scholarly journals Active Shape Model Search using Local Grey-Level Models: A Quantitative Evaluation

1993 ◽  
Author(s):  
T. F. Cootes ◽  
C. J. Taylor
Keyword(s):  
Quantitative Evaluation ◽  
Active Shape Model ◽  
Grey Level ◽  
Shape Model ◽  
Model Search ◽  
Active Shape

scholarly journals The interleaved partial active shape model search (IPASM) algorithm – Preliminary results of a novel approach towards 3D ultrasound-based bone surface reconstruction

10.29007/ckw2 ◽  
2018 ◽  
Author(s):  
Christoph Hänisch ◽  
Benjamin Hohlmann ◽  
Klaus Radermacher
Keyword(s):  
Search Algorithm ◽  
Imaging Modality ◽  
Statistical Shape Model ◽  
Training Data ◽  
Active Shape Model ◽  
Imaging Data ◽  
Shape Model ◽  
Underlying Distribution ◽  
Model Search ◽  
Active Shape

In applications such as biomechanical simulations or implant planning, bone surfaces of the knee are most often reconstructed from computed tomography or magnetic resonance imaging data. Here, ultrasound (US) might serve as an alternative imaging modality. However, established methods cannot directly be transferred to US due to differences in imaging quality and underlying physics.In this paper, we present a generalisation of the well-known active shape model search algorithm (ASM) that allows for segmenting various structures in US volume images that are too large to be captured with a single recording. The multi-view segmentation approach uses a-priori knowledge in the form of a statistical shape model (SSM) as is the case with the classical ASM. This allows to extrapolate missing information and to generate shapes that comply with the underlying distribution of some training data. The main differences are, however, that the SSM is not only adapted to a single image but to multiple images and that the adaption process is interleaved. As a result, within each iteration the surface information of all sub-volumes is propagated and used in all subsequent steps.In-silico tests were conducted to investigate how this algorithm would perform in real tracked US data. US volume images were split in slightly overlapping sub-volumes, noise was added, and the alignment was distorted. We could show that the algorithm is capable of reconstructing shapes in the lower millimetre range and for some cases even with submillimetric accuracy. The algorithm is hardly affected by orientation errors below 5 degrees and displacement errors below 5 mm; above these limits, the average absolute SDE as well as its associated variance increases.

Human face shape classification method based on active shape model

2009 ◽  
Vol 29 (10) ◽  
pp. 2710-2712 ◽  
Author(s):  
Li-qiang DU ◽  
Peng JIA ◽  
Zong-tan ZHOU ◽  
De-wen HU
Keyword(s):  
Classification Method ◽  
Active Shape Model ◽  
Shape Classification ◽  
Human Face ◽  
Shape Model ◽  
Face Shape ◽  
Active Shape
Sign in / Sign up

Export Citation Format

Share Document