A study of validity of stochastic models for sequential image processing procedures

摘要

We have proposed stochastic models of sequential image processing procedures for detecting abnormal shadows in medical images and developed automated procedure construction process by using this models. This paper evaluates the validity of the stochastic models for three kinds of the sequential image processing procedures which consists of three local processes, "filtering and binarization", "intermediate decision" and "final decision". At first, the validity of assumptions introduced in the modeling process is tested. Secondly, we evaluate the validity of the equations which show relationships between performances of local processes and that of the whole procedure.