this is an IP camera daemon that reads video frames from the camera and publishes them to connected HTTP clients. You can find the source code at https://github.com/rhuitl/uClinux/blob/ipcamd/user/ipcamd/
For the HTTP server part, libmicrohttpd is used.
A simple motion detection algorithm is implemented and runs fast enough to keep up with the frame rate. It works by decoding a sub-sampled version of the image (80x60 for 640x480 JPEG frames) and creating a moving average background image. Image frames can then be compared against the background to detection motion. Right now no action is performed when motion is detected, but depending on the application one could trigger an alarm or suspend the HTTP transmission (sending only keep-alives) until motion is detected.
A very experimental feature is SSL signing of a video frame's hash value. This way one could make sure no one is tampering with the video frames as they are transmitted wirelessly. AFAIR the signing takes a few hundred miliseconds, so that might be a bit too slow.
Please check the commit log to find out how to setup your uClinux tree to build ipcamd. You'll also have to add libmicrohttpd to the lib/ directory (see https://github.com/rhuitl/uClinux/commits/master/lib/libmicrohttpd
Command line options:
-d | --device name Video device name [/dev/video]
-h | --help Print this message
-m | --mmap Use memory mapped buffers
-r | --read Use read() calls
-u | --userp Use application allocated buffers
-p | --private Use private mapping (use with -m)
-c | --count Number of frames to grab
-w | --width Set width (640 or 320), height is auto-selected
-j | --decode-jpeg Decode DC coefficients of JPEG, repeat to save as /tmp/test.pgm
-s | --sign-jpeg Sign SHA1 hash of JPEG data
-D | --dhcp Start dhcpcd to handle network setup
-v | --verbose Be verbose, e.g., print video frames
If it works with your webcam driver, you should use mmap() (-m instead of -r or -u) in order to minimize copying of image data.