Maximum CMU width range

Does anyone know the amount of pixels in the x axis of the camera(servos not moving)? From what I understand, Easyc has those values, but we’re using MPLab.
Thanks. :slight_smile:

The answer is in Kevin Watson’s “Bells and whistles” version of the camera code, in tracking.h:

// parameters for CMUcam2 with OV7620 camera module
#define IMAGE_WIDTH 159
#define IMAGE_HEIGHT 239

Those are in pixels.

Happy hacking,


Also, you get about 60 degrees of sight in the wide direction.

Hmm, I think I measured it to be about 50 degrees last week…back to the drawing board for me…:frowning:

Then again, my FOV measurements suggested that the longer dimension (the height) has a narrower FOV than the shorter dimension (the width). I went over my math a few times and had someone else check it over, so I’m pretty sure my measurements are OK.

I don’t know; I’ll take another look at it either tonight or tomorrow night.


Huh, we had it 46.6 degrees…:confused:

Thanks again! :smiley:

Does anyone know the exact(± 5 degrees) width range?

It probably depends a bit on how you measure the view angle.

And yes, the pixel dimensions 160x240, makes the camera’s field of view taller than it is wide - unlike a TV screen.


Check again. The width of 160 is in “half resolution” mode, with each reported pixel being twice as wide as it is tall. The actual aspect ratio is a perfectly TV-compatible 320x240.

I stand corrected. That is not mentioned in tracking.h. You found that in the CMU manufacturer’s specs?

So does anyone have the estimate? I’m not quite sure what the resolutions tell about the camera’s width view.

Just to avoid some confusion here, Alan is correct when he says the imager used in the CMUcam2 is larger than the values found in tracking.h. To get reasonable update rates from the camera, the microcontroller down-samples the image to 160 x 240 and then does all operations on that smaller image.


Which suggests that vertical, say height, calculations are more accurate.