Body surface imaging
View Patent ↗An embodiment of the invention may provide a portable, inexpensive two-view 3D stereo vision imaging system, which acquires a 3D surface model and dimensions of an object. The system may comprise front-side and back-side stereo imagers which each have a projector and at least two digital cameras to image the object from different perspectives. An embodiment may include a method for reconstructing an image of a human body from the data of a two-view body scanner by obtaining a front scan image data point set and a back scan image data point set. A smooth body image may be gained by processing the data point sets using the following steps: (1) data resampling; (2) initial mesh generation; (3) mesh simplification; and (4) mesh subdivision and optimization.
1. A method comprising:
determining a front image data set based on stereoscopically imaging a front-side of a subject using first and second front-side digital cameras collectively positioned in a first static position;
projecting a front visual pattern on the front-side of the subject to create a front artificial texture using a front projector;
determining a back image data set based on stereoscopically imaging a back-side of the subject using first and second back-side digital cameras collectively positioned in a second static position;
projecting a back visual pattern on the back-side of the subject to create a back artificial texture using a back projector;
determining a mesh, which fills at least a portion of a non-imaged occlusion between the front and back image data sets, based on the front and back image data sets;
optimizing a disparity map based on
E
(
f
)
=
∑
i
∈
P
D
p
(
f
p
)
+
∑
(
p
,
q
)
∈
N
V
p
,
q
(
f
p
,
f
p
)
;
and
determining a three-dimensional whole body surface image, without repositioning any of the front-side or back-side cameras from the first and second static positions, based on the (a) mesh, (b) front and back artificial textures, (c) the front and back image data sets; and (d) the optimized disparity map.
2. The method of claim 1 including simultaneously and instantaneously capturing stereoscopic images of the front-side and back-side of the subject using all of the front-side and back-side digital cameras.
3. The method of claim 1 including determining the three-dimensional whole body surface image at a sub-pixel resolution.
4. The method of claim 1 including zippering the front and back image sets, using an Add-Delete method, to create the mesh.
5. The method of claim 1 including:
adding points to an edge of one of the front and back image data sets and weighting the added points to determine an initial mesh;
soft-deleting the weighted added points to determine a control mesh; and
subdividing the control mesh to determine the mesh.
6. The method of claim 1 including:
resampling the front image data set, on a front regular grid, to determine a front mesh and resampling the back image data set, on a back regular grid, to determine a back mesh;
generating an initial mesh that couples the front mesh to the back mesh;
using soft-deletion to simplify the initial mesh to determine a control mesh; and
subdividing the control mesh to fit original image data, acquired from the first and second front-side and back-side cameras, and determine the mesh.
7. The method of claim 6 including:
storing the control mesh into a non-volatile memory included in a computer; and
discarding the original image data and then discontinuing power to the computer followed by reconstructing the mesh based on the stored control mesh.
8. The method of claim 1 including:
determining a disparity space image and segmenting the subject from a related background to determine the disparity map; and
refining the disparity map, via local matching on a pixel-by-pixel basis followed by global optimization, to determine the optimized disparity map;
wherein the front-side corresponds to an anterior side of the subject and the back-side corresponds to a posterior side of the subject.
9. The method of claim 1 including, based on the three-dimensional whole body surface image, determining one or more of the following parameters for the subject including body volume, obesity, medication dosage, and body fat.
10. An article comprising a non-transitory medium storing instructions that enable a processor-based system to:
determine front and back image data sets based on a stereoscopically imaged front-side and back-side of a subject respectively imaged using first and second front-side digital cameras collectively positioned in a first static position and first and second back-side digital cameras collectively positioned in a second static position;
determine a mesh, which fills at least a portion of a non-imaged occlusion between the front and back image data sets, based on the front and back image data sets; and
optimize a disparity map based on
E
(
f
)
=
∑
i
∈
P
D
p
(
f
p
)
+
∑
(
p
,
q
)
∈
N
V
p
,
q
(
f
p
,
f
p
)
;
determine a three-dimensional whole body surface image, without having repositioned any of the front-side or back-side cameras from the first and second static positions, based on the (a) mesh, (b) the front and back image data sets, and (c) the optimized disparity map.
11. The article of claim 10 storing instructions that enable the system to determine the three-dimensional whole body surface image based on having used a total maximum plurality of cameras that does not exceed eight cameras.
12. The article of claim 10 storing instructions that enable the system to zipper the front and back image sets, using an Add-Delete method, to create the mesh.
13. An apparatus comprising:
a non-transitory memory to receive image data;
a processor, coupled to the memory, to determine (a) front and back image data sets based on stereoscopically imaging a front-side and a back-side of a subject respectively using first and second front-side digital cameras collectively positioned in a first static position and first and second back-side digital cameras collectively positioned in a second static position; (b) a mesh, which fills at least a portion of a non-imaged occlusion between the front and back image data sets, based on the front and back image data sets; an optimized disparity map based on
E
(
f
)
=
∑
i
∈
P
D
p
(
f
p
)
+
∑
(
p
,
q
)
∈
N
V
p
,
q
(
f
p
,
f
p
)
;
and (c) a three-dimensional whole body surface image, without repositioning any of the front-side or back-side cameras from the first and second static positions, based on the mesh, the optimized data map, and the front and back image data sets.
14. The apparatus of claim 13 , wherein the apparatus is portable.
15. The apparatus of claim 14 , wherein the apparatus includes a total maximum plurality of cameras that does not exceed four cameras.
16. The apparatus of claim 13 , wherein processor is to, based on the three-dimensional whole body surface image, determine one or more of the following parameters for the subject including body circumference, body volume, obesity, medication dosage, body fat, and apparel sizing.
17. The apparatus of claim 13 , wherein the apparatus is configured to passively determine the three-dimensional whole body surface image.