Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
research.imlop [2014/07/06 19:25]
sbillin3@johnshopkins.edu intranet:research:icop renamed to research.imlop (moving from intranet to public research page)
research.imlop [2019/08/07 12:01] (current)
Line 1: Line 1:
-====== Iterative ​Closest ​Oriented Point Registration (ICOP) ======+====== Iterative ​Most Likely ​Oriented Point Registration (IMLOP) ======
  
 ===== Project Overview ===== ===== Project Overview =====
-The iterative closest point (ICP) algorithm is a popular method for registering geometric representations of 3D shapes and has been extensively applied to problems of rigid body shape alignment in medical research. ICP-based algorithms seek optimal alignment by iterating two key steps: a correspondence phase that computes ​closest ​point pairs and a registration phase that computes the transform to align the point pairs.+The iterative closest point (ICP) algorithm is a popular method for registering geometric representations of 3D shapes and has been extensively applied to problems of rigid body shape alignment in medical research. ​Many variants of ICP have been introduced. All ICP-based algorithms seek optimal alignment by iterating two key algorithmic ​steps: a correspondence phase that computes ​a match on the target shape for each point comprising the source shape and a registration phase that computes the rigid body transformation that optimally aligns ​the matched ​point pairs.
  
-To date, ICP variants have relied on point position information for computing shape alignment. ​We have a developed ​a new algorithm, namely Iterative ​Closest ​Oriented Point (ICOP), which incorporates both position and surface normal information to compute ​a shape alignment. ​Position and orientation information for each point are combined in a probabilistic framework and incorporated ​into both registration and correspondence phases of the algorithm. It is hypothesized ​that using this richer descriptor set may improve ​registration accuracy and robustness ​when orientations are known or may be determinedespecially ​in case of sparsely sampled data.+We introduce ​a new probabilistic variant of the ICP algorithm, namely Iterative ​Most Likely ​Oriented Point (IMLOP), which incorporates both surface ​position and surface normal information to compute ​an alignment ​of two shapesprobabilistic framework ​is used to incorporate the position ​and orientation data into both the correspondence ​and registration ​phases of the algorithm, where orientation data is treated as Fisher distributed and position data is treated as Gaussian distributed. For the correspondence phase, we present a novel search algorithm ​that efficiently computes the most probable matches considering both position and orientation information of the target shape and of the source points being matched. For the registration ​phase, we present a closed-form solution to the problem of computing the rigid body transformation that maximizes the likelihood of the oriented point matches under the probabilistic noise model. 
 + 
 +Experiments by simulation using human femur data segmented from CT imaging indicate that IMLOP is able to register shapes with a higher degree of accuracy ​than ICP. By evaluating the final residual error in both position ​and orientation of the final matches, IMLOP is able to automatically detect ​when a "​correct"​ registration has been achieved. We demonstrate that IMLOP'​s criteria for detecting a registration failure is much more robust than what is possible using the sub-set of this criteria that is available to ICP. Finally, in our tests IMLOP computed a registration solution is less than half the run-time ​of ICP on average. 
 + 
 +{{ ::​research:​imlop:​imlop_testresutls_tre_vs_matchresiduals.png?600 |}}
  
 ===== Project Personnel ===== ===== Project Personnel =====
   * Seth Billings   * Seth Billings
-  * Emad Boctor 
   * Russell Taylor   * Russell Taylor
  
 ===== Funding ===== ===== Funding =====
   * National Science Foundation, Graduate Research Fellowship Program (NSF GRFP)   * National Science Foundation, Graduate Research Fellowship Program (NSF GRFP)
-  * National Institutes of Health, Individual Graduate Partnership Program (NIH GPP)  
-  * Gift from Intuitive Surgical 
   * Johns Hopkins University Internal Funds   * Johns Hopkins University Internal Funds
  
 ===== Affiliated labs ===== ===== Affiliated labs =====
   * [[:​start|Computer Integrated Interventional Systems Laboratory]]   * [[:​start|Computer Integrated Interventional Systems Laboratory]]
-  * [[https://​musiic.lcsr.jhu.edu/​Main_Page|Medical UltraSound Imaging and Intervention Collaboration (MUSiiC)]] 
  
 ===== Publications ===== ===== Publications =====
 +  * Seth Billings and Russell Taylor, //Iterative Most Likely Oriented Point Registration//,​ Accepted to MICCAI 2014
  
 +TODO following publication:​
 +The final publication is available at Springer via http://​dx.doli.org/​[insert DOI]
research.imlop.1404689154.txt.gz · Last modified: 2019/08/07 12:05 (external edit)




ERC CISST    LCSR    WSE    JHU