A LEARNING METHOD FOR AN IMAGE ANALYSIS SYSTEM FOR USE IN THE ANALYSIS OF AN OBJECT AS WELL AS USES OF THE METHOD

摘要

In the teaching of an image analysis system, a video image of a reference object is captured. Several areas, each of which represents a class (22, 23), are marked on the video image, which is divided into pixels having their separate identification, such as colour. Once all classes have been formed, each class is given a colour which is assigned to all pixels in the class irrespective of identification. Pixels which may belong to more than one class, a so-called conflict class (24), and pixels which do not belong to any class, a so-called 0 class (25), are transferred to a residual class. Then the entire video image is processed in that the pixels which are not represented in the zero class (25) or the conflict class (24), are given a colour value from the class having a pixel identification which is closest to the pixel concerned. The principles of the invention may be applied in general in connection with analysis techniques where objects may be characterized by means of an N-dimensional set of parameters. This provides a method which can teach an image analysis system in a quick and safe manner for use e.g. in connection with analysis of cells/tissue, analysis of cuts of meat, metallurgical analyses and other similar fields.