We present the information system that allows us to minimize errors in the iris recognition by varying the settings of iris recognition algorithms. The information system has a modular design that allows replacement of iris recognition algorithms at any stage for improving system performance and recognition quality. We propose a method for improving the iris detection accuracy by changing parameters of the integro-differential operator used for localization of iris boundaries. It is shown that the application of the integro-differential operator along a certain contour allows us to improve the iris detection accuracy. We demonstrate that changing parameters of the Gabor filter for normalized iris image parameterization also improves the quality of iris recognition. Conducted numerical experiments allow us to select optimal parameters for the Gabor filter. It is shown that a number of shifts in a comparison process of iris images binary codes affects the overall performance and recognition error. Thus, multiple shifts significantly reduce the recognition error due to getting invariance to eye rotation during the iris scanning.