The problem of the most exact determination of the observed signal position from noisy experimental data distorted by a measuring device is considered on one of the most typical example of the measurements of an elementary particles that passes a discrete detector producing an electron shower. Then during the process of registering the signal is to be analog-to-digit converted and stored as a histogram. The basic problem is: to reconstruct the original signal position and other its parameters from the registered histogram. However, such hindering factors, as the signal distortions due to the digitizing process, the presence of background noise and signal overlapping due to the high occupancy of the majority of modern experimental systems, demand to combine two statistic approaches: parametric and non-parametric ones, in order to find an efficient solution of the problem.