89 highest expected or required accuracy of the instrument or compo nent tested. The conversion can be expressed symbolically as: (So -T D„) x: (C, dC) S0 -f- D0. Cs represents the process of conversion and dC the imperfections oï it. S0 is the digital equivalent of S0> and if dC is insignificant D0 represents the digital equivalent of D0. In detective tests the aim of data processing is to separate the regular and random errors as good as possible, to estimate their magnitude and to detect their sources. The corresponding accuracies should also be estimated. The theoretical tools for achieving this are the mathematical models linking the input, the process with undesired disturbances, and the output. The choice of the mathe matical model and the structure of the testing process have been made according to the same considerations. The mathematical model, written in general terms, is: (Si -j- Dt) ijxjPj-f cLP) m (S0 -f- D0) m input process output model model model In the testing of instruments, the input and output models may be regarded as known. The establishment of the model of the process with disturbances is a rather delicate problem. The relevant ma thematical modeling applies to phenomena involved in the flow of the semantic or/and metric information. Generally, the mathematical models should fulfil the following requirements I) they should be versatile thus applicable to a larger group of instruments. Specific models should be simplifications of the general one. II) they should represent the process and the disturbances involved sufficiently accurately. The accuracy should be matched with the expected maximum useful information contained in the input and output models. Ill) they should be adapted to the available means and possibili ties for data processing (e.g. computers). In the ideal case, where the input and the process (including conversion) are perfect, there is a simple relation: Si x Ps S0, or in terms of models: Sim x Psm Som- For linear or "linearised" processes, this expression can be sub tracted from the corresponding one, valid for the imperfect case.

Digitale Tijdschriftenarchief Stichting De Hollandse Cirkel en Geo Informatie Nederland

Tijdschrift voor Kadaster en Landmeetkunde (KenL) | 1967 | | pagina 35