Active appearance model: Difference between revisions

Revision as of 18:35, 2 August 2016

Revision as of 13:28, 18 February 2019

Line 1:

An '''active appearance model''' ('''AAM''') is a [[computer vision]] algorithm for matching a [[statistical model]] of object shape and appearance to a new image. They are built during a training phase. A set of images, together with coordinates of landmarks that appear in all of the images, is provided to the training supervisor.

The model was first introduced by Edwards, Cootes and Taylor in the context of face analysis at the 3rd International Conference on Face and Gesture Recognition, 1998.<ref>{{Cite book | last1 = Edwards | first1 = G. J. | last2 = Taylor | first2 = C. J. | last3 = Cootes | first3 = T. F. | doi = 10.1109/AFGR.1998.670965 | chapter = Interpreting face images using active appearance models | title = Proceedings Third IEEE International Conference on Automatic Face and Gesture Recognition | pages = 300 | year = 1998 | isbn = 978-0-8186-8344-92 | pmid = | pmc = | citeseerx = 10.1.1.33.1784 }}</ref> Cootes, Edwards and Taylor further described the approach as a general method in computer vision at the European Conference on Computer Vision in the same year.<ref>{{Cite book | last1 = Cootes | first1 = T. F. | last2 = Edwards | first2 = G. J. | last3 = Taylor | first3 = C. J. | doi = 10.1007/BFb0054760 | chapter = Active appearance models | title = Computer Vision — ECCV'98 | series = Lecture Notes in Computer Science | volume = 1407 | pages = 484 | year = 1998 | isbn = 978-3-540-64613-26 | pmid = | pmc = | citeseerx = 10.1.1.374.7954 }}</ref><ref>{{Cite journal | last1 = Cootes | first1 = T. F. | last2 = Edwards | first2 = G. J. | last3 = Taylor | first3 = C. J. | doi = 10.1109/34.927467 | title = Active appearance models | journal = IEEE Transactions on Pattern Analysis and Machine Intelligence | volume = 23 | issue = 6 | pages = 681 | year = 2001 | pmid = | pmc = | citeseerx = 10.1.1.128.4967 }}</ref> The approach is widely used for matching and tracking faces and for [[medical imaging|medical image interpretation]].

The algorithm uses the difference between the current estimate of appearance and the target image to drive an [[Optimization (mathematics)|optimization]] process.