial and error and that includes masking the detectors. You could use a portable IR source (there are battery video camera lights which are IR, not full spectrum light) to define the borders of the field, but the borders will be true only for the concentrated IR source,  a warm body would produce completely different borders).  And even with masks on the PR511s detector windows, the edge of field detection, and the sensitivity of the detector will not be consistant.

If all you say is realistic and typical for motion detection, then the situation is distressing, to say the least.

The variance you claim for borders still puzzles me;  I just can't get my head around how a sensor's performance variances can cause it to "see" beyond the "hard stops" (masks) such that it would produce the varying boundaries that you describe. 

[dave w]

The variance you claim for borders still puzzles me;  I just can't get my head around how a sensor's performance variances can cause it to "see" beyond the "hard stops" (masks) such that it would produce the varying boundaries that you describe. 

The sensors detection sensitivity (performance) remains the same. What the sensor  is detecting (IR emission) is the variation. 

A video camera "sees" farther on a clear day than a foggy day. Apply that axiom to the motion sensor with ambient temperature and body size being the fog. Masking will restrick the width of the field just like zooming a camera lens but the fog is still there.

[MD Corie]

It also sounds to me like "precision" that I am "requiring" is being over-estimated:  I really only need to segregate the detection areas into about 6-8 "quadrants" around the building, and be able to tell which quadrant motion has been detected in - without getting triggers on two or more sensors for any given motion.  For example, if motion is occuring to the northwest of the building, I need it to trigger only the northwest sensor, and not the north, northwest and west sensors. 

I'm sure that given enough time and motivation it's entirely possible to do almost what you need.

Imagine the purely hypothetical set-up...

All motion sensors are active until motion is detected in one zone. As soon as motion is detected in a single zone all other motion sensors become deactivated. When motion is no longer sensed in the first zone all sensors become active again. Scenario repeats.

This could possibly be achieved with clever use of macros and flags. Sensor one sets a flag so that motion on other sensors is ignored until sensor one no longer sees motion.

Obviously there must be 'blind' space between the sensors for this to occur - the blind space must be fairly significant to allow for heat, light and temperature variations that will naturally occur.


You can see the variance between two seemingly identical motion sensors by mounting them immediately next to each other and running a set of tests to see which one activates and when - the angle and range will differ even on high end motion sensors.


{Rhetorical Question}
What do you want to happen if your motion source tracks around your property (looking in different windows, trying doors, climbing downpipes)?
{/Rhetorical Question}

Intruder approaches from NorthWest (NW motion detected), intruder moves left (N motion ignored), intruder moves further and smashes window on NorthEast side of house as camera system happily records no motion to NorthWest.

Intruder approaches from NorthWest as intruder 2 approaches from North - since you only want motion detection in a single zone that must mean that only one intruder is detected.



If you need to detect movement within the view of the camera and nowhere else, you MUST use the camera to detect the movement.
If you need to vaguely detect motion somewhere near the camera, then you have all the stuff and you just need to set it out as best you can.

There's no polite way of saying it, but although you may think that a video detector is not practical for your application, I think the truth is that your application is not practical unless there's something vital that I'm missing - it makes absolutley no sense to focus motion on a single zone and ignore any and all motion in other zones.

I'll stand aside now and wait to see what happens

I think your "lock-out" proposal would be simple to implement, and would be a reasonable work-around if the sensors cannot be prevented from "overlapping" their detection areas.  I was hoping to be able to prevent the overlaps, though.

Your rhetorical scenario is - unfortunately - exactly what occurs under current conditions here - at least in the general context.  (As far as I know, there have been no instances of trying doors and climbing down pipes... but motion frequently moves from sensor field to sensor field in rapid succession... and the macro-based system can't keep up with the movement from zone to zone.  (This was not the case when only the hardware was being used to control the cameras directly, and the slow response is a direct result of the slowness that the macros run, and problems with a new macro trying to run when a previous one is still running - but all that is the topic of a different thread).

Regarding the multiple-intruder scenario, that is indeed a problem, but the inherent limitations of the camera system mean that only one camera can be operating at any particular instant, anyway.  However, the scheme in use says that the most recently-detected motion will activate the corresponding camera in all cases.  If motion were detected in more than one zone exactly at the same time, then PLC collisions would likely result in both motions being missed, as the trigger signals would tromp on each other.  If the triggers occur far enough apart that they don't get stomped, then the first motion would trigger the first camera, and the second motion would then switch the view to the second camera, and so forth.  If the triggers came far enough apart in time that the cameras could come on, then it might be that a glimpse of the earlier motion would occur before the cameras would switch, but in any case, only the last-triggered view would remain in effect.  Obviously, only one area could be observed at length, or else some sort of scan scheme might be implemented, to periodically switch among the active views.  None of this is ideal for observing the entire situation, but then again, the multiple intruders is a relatively unlikely scenario... and since we are already pushing the envelope of capabilities in detecting just one intruder, there's no point in trying to stretch the system beyond that.

As far as the comment about ignoring motion in all but one "zone", I'm not sure where that comes from, because that is not at all my intent.  I merely want to get the right camera to turn on when motion occurs within its field of view, without any desire to suppress other legitimate detections.  The problem I have now is that many of the detectors will trigger on motion that occurs within multiple fields of view instead of only within the field of the associated camera, and thus often results in the wrong camera being turned on.  I'm seeking to prevent that overlap, not to suppress the other sensors.

[MD Corie]

The variance you claim for borders still puzzles me;  I just can't get my head around how a sensor's performance variances can cause it to "see" beyond the "hard stops" (masks) such that it would produce the varying boundaries that you describe. 

The sensors detection sensitivity (performance) remains the same. What the sensor  is detecting (IR emission) is the variation. 

A video camera "sees" farther on a clear day than a foggy day. Apply that axiom to the motion sensor with ambient temperature and body size being the fog. Masking will restrick the width of the field just like zooming a camera lens but the fog is still there.

OK, but my point is that the variations should affect only the depth of view, not the width - which is what I'm concerned with as far as preventing overlap of the sensors.

[systemdm]

MD Corie,

This is not an X10 solution, but have you tried BlueIris.   You can set up your cameras and mask off what areas you don't want to detect motion.  Great piece of software.   I belive they even have a deal with Foscam IP cameras to get a discount on the software.

[beelocks]

OK, but my point is that the variations should affect only the depth of view, not the width - which is what I'm concerned with as far as preventing overlap of the sensors.

Just because you WANT something to work in a certain way, that does not make it so.  

Another analogy for you.
If you wear glasses please remove them and try to detect the exact edges of a person just outside your comfortable range of vision. If you don't wear glasses, please borrow some, put them on and try the same experiment.
Now, close one eye, stand on one leg and ask the blurry person to raise their left arm.
Next turn sideways so that you can see you blurry friend's arm out of the corner of your blurry eye.
Ask them to raise and lower their arm at random.
What you are seeing is approximately what your motion sensor 'sees' at the edge of its detection zone - a general blur that is almost certainly somewhere close to being seen, but sometimes not.
Now try to describe that exact vague location to someone standing just to your right and ask them to take pictures of what they can see.

You cannot change that with masks, you cannot change that with expensive motion sensors. This is the nature of motion sensors. With clever macros and flags you CAN do what you're asking even with overlapping sensors, but only if your subject is co-operative and moves slowly enough for your macros to keep up.


Try some alternative method of keeping intruders away instead of simply noting that they were maybe there by chasin them with a camera system that is inadequate for your needs.

A sign in your yard that states...

Dear Intruder,
I'd really like to know that you were in my yard, but I don't have a surveillance system that quite works the way I'd like it to. If you wouldn't mind slipping a slightly out-of-focus picture of yourself under my front door I'd be really grateful.
Oh, by the way, I have nothing worth stealing. Feel free to look around, but please don't break in. I promise you, there's nothing worth taking; well, unless you think my cat is valuable.
Thanks.

[/b]


Yup, I think that should work.  

[Noam]

From the macros that you posted in the other thread, it looks like you are trying to make a fancy motion-tracking video system out of X10 parts.
I don't think you'll be able to do it, as the components are just not good enough to do what you want. I think you'll need a motion-tracking system that is computer-based, and uses the changing images themselves to detect the motion, not relying on inaccurate sensors to do it.

[MD Corie]

OK, but my point is that the variations should affect only the depth of view, not the width - which is what I'm concerned with as far as preventing overlap of the sensors.

Just because you WANT something to work in a certain way, that does not make it so.  

... be really grateful.
Oh, by the way, I have nothing worth stealing. Feel free to look around, but please don't break in. I promise you, there's nothing worth taking; well, unless you think my cat is valuable.
Thanks. [/center][/b]


Yup, I think that should work.  

Clearly, wanting something to work as it should does not make it so when dealing with these devices...    However, there are certain laws of physics that seem to be violated in these instances, so I'm trying to find the explanation(s) of what otherwise appears to be "heresy".

This whole diversion is why I really hated to expand this from a limited technical question to a full-fledged examination of of my system and devolving into a critique of it, that seems largely based on some twisted assumptions of my needs, my requirements, with stuff from somewhere out in left field thrown in for good measure.

The crux of the issue with an overall system "evaluation" is that as fouled up as it may be for whatever reason, it's all I have to work with - and all I'll likely be able to have - so my objective is to make it work as well as it can... and one problem to be solved is how to segregate the viewing zones of motion sensors.

Even before ever posting this question, I was painfully aware of the limitations of capabilities that these sensors exhibit... so there's no reason to dwell on those.  But, what I fail to understand is why it is supposedly not possible to restrict the field of view of these sensors (by some means), so that they do not react to anything outside of that field - without shutting them down entirely (as seems to be the only alternative brought forth).

The other thing that bothers me is that it would appear that the behavior of these sensors is so random that it cannot be characterized in any way... because if it was possible to characterize their behavior, then it should be possible to make use of that to create a work-around.  Whether that would be a practical, usable solution is another question... I just find it hard to believe that their behavior is so random that there is no way to get a handle on it.

And the kicker is that these devices were marketed to me as being not only capable of doing this job, but actually intended to do it.  Apparently, that was a patently bogus claim.

[HA Dave]

..... And the kicker is that these devices were marketed to me as being not only capable of doing this job, but actually intended to do it.  Apparently, that was a patently bogus claim.

This is the "holiday season" around my house. And every year my sister-inlaw, my nephews wife, and I... take copious pictures at family gatherings. And each year... certain people.... aren't in the pictures. Some people just don't like having their pictures taken (for whatever reason) and they simply avoid the situation. It isn't difficult to merely look away... or just look down.

When outside.... wearing a hat pulled down, or a hoodie, big sunglasses, a hand to the face, keeping the sun on the back of ones head prevents decent video or photos. Everyone knows that. Let me repeat this.... EVERY human being on the planet knows how to avoid having their image being captured by a camera.

Obsessing about a part of a security setup is completely normal. We all do it! I remember reading of a forum member who mounted his security alarm console on his ceiling. His solution to what he reasoned was an access flaw. It normal for people to see the flaws in the systems they create. When we engineer our own systems.... we run the risk of over-engineering them as well.

All systems have weaknesses. We have to accept a certain amount of vulnerability within our setups.

Camera security is the weakest link of any security system. That cannot be changed. Cameras add a level of deterrent. That's it, that's all they do. Sure with the millions of security cameras running 24/7... people are caught... doing all sorts of things. Many people have been prosecuted  with X10 camera images. But I wouldn't make that my goal.

Would be intruders pick their targets based on easiness, risk, and reward. Making your home harder to get into... and with greater risk of being caught... should be the goal. If you've created a "reward rich target" that is another matter entirely (for which you should seek professional advice).

A sign that warns of cameras.... isn't a bad idea. I use a voice announcement as seen in my YouTube video.

[dave w]

OK, but my point is that the variations should affect only the depth of view, not the width - which is what I'm concerned with as far as preventing overlap of the sensors.

Bogus.
You continue to make the same wrong assumptions when it has been clearly explained by multiple people. Why would the width of field NOT be effected?
My fun meter is pegged - I'm outta here.

[beelocks]

The laws of physics are not being violated in any way at all. There are no magical elves, pixies or dwarves living in your sensors that are able to bend, twist or otherwise muck about with gravity, radiance or light (but there may be gremlins - please note that gremlins do not really exist, they are simply an excuse used by people unable to come up with solutions).

You simply do not understand...
1) What a camera system is supposed to do,
2) How motion sensors actually operate,
3) How to integrate one with the other,
4) How to listen when you are told the same thing multiple times,
5) All of the above, or
6) Gremlins

If I knew how to put a poll on this I'm sure you would get an overwhelming vote for 5)

Now, if you don't mind, I'm taking my ball and going home. 

[MD Corie]

OK, but my point is that the variations should affect only the depth of view, not the width - which is what I'm concerned with as far as preventing overlap of the sensors.

Bogus.
You continue to make the same wrong assumptions when it has been clearly explained by multiple people. Why would the width of field NOT be effected?
My fun meter is pegged - I'm outta here.

I'm the one making wrong assumptions???  Wow!  That's rich... considering how many wrong assumptions people have been attributing to my system, on basis that I have no clue about.  For example, "security" against criminals is, at best, a side benefit of my system - not its raison d'etre - so whether or not it can accomplish that shouldn't be the main issue here... yet people have somehow decided by some sort of concensus that it is indeed why my system is in place.  Talk about wrong assumptions!

Be that as it may, let my try (once again) to explain the laws of physics that I'm referring to.  Because people seem to like analogies, here's mine, using a camera as an analogy for a motion sensor:

Let's say a camera is installed, having a 60-degree wide field of view.  Under optimum environmental conditions, this camera "sees" whatever exists within that 60-degree field of view, within the constraints of the camera's capabilities.  (For instance, a low-res camera might not be able to distinguish objects smaller than, say, 1' tall at distances beyond, say, 30' from the camera... but can distinguish objects larger than, say, 5' tall at distances closer than, say, 80' from the camera).  Now, the camera's characteristics or optical factors may prevent it from "seeing" some or all such objects when certain adverse environmental conditions exist - such as rain, snow, fog, darkness, etc. (that might impair the camera's ability to view part or all of it's optimum field of view), or even temperature extremes (that might impair the operation of the camera itself, preventing it from "distinquishing" all objects within of its usual field of view).

The camera's physical width of field (and therefore the effective boundaries of that field, as imposed on the surrounding terrain) is determined by the physics of the detector and optics of the camera, and therefore will not change (barring any physical deformation of the camera's optics due to extreme temperatures or whatever).  Variations of other environmental conditions might affect how far away (depth of field) the camera can see objects at any given time, but will not affect how wide the field of view is, because that remains limited by the physical constraints of the optics.  Further, the maximum width of view could be narrowed by placing optically-opaque objects near the camera on the sides of its view.  Again, environmental conditions may affect how far the camera can see effectively while these objects restrict the width of view, but the environmental conditions will not alter the width of view.

If we can agree that a camera is a fair analogy of a motion sensor (as was supposed by others earlier in this thread), then let me use this analogy in an attempt to make my point:

Let's say there are two cameras mounted such that their width of views partially overlap at some distance from the cameras, making it possible for both cameras to see the same objects while those objects are in the areas of the terrain where the views of the two cameras overlap.  If it is desired to prevent both cameras from viewing the same objects simultaneously, then the camera(s) would need to be adjusted away from each other, or else some blocking object(s) would have to be imposed on the adjacent side(s) of one or both cameras' field of view, so that no overlap of views would exist.  Once configured this way, no amount of normal environmental variations would alter the physical limits of the views, to somehow re-impose an overlap condition.  Same thing applies to motion sensors:  Affects of environmental variations are not going to magically overcome the physical characteristics and expand the width of view so that the sensors' fields of view would again overlap, as some people are claiming.

In order to set up the two cameras so that they view adjacent areas of the terrain without any overlap would require trial and error, by adjusting the cameras and/or any blocking objects and looking at the on-screen image to find out where its boundaries fall in relation to the viewed terrain.  Presumably, one would make these observations under reasonably optimal environmental conditions, yet it would be possible to make them under adverse conditions, even in total darkness, by taking some measure to "enhance" the spot being "mapped" - for instance, by placing a small bright light at the spot of interest, or by moving such a light into the field of view from the side until it shows up on the screen.  The spot in the terrain where this occurs can then be "mapped" as the edge of the viewed area.  Similarly, with motion sensors, it should be possible to "map" the physical sides of the field of view by using some IR "spot source" that is intense enough to overcome any "weak" detection due to adverse environmental conditions, and moving it towards the field of view from the side until it comes into the physical view of the sensor... but the concensus here has been that this is not possible under any circumstances.  I find this hard to believe because it would mean that the motion sensors must "detect" only at random - in which case, they would not be sensors, but just noise sources.

My question all along has been what can be used to provide an intense "spot source" of IR that will definitely trigger a sensor when it moves within the sensor's field, or if there is a better technique that would ensure detections (like maybe how fast or how far to move the "target")... because everything I've tried so far does not seem to have enough "umph" to ensure that the sensor will "see" it move.  I fail to understand why any actual or assumed application of my sensors has any bearing on this... at least not beyond establishing the ambient conditions that it would be used in - yet that seems to be the direction this discussion has been pushed into.

[beelocks]

OK. I got it now.

I'll agree with you. Physics is broken. 

Call your local reality inspector and have him run a standard testing sequence to work out exactly why radiant heat patterns do not work the same way as visual light patterns in your locale. I'm sure that when you speak to him he'll bring several friends with white jackets. It's likely his friends will bring you a special coat with the fastenings down the back and extra long sleeves with strings on the ends 

Last time I needed to get reality adjusted, it got pretty expensive - you may find it a lot less expensive to just move.