Unless you want to devise your own audio and video encoding/decoding routines, you'll probably want to use some existing routines. They do exist, in Linux world. So, you need a microcontroller with memory management and sufficient throughput and ROM/RAM to run Linux, unless you want to write your own routines. For a Linux capable processor, you could google for that, to identify the microcontroller you want. Then, you need software that can handle video calling - you could write your own, or you could use standards based methods like SIP. So, yes it is possible (since such devices exist), and you need some experience in developing Linux based applications.