Image segmentation by assigning classes to adaptive mesh primitives
AssigneeKoninklijke Philips N.V.
InventorsJens Von Berg, Michael Kaus
Life of this patent
Estimating the term…
Abstract
Quantification of metric or functional parameters often requires image segmentation. A crucial part of such method is the model of the surface characteristics of the object of interest (features), which drives the deformable surface towards the object boundary in the image. According to the present invention, sections of the mesh are assigned to different classes for different features. According to the present invention, the assignment of mesh sections to the classes is adapted by using actual feature information from the unseen image. Advantageously, this allows for an adaptation of the feature category to which the mesh section is assigned and thereby allows an improved segmentation of the object.
Claims20
A method of segmenting a selected region from a multi-dimensional dataset, via a computing device, wherein an adaptive mesh comprising a plurality of mesh primitives is to be adapted to the selected region, the method comprising the steps of:
- sampling data from the multi-dimensional dataset;
- estimating a probability, via the computing device, that a mesh primitive of the plurality of mesh primitives belongs to a class of a plurality of classes by using the sampled data from the multi-dimensional dataset by means of an estimator, wherein the probability is a function of a frequency of occurrence of the mesh primitive within a training set;
- and
- applying an energy function corresponding to the class of the plurality of classes, via the computing device, to adapt the mesh primitive of the plurality of mesh primitives to the selected region.
A method of assigning classes to mesh primitives of an adaptive mesh, via a computing device, for segmenting an object from an image, the method comprising the steps of:
- estimating, via a computing device, a probability that a mesh primitive of the mesh primitives belongs to a class of the classes by using a feature probe from the image by means of an estimator, wherein the probability is a function of a parametric density function used to model data from a training set;
- and
- assigning, via a computing device, a class to the mesh primitive on the basis of the estimated probability.
A method of generating a shape model for segmentation of an object from a multi-dimensional dataset, via a computing device, wherein the shape model includes an adaptive mesh with a plurality of mesh primitives to be adapted to the object, the method comprising the steps of
- (a) determining, via the computing device, an a priori probability that a mesh primitive of the plurality of mesh primitives belongs to a class of a plurality of classes relating to features of the object, wherein the a priori probability is derived from a frequency of occurrence of the mesh primitive within a training set;
- (b) estimating, via the computing device, a class conditional probability density function for a feature vector at a position in the multi-dimensional dataset for the plurality of classes by using data from a training model;
- (c) determining, via the computing device, an a posteriori probability that the feature vector belongs to the class of the plurality of classes.
An image processing device comprising:
- a memory for storing a multi-dimensional dataset;
- and
- a calculation unit for segmenting a selected region from the multi-dimensional dataset, wherein an adaptive mesh comprising a plurality of mesh primitives is to be adapted to the selected region, the calculation unit being constructed to sample data from the multi-dimensional dataset;
- estimate a probability that a mesh primitive of the plurality of mesh primitives belongs to a class of a plurality of classes by using the sampled data from the multi-dimensional dataset by means of an estimator, wherein the probability is derived from a frequency of occurrence of the mesh primitive within a training set;
- and apply an energy function corresponding to the class of the plurality of classes to adapt the mesh primitive of the plurality of mesh primitives to the selected region.
An image processing device comprising:
- a memory for storing an image;
- a calculation unit for:
- assigning classes to mesh primitives of an adaptive mesh for segmenting an object from the image;
- estimating a probability that a mesh primitive of the mesh primitives belongs to a class of the classes by using a feature probe from the image by means of an estimator, wherein the probability is derived from a calculation of a parametric density function used to model data from a training set;
- and assigning a class to the mesh primitive on the basis of the estimated probability.
A computer readable storage medium encoded with computer executable instructions for an image processing device for assigning classes to mesh primitives of an adaptive mesh for segmenting an object from the image, the computer executable instructions, when executed by a processor cause the process to perform the following steps:
- estimating a probability that a mesh primitive of the mesh primitives belongs to a class of the classes by using a feature probe from the image by means of an estimator, wherein the probability is derived from both a frequency of occurrence of the mesh primitive within a training set and a calculation of a parametric density function used to model data from the training set;
- and
- assigning a class to the mesh primitive on the basis of the estimated probability.
Dependencies are inferred from the claim text.