Alignment by Maximization of Mutual Information

By Paul A. Viola, William M. Wells. 1995

A new information-theoretic approach is presented for finding the pose of an object in an image. The technique does not require information about the surface properties of the object, besides its shape, and is robust with respect to variations of illumination. In our derivation few assumptions are made about the nature of the imaging process. As a result the algorithms are quite general and may foreseeably be used in a wide variety of imaging situations.Experiments are presented that demonstrate the approach registering magnetic resonance (MR) images, aligning a complex 3D object model to real scenes including clutter and occlusion, tracking a human head in a video sequence and aligning a view-based 2D object model to real images.The method is based on a formulation of the mutual information between the model and the image. As applied here the technique is intensity-based, rather than feature-based. It works well in domains where edge or gradient-magnitude based methods have difficulty, yet it is more robust than traditional correlation. Additionally, it has an efficient implementation that is based on stochastic approximation.

Subfields: Computer Vision and Pattern Recognition, Information Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Information Theory

Citations in the corpus

SIFT Flow: Dense Correspondence across Scenes and Its Applications (2011, methodology): “However, these methods still typically require objects to be salient, similar, with limited background • Ce Liu is with Microsoft Research New England, One Memorial Drive, Cambridge, MA 02142.”
Evaluation of Cost Functions for Stereo Matching (2007, unknown): “The third ﬁlter is a mean ﬁlter , which aims to compensate powerful method handling complex radiometric relationships between two images [20].”
A robust method for registration of three-dimensional knee implant models to two-dimensional fluoroscopy images (2003, background): “A variety of image difference measures can be used [20], such as pattern intensity [19], gradient difference [11], and cross-correlation [10].”
Unsupervised End-to-end Learning for Deformable Medical Image Registration (2017, background): “Though in some cases parametric transformation models (such as rigid, affine and B-Splines transformation) are enough to recover the underlying deformations [33], [38], a more flexible non-parametric transformation model allowing for arbitrary local deformations is usually needed [4], [5].”
Deformable medical image registration: setting the state of the art with discrete methods. (2011, background): “(29) we need to determine the transformation under the”
Image registration methods: a survey (2003, methodology): “[201] maxima of MI was found using the gradient descent optimization method.”
An integrated visualization system for surgical planning and guidance using image fusion and an open MR (2001, methodology): “The 3D Slicer supports manual rigid registration as well as automatic registration by maximization of mutual information (MI) (23,24).”
Image change detection algorithms: a systematic survey (2005, background): “The idea is related to mutual information measures used for image registration [53].”
Class-Based Feature Matching Across Unrestricted Transformations (2008, background): “A scheme that can cope with complex intensity transformations was described in [18].”
elastix: A Toolbox for Intensity-Based Medical Image Registration (2010, background): “tion (MI) [26], [44]–[46], and normalized mutual information”
A Review of Geometric Transformations for Nonrigid Body Registration (2008, methodology): “These consisted of rigid-body [150], [151] and nonrigid registration algorithms and the proportional squaring method of Talairach [40] which requires manual input.”
Cumulative residual entropy: a new measure of information (2004, unknown): “The most popular cost function presented in literature has been the mutual information (MI), , between the two given images which is maximized over the class of transformations mentioned earlier [18], [3].”
PET-guided delineation of radiation therapy treatment volumes: a survey of image segmentation techniques (2010, methodology): “Mutual information This is a measure of uncertainty (entropy), which is routinely used with registration algorithms [196]; however, the metric is applicable to segmentation analysis as well [197].”