cmucam3-hardware (#1) - Frame differencing problem (#193) - Message List

Frame differencing problem

Hi, I am currently working on a project and having a little difficult time with using the frame differencing method. I am trying to use the FD (frame difference) to detect an object that is moving within the camera view. Even though the camera response with the T packet after the command was sent, it does not seem to have the right response. There was only a little different as the object was moving in front of the camera. All I got was the T packet of this format "T 4 4 1 1 8 8 ..." with the value of 4 4 fluctuating just a little bit. I did try to use the LF command to load the frame but it did not really help out. It looks like the return response of the FD command shows only the x and y coordinate pixels of only the first 8x8 grid instead of the whole 88x143 resolution view.

Does anyone has the same problem that I am having? If you have figured out or know how to use the frame differencing (FD) command to detect the movement, please give me some tips and/or ideas on what is needed to be done to get the "correct" result, i.e. having the return response shows the x and y pixel coordinate based on the whole view.

I would really appreciate the help.

Thank you in advance

DK

  • Message #489

    Frame differencing on the CMUcam2 operates on 64 regions of the image (8x8). In hi-res mode on the CMUcam3's emulation of the CMUcam2, it operates on a 16x16 images. You can modify that code to increase the resolution if you want.

    The values you are seeing in the above example basically indicate that the camera sees the entire image as changing.