Under the state of the art technolgoy, pattern recognition is not a science, it's procedures are neither universal, nor fundamental - more.
Based on a physics theory of information we are able to make our programs understand what they are seeing, in a way similar to what a human being is doing (as described by Rizzolatti et al.).
This display is created by a program, which recognizes human beings. The program does not use any motion detection. To the best to our knowledge, this is the first program able to recognize human beings (that is to identify them without motion detection) in a natural environment.
Note: this prototype program was created with an effort of about one man-year of work.
Also from a technical point of view, our programs are different from the state of the art: they perform only very simple operations, there are no probability functions and no decision making => linear computing. A detailed discussion can be found here.
Note, that linear computing is in particular suited for the CELL processor (IBM). We are in the process of implementing the general man identifier program on the CELL processor.
Isomorph work in progress: