Revision history [back]

Hi, I think I have found the answer.

The problem is how to treat the "pixel binning" fields of the Camera Info Messages. The driver I am using actually publishes that the camera has already processed the original image and has used a 2x2 pixel binning to reduce the size of the image alongo with the new (reduced) dimension of the image. The package image_proc however, uses that information once again to generate an empty matrix for the rectified image. i.e.: My camera driver uses Format7 Video Mode 0 to produce a 1024x1024 image (after binning its original 2048x2048 using 2x2 bins).

The problem is that it also publishes the binning information, so

binning_x: 2
binning_y: 2

Which causes some problems for image_proc for it tries to generate a (1024/2)x(1024/2) empty matrix which will be partially filled up with the rectified frame...

See ya