Object recognition using a local adaptive method with digital image data, in which possible objects are selected by creation of binary images that are then used with gray value vectors to identify the most likely objects

摘要

Method for local adaptive object recognition of digital image material, whereby a digitized image is converted into binary code using a number of binary thresholds so that a number of binary images result. Each binary image is subjected to a linkage analysis in order to segment linked areas as possible objects. Characteristics such as gray value vectors and regions of interest are stored for each possible object. To select one of multiple possible objects one ROI is selected from a cluster of ROIs and with it the corresponding possible object. The invention also relates to a corresponding use of the method in industrial applications, especially for object recognition and completeness checking.