A pattern recognition system should be composed of a sensor that gathers the observations to classify such a digital camera, a feature extraction system that transforms the information captured in numerical or symbolic, and a system of classification or description of the information measure. A pattern recognition system must have a sensor in charged of getting the data to classify for example a digital camera, a characteristic collector system that changes the captured information to numeric or symbolic values, and a classification system of the measured information. In this chapter we focus on the statistical classification. Its main classifier is based on probability theory. It is historically the first approach and that there was probably the most developed, but it does not mean that its development and research has been finished yet. We begin with the moments used in statistics and which are basic, which can be used to find patterns within a data base by applying the statistical patterns in the recognition of particular facial expressions in digital video processing. Despite the good results that give us the recognition of patterns using the moment , these are not enough in many cases, so we need to use other mathematical operations such as correlation and covariance as well as other concepts derived from these equations. Most of the processes studied have historically been analyzed in an instant of time, but I really did not occur in a second. Any experiment is deterministic; all have a degree of uncertainty. Researchers have developed a theory of probabilities, although again apply with respect to a viewpoint determistico. On why we must return to stochastic processes and add a clear perspective toward randomized experiments. Although there are many phenomena which can be applied the probability, we will focus on the affective computing, for the recognition of facial expressions, because it is a process that must be treated with this efoque. To accomplish this, first locate the face and then the characteristic points of the face; Once done, apply the statistical analysis with the expectation and variance in a process that varies in time, and therefore change these values. These functions can be characterized with Gaussian or Rayleigh. Although results are promising. It is evident that with increasing the number of stored data, requires more resources or more efficient implementation of algorithms for analysis and interpretation of data. 24