[Unresolved] PixelCheckSum Functionally help

[Negative1]

Hi all,

Little background on my script:

I'm looking at a Black and White live camera image to detect a change using pixelchecksum. However the returned value is always different even if there isn't a change detectable to a human eye.

I have guessed that the return value is larger the more "white" in the image (keep in mind it's a black and white image). Is this correct? If so My plan is to save the pixelchecksum value to an array 50 times or so and remove the 10 smallest and 10 largest values then average them and use that value to determine how much white is in the screen. I only need to know the point in which the camera image is the brightest (it's going to move like a bell curve and I need to find the peak point) Also its not a solid uniform intensity image.

Any help with pixelchecksum or the logic of how I plan to use it would be appreciated. Please also if you have a better idea of how to determine the peak brightness that would be great too.

Thanks all

-1

Edited April 12, 2010 by Negative1

[Makaule]

If you are trying to compire WebCam view with PixelCheckSum, then forget it. The thing is that you wont get 2 identicaly views and right after it detects not same part it tells that "something" had changed. You would need to use other functions to get what you need. Instead of PixelCheckSum, you may try use PixelSearch.

By the way, how did you make you camera to work in Black & White mode? Did you make it with some command in AutoIT?

Edited April 12, 2010 by Makaule

[Negative1]

If you are trying to compire WebCam view with PixelCheckSum, then forget it. The thing is that you wont get 2 identicaly views and right after it detects not same part it tells that "something" had changed. You would need to use other functions to get what you need.

By the way, how did you make you camera to work in Black & White mode? Did you make it with some command in AutoIT?

Thanks for the reply but it is a infrared camera since there is no color in the infrared spectrum it appears as black and white.

[AndyG]

Hi,

because of the Image Noise it is nearly impossible to get two exact the same images from a digital camera. The CCD Sensor produces some "random" information depending of electrical voltage and temperature. 

You may try to count the number of the black or white "pixel" in different areas of the image to "calibrate" the image (with no visible movement).

Example: Area01 has 10x10 Pixel, the "Noise" produces from (minimum) 25 to 50 (maximum) black pixel without any motion infront of the camera. If there were suddenly 70 black pixel in this area, and something similar is going on in Area02 and Area03 (where are next to Area01) then a probable motion infront of the camera could be detected.....maybe....

Andy

[Makaule]

Yes, AndyG, you are almost right, but if you add little tolerance value you would get what you need (Tolerance value 5-15 should be enough).