Bayesian 3D shape from silhouettes

Kim, Dong-Hoon; Ruttle, Jonathan; Dahyot, Rozenn

doi:10.1016/j.dsp.2013.06.007

Cited by 8 publications

(11 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…pixel neighborhood information in the depth map). Colour information can also be included in the objective function to estimate a coloured surface [39]. As a simple final stage in Fig.…”

Section: Discussionmentioning

confidence: 99%

“…11). As an extension to [36,39], the last function F 3 relates the depth values to the latent 3D locations. The joint density function p x;W ðx; WÞ can be approximated by its empirical estimate using the observations in ..;C .…”

Section: Link Function Fmentioning

confidence: 99%

“…Note that this modelling links explicitly the observed quantity x from the cameras with the additive perturbation . The first two functions ðF 1 ; F 2 Þ link the pixel positions to the latent 3D locations and was used in Ruttle et al's modelling to infer shape-from-silhouettes [36,39] (cf. Eq.…”

Section: Link Function Fmentioning

confidence: 99%

“…Both Ruttle and Kim [36,39] considered that the camera matrices associated with each recorded image are known exactly. In practice however, despite careful calibration, errors occur on the extrinsic camera parameters.…”

Section: Gr 2 T For 3d Shape Reconstruction From Multiple Viewsmentioning

confidence: 99%

“…Kim et al [39] proposed to extend the link function (11) to infer a coloured 3D surface from colour images captured from multiple views. Both Ruttle and Kim [36,39] considered that the camera matrices associated with each recorded image are known exactly.…”

Section: Gr 2 T For 3d Shape Reconstruction From Multiple Viewsmentioning

confidence: 99%

See 4 more Smart Citations

Robust shape from depth images with GR2T

Ruttle

Arellano

Dahyot

2014

Pattern Recognition Letters

Self Cite

View full text Add to dashboard Cite

a b s t r a c tThis paper proposes to infer accurately a 3D shape of an object captured by a depth camera from multiple view points. The Generalised Relaxed Radon Transform (GR 2 T) [1] is used here to merge all depth images in a robust kernel density estimate that models the surface of an object in the 3D space. The kernel is tailored to capture the uncertainty associated with each pixel in the depth images. The resulting cost function is suitable for stochastic exploration with gradient ascent algorithms when the noise of the observations is modelled with a differentiable distribution. When merging several depth images captured from several view points, extrinsic camera parameters need to be known accurately, and we extend GR 2 T to also estimate these nuisance parameters. We illustrate qualitatively the performance of our modelling and we assess quantitatively the accuracy of our 3D shape reconstructions computed from depth images captured with a Kinect camera.

show abstract

“…pixel neighborhood information in the depth map). Colour information can also be included in the objective function to estimate a coloured surface [39]. As a simple final stage in Fig.…”

Section: Discussionmentioning

confidence: 99%

Section: Link Function Fmentioning

confidence: 99%

Section: Link Function Fmentioning

confidence: 99%

Section: Gr 2 T For 3d Shape Reconstruction From Multiple Viewsmentioning

confidence: 99%

Section: Gr 2 T For 3d Shape Reconstruction From Multiple Viewsmentioning

confidence: 99%

See 3 more Smart Citations

Robust shape from depth images with GR2T

Ruttle

Arellano

Dahyot

2014

Pattern Recognition Letters

Self Cite

View full text Add to dashboard Cite

show abstract

3D morphometry of red blood cells by digital holography

et al. 2014

View full text Add to dashboard Cite

Three dimensional (3D) morphometric analysis of flowing and not-adherent cells is an important aspect for diagnostic purposes. However, diagnostics tools need to be quantitative, label-free and, as much as possible, accurate. Recently, a simple holographic approach, based on shape from silhouette algorithm, has been demonstrated for accurate calculation of cells biovolume and displaying their 3D shapes. Such approach has been adopted in combination with holographic optical tweezers and successfully applied to cells with convex shape. Nevertheless, unfortunately, the method fails in case of specimen with concave surfaces. Here, we propose an effective approach to achieve correct 3D shape measurement that can be extended in case of cells having concave surfaces, thus overcoming the limit of the previous technique. We prove the new procedure for healthy red blood cells (RBCs) (i.e., discocytes) having a concave surface in their central region.Comparative analysis of experimental results with a theoretical 3D geometrical model of RBC is discussed in order to evaluate accuracy of the proposed approach. Finally, we show that the method can be also useful to classify, in terms of morphology, different varieties of RBCs. V C 2014 International Society for Advancement of Cytometry Key terms red blood cells; morphological analysis; microfluidics; microscopy; digital holography; optical tweezers IN recent years, digital holography (DH) has gained credit as non-invasive and label-free diagnostic method in several research fields (1-6). Various microscope DH configurations have been proven to be powerful diagnostic tools for investigating micro-objects (7). For example, off-axis DH in microscopy is a very effective process for achieving high-precision, quantitative, phase-contrast mapping (QPM) of biological specimens as it allows measurements of optical thickness with nanometric accuracy (7). The main advantage of DH is the ability to furnish quantitative and intensity-independent microscope QPMs thus allowing quasi real-time and long time-lapse investigation of in vitro biological specimens (8-11). Nowadays, one of the main challenging goals in the field of bio-microscopy would be a diagnostic analysis through quantitative, label-free, and accurate 3D morphometric characterization. Several techniques have been proposed for the 3D morphological study of biological specimens, such as for example optical coherent tomography (12-16) and confocal microscopy (17-19), but they are not the most suitable methods in case of dynamic events such as flowing and not-adherent cells. Recently, a simple and fast method, based on shape from silhouette (SFS) algorithm, provided an effective solution for calculating cells biovolume and displaying their 3D shape (20). In general, 3D shape evaluation of an object is possible by observing a cell along different directions (21,22). In Ref. 20), the experimental arrangement was a combination of a DH microscope and optical tweezers (OT) (23). OT was employed for inducing selfrotation of the...

show abstract

Deep Shape from a Low Number of Silhouettes

Dahyot

Prasad

2016

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Despite strong progress in the field of 3D reconstruction from multiple views, holes on objects, transparency of objects and textureless scenes, continue to be open challenges. On the other hand, silhouette based reconstruction techniques ease the dependency of 3d reconstruction on image pixels but need a large number of silhouettes to be available from multiple views. In this paper, a novel end to end pipeline is proposed to produce high quality reconstruction from a low number of silhouettes, the core of which is a deep shape reconstruction architecture. Evaluations on ShapeNet [1] show good quality of reconstruction compared with ground truth.

show abstract

Bayesian 3D shape from silhouettes

Cited by 8 publications

References 36 publications

Robust shape from depth images with GR2T

Robust shape from depth images with GR2T

3D morphometry of red blood cells by digital holography

Deep Shape from a Low Number of Silhouettes

Contact Info

Product

Resources

About