Method for the three-dimensional detection of objects

A method for the three-dimensional detection of objects, the method comprising the steps of:

• detecting a scene by means of an image detection device, wherein a camera image is produced;
• illuminating the scene with a pattern, with the pattern being irregular at least locally, wherein an image is produced as a reference image in accordance with the pattern;
• fixing a correlation window as a section of the camera image;
• fixing a plurality of search windows as sections of a taught reference image, wherein different search windows reproduce different regions of the reference image;
• determining the search window which corresponds to the correlation window by means of a correlation process,
• applying a binarization scheme to the correlation window and to the search windows in the correlation process, in which the respective points of at least one randomly selected point pair are compared to one another within the correlation window and within the search windows, with a respective resultant vector being formed for the correlation window and for the search windows;
• carrying out a correlation of the resultant vector of the correlation window and of the resultant vectors of the search windows in order to determine the search window which corresponds to the correlation window;
• and
• selecting the at least one randomly selected point pair by means of a plurality of iterations, with the at least one randomly selected point pair being reselected randomly for each iteration and with that at least one randomly selected point pair being used for the correlation process which achieves the best association between the correlation window and the search window which corresponds to the correlation window.