System and method for identifying features in the digital image based on the rule extraction

摘要

Method for classifying features in a digital medical image is a step of providing a plurality of feature points in N-dimensional space, which is an element contained in one of the two sets, each feature point is, the step of providing and, determining the classification plane is separated from the feature points in the set of second in a set of two feature points in the set of the first one of the two sets, and converting the classification plane The step of a (32), has a positive component, the feature regions for one or more feature points included in the set of one, the normal vector to the classification plane is the conversion of the n a unit hypercube in the space in the converted having an axis, and wherein the step of converting, to obtain a upper bound along each of the axes of the n unit hypercube and (33), and inverse transforming the upper bound The step of removing from the set of the one, to obtain a new rule with the one or more feature points included in the set and the one (34), feature points encompassed by the new rule above (35) I include a.