Development of the Statistical Pattern Recognition Methodology and Its Application to Quality Control

摘要

Various aspects of statistical pattern recognition are reviewed. In the area of feature selection, feature set search algorithms for selecting a subset of pattern descriptors from a set of measurements are comparatively studied. Decision making systems are then considered, particularly as regards measurement errors and the subsequent increase in the error probability of the system. In the realm of pattern classification, the problem of designing a pattern recognition system for processing incomplete pattern vectors is discussed. Next, it is shown that the probability distribution of empirical error count estimate is not binomial and a recursive formula for determining this distribution is given. Finally, pattern recognition techniques are applied to the problem of automatic inspection of products by lots. Alternative approaches to acceptance sampling plan design are proposed.