Capture of arm-muscle deformations using a depth-camera

@inproceedings{robertini2013capture,
  title={Capture of arm-muscle deformations using a depth-camera},
  author={Robertini, Nadia and Neumann, Thomas and Varanasi, Kiran and Theobalt, Christian},
  booktitle={Proceedings of the 10th European Conference on Visual Media Production},
  pages={12},
  year={2013},
  organization={ACM}
}
keywords RGBD , Kinect , PA , IB , CV , DB , ***** , BMM , ML

Abstract
Modeling realistic skin deformations due to underneath muscle bul- 
ging has a wide range of applications in medicine, entertainment 
and art. Current acquisition systems based on dense markers and 
multiple synchronized cameras are able to record and reproduce 
fine-scale skin deformations with sufficient quality. However, the 
complexity and the high cost of these systems severely limit their 
applicability. In this paper, we propose a method for reconstructing 
fine-scale arm muscle deformations using the Kinect depth cam- 
era. The captured data from the depth camera has no temporal 
contiguity and suffers from noise and sensory artifacts, and thus 
unsuitable by itself for potential applications in visual media pro- 
duction or biomechanics. We process noisy depth input to ob- 
tain spatio-temporally consistent 3D mesh reconstructions show- 
ing fine-scale muscle bulges over time. Our main contribution is 
the incorporation of statistical deformation priors into the spatio- 
temporal mesh registration progress. We obtain these priors from 
a previous dataset of a limited number of physiologically different 
actors captured using a high fidelity acquisition setup, and these 
priors help provide a better initialization for the ultimate non-rigid 
surface refinement that models deformations beyond the range of 
the previous dataset. Thus, our method is an easily scalable frame- 
work for bootstrapping the statistical muscle deformation model, 
by extending the set of subjects through a Kinect based acquisi- 
tion process. We validate our spatio-temporal surface registration 
method on several arm movements performed by people of differ- 
ent body shapes. 

Puede servir como una herramienta en el paso de postural analysis para mejorar el proceso con mejor informacion biomecánica

problem: reocnstr con gran calidad deform muscular

Kinect sirve pro tiene  problemas que no lo hacen util para obtener esta reconstriccion con gran calidad
lo que proponen: un metodo que lo hace de depth images de kinect para la region del brazo y hombro. Ayudando a este proceso que sea posible con dispositivos baratos kamagra pas cher.
Presentan alternativas para modelar deformaciones

 

Basicamente resuelven el problema de modelar realisticamente deformaciones del cuerpo humano con distintos fines, entre ellos usarlos para modelos internos del musculo en biomecánica!!.
No se como de aca podemos ir a medir algun otro valor del estado o fuerza del músculo pero puede ser representativo del estado interno.
Es más modelamiento de deformaciones 

En resumen este trabajo presenta una forma de modelar la deformaciones de la piel para obtener como posible aplicación ayudar al calculo de la biomecánica pero es usando una camara de profundidad no se si sirve o como se puede hacer con una camara rgb convencional

Me parece que puede servir para usar ese modelo generado como fuente de información o para extraer de el las features, pero se complejiza la cosa me parece.

Los datasets que tienen al parecer están publicamente available

In \cite{robertinicapture} the authors propose an economic and non intrusive method to reconstruct skin deformation due to internal muscle bulging using RGBD sensors like microsoft Kinect and prior deformation data.
The way that they achieve this is processing the depth input to reconstruct the 3D meshes helping the process with data of muscle deformation previously aquired by a more precise (but more intrusive and expensive) multicamera acquisition system.
In general view their method take noisy dpeth frames and in a first step performs a data cleaning process. After that they try to fit a template to the depth points using an Iterative Closest  Point algorithm. Acorrding to the pose and shape they imppose restrictions to get a better alignment of the mesh with the measurements. A final refinment step is performed to get better and more realistic models.
The approach proposed here could be usefull in our research as a preprocesing step to get more accurate 3D Meshes of the deformation of the skin that could be used as input in our process or approach.