The technology used here to do the image processing is GLSL, in particular fragment shaders (aka pixel shaders). GLSL is a very small C-like language that's become a standard for GPUs. GLSL code gets sent (as a string) to the GPU by javascript via the WebGL API.

Seriously is a JS library for handling the boilerplate of WebGL, composing and compositing multiple shaders in a pipeline/graph, and adjusting their parameters. In addition it comes with a bunch of pre-written shaders.

Shaders themselves are a lot of fun to write, IMO. A pixel shader is just a function that computes a color given an input pixel location. (A shader can also take in additional input such as "textures" to sample from, e.g. a video frame.)

The shader is run with massive parallelization in the GPU. In theory every pixel can be processed simultaneously. This is how these effects can run in real time.

Here are some more examples of what you can do with pixel shaders (including sampling from the webcam in the browser -- should work in Chrome and FF), http://pixelshaders.com/examples/

I'm in the process of writing an interactive book about pixel shaders, http://pixelshaders.com/