Apparatus for segmenting an object comprising sub-objects

A method of segmenting an object comprising sub-objects shown in an object image (18), the method comprising the following steps:

• generating a feature image (19) showing features related to intermediate regions between the sub-objects by means of a feature image generation unit (2),
• segmenting the sub-objects, using the object image and the feature image, by means of a segmentation unit (3);
• wherein segmenting the sub-objects comprises:
• providing a model object comprising model sub-objects, pre-positioning the model object in the object image and the feature image, and
• adapting the model object to the object in the object image and in the feature image;
• wherein adapting the model object is performed such that a surface of a model sub-object facing another model sub-object is adapted to the feature image, and such that a surface of the model object which does not face a model sub-object is adapted to the object image.

A nontransitory computer-readable medium having embodied thereon a computer program for segmenting an object comprising sub-objects shown in an object image (18), the computer program comprising program code means for, when the computer program is run on a computer, carrying out the steps of a method of segmenting an object comprising sub-objects shown in an object image, the method comprising the following steps:

• generating a feature image showing features related to intermediate regions between the sub-objects by means of a feature image generation unit,
• segmenting the sub-objects, using the object image and the feature image, by means of a segmentation unit;
• wherein segmenting the sub-objects comprises:
• providing a model object comprising model sub-objects,
• pre-positioning the model object in the object image and the feature image, and
• adapting the model object to the object in the object image and in the feature image;
• wherein adapting the model object is performed such that a surface of a model sub-object facing another model sub-object is adapted to the feature image, and such that a surface of the model object which does not face a model sub-object is adapted to the object image.