wuulsoftware

i want to know how i can correct this problem: when the luminosity appears(like a sun), the sw detect like a movement...

somebody can help me?

tks!

That's a very difficult problem - the program works by comparing one camera frame to the previous frame and working out the percentage of pixels that have changed, so if the sun suddenly appears from behind a cloud and lights up the room, a good percentage of pixels will suddenly change colour which can trigger movement detection.

I can't think of any simple way around this I'm afraid, other than to try to minimise the amount of natural light that the camera is exposed to, e.g. not positioning the camera near a window.

You could try increasing the movement trigger threshold so that it is less sensitive, but this may result in actual movements not getting detected. You will have to play around with the setting to find the best value.

Sorry I can't be much more help - if anyone else has any suggestions please share them.

OK!

I´m trying to know the best value to set in the movement trigger, without lose in movement detected.

Thanks!!!

I'm having the same problem. I have an idea I would like to test - but am stuggling to get Turbo Delphi up and running

The idea is to add another panel. You have to place the panel on a place on the image where you are almost certain there won't be real movement.

Then I want to automatically adjust the slider to the percentage of "movement" in this panel plus 1 or 2 percent. This way the setting for movement is automatticaly adjusted for distortion..

Maby someone else had a working Delphi and can get this Idea on the road?

Have you tried defining the border area that the program examines? By default it scans the entire image for movement, but if you click the grid icon at the top you drag an area to monitor, and anything outside this will be ignored.

Drag the blue border anywhere on the screen to move it, to change size hold SHIFT down while dragging.

Yep, I tried it. But the problem is that in the rectangle there is 'movement' from webcam-noise when it's becoming dark. Don't exactly know why, but after sunset the standard percentage of difference between images is going up from 0 to 15% because of this noise.

So my idea was to declare another frame in which I dont expect any movement. By moving the movement-treshold by the percentage of movement in this frame I would be able to detect real movement in the main frame.

But I still don't have a Delphi in which to check out my idea So many plans so little time

Do you mean use a "base-image" that every image is compared to? If so, that feature is already present (can't remember exactly how to configure it as I haven't run it for ages but I remember it's pretty obvious).

No, the idea is to use a part of the image from the webcam as a indicator for the amount of distortion of the image.

Why? When I point my cam to my car. And I select a square around my car to protect the car around sun-down the webcam-images start to get noise in the images which worsens during the night. After sunset the image is distortion free. So when I set the movement indicator at 10% after sunset moveaction starts making images until sunset... Just because the distortion leads to a difference between images of more than 10%

What I want to do is to select another part of the same image, where I don't expect any movement at all. And use that part of the image to calculate the amount of distortion. Simply by comparing the pixels in that part of the image with the previous image, as you do with the detection zone.

Then I want to add the percentage of difference in this "reference zone" to the threshold for movement-detection. So when I set the threshold to 10% and there is 15% of distortion the threshold would move to 25% This way I hope that after midnight moveaction will correct for the webcam-noise.