Recognition of image object with different picture size elements

The problem of textual image recognition when some symbols have different picture size is considered. The author presents a new statistical image object recognition method based on the use of Gaussian mixture densities in the context of the Bayesian decision rule. In addition to this, the methods based on the distance between images and classes are examined. To assess the threshold in the measuring method of the correspondence between partitions of a finite set of objects, a solution based on mixture distribution division is presented. The ideas are illustrated with an example of real text recognition procedure.