Adaptive processing in time of flight imaging
Examples are disclosed herein relating to signal processing in a time-of-flight (ToF) system. One example provides, a method comprising emitting, via a light source, amplitude-modulated light toward an object, acquiring, via an image sensor comprising a plurality of pixels, a plurality of image frames capturing light emitted from the light source that is reflected by the object, wherein the plurality of image frames are acquired at two or more different frequencies of the amplitude-modulated light and collectively form a multifrequency frame, and for each pixel of the multifrequency frame, determining a brightness level, applying an adaptive denoising process by setting a kernel size based on the brightness level, and performing a phase unwrapping process to determine a depth value for the pixel.
1. A method for signal processing in a time of flight system, the method comprising:
emitting, via a light source, amplitude-modulated light toward an object;
acquiring, via an image sensor comprising a plurality of pixels, a plurality of image frames capturing light emitted from the light source that is reflected by the object, wherein the plurality of image frames are acquired at two or more different frequencies of the amplitude-modulated light and collectively form a multifrequency frame; and
for each pixel of the multifrequency frame,
determining a brightness level,
applying an adaptive denoising process by setting a kernel size based on the brightness level, and
performing a phase unwrapping process to determine a depth value for the pixel.
2. The method of claim 1 , wherein setting the kernel size comprises setting a larger kernel for a lower brightness level, and setting a smaller kernel for a higher brightness level.
3. The method of claim 2 , wherein setting a smaller kernel comprises adding zeroes to a kernel.
4. The method of claim 1 , wherein applying the adaptive denoising process comprises applying:
S
(
m
,
n
,
k
)
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
S
~
(
m
,
n
,
k
)
with
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
=
1
wherein
{tilde over (S)}(m, n, k) is a signal corresponding to pixel (m,n) for a frequency k, of an array of size (M, N) with {m∈1, 2, . . . , M}, and {n∈1, 2, . . . , N}, for a frequency k∈1, 2, . . . , K,
S(m, n, k) is a signal with reduced noise, represented as a weighted sum of neighboring pixels, {i∈−I, −I+1, . . . , I} and {j∈−J, −J+1, . . . , J}, and
λ(i, j, k) are edge-preserving normalized weighting coefficients.
5. The method of claim 4 , wherein
λ
(
i
,
j
,
k
)
=
1
Norm
(
m
,
n
,
k
)
ξ
(
i
,
j
,
k
)
,
wherein
ξ
(
i
,
j
,
k
)
=
exp
{
-
[
(
m
-
i
)
2
+
(
n
-
j
)
2
]
/
σ
sm
2
}
exp
{
-
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
/
σ
sh
2
]
,
and
Norm
(
m
,
n
,
k
)
=
∑
i
=
-
I
I
∑
j
=
-
J
J
ξ
(
i
,
j
,
k
)
,
and wherein σ sm 2 and σ sh 2 are parameters to adjust for one or more of smoothing and sharpening.
6. The method of claim 4 , wherein
λ
(
i
,
j
,
k
)
=
1
Norm
(
m
,
n
,
k
)
ξ
(
i
,
j
,
k
)
,
wherein
ξ( i,j,k )=γ( i,j,k ) I τ ( m,n,k ),
wherein
γ
(
i
,
j
,
k
)
→
1
-
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
σ
sh
2
∑
i
=
-
I
I
∑
j
=
-
J
J
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
,
and wherein
I
τ
(
m
,
n
,
k
)
=
{
0
if
γ
(
i
,
j
,
k
)
<
τ
(
m
,
n
,
k
)
1
if
γ
(
i
,
j
,
k
)
≥
τ
(
m
,
n
,
k
)
.
7. The method of claim 6 , wherein τ(m, n, k) is variable.
8. The method of claim 1 , wherein performing the phase unwrapping process comprises performing adaptive phase unwrapping based upon the brightness level.
9. The method of claim 8 , wherein performing the adaptive phase unwrapping based upon the brightness level comprises applying
{
cos
[
ϕ
~
(
i
,
j
,
k
)
]
=
S
~
r
(
i
,
j
,
k
)
(
i
,
j
,
k
)
sin
[
ϕ
~
(
i
,
j
,
k
)
]
=
S
~
i
(
i
,
j
,
k
)
(
i
,
j
,
k
)
,
wherein
(
i
,
j
,
k
)
=
(
S
~
r
(
i
,
j
,
k
)
)
2
+
(
S
~
i
(
i
,
j
,
k
)
)
2
10. The method of claim 9 , wherein cos [ϕ(i, j, k)] and sin [ϕ(i, j, k)] represent a signal with reduced noise and are determined by:
{
cos
[
ϕ
(
i
,
j
,
k
)
]
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
cos
[
ϕ
~
(
i
,
j
,
k
)
]
sin
[
ϕ
(
i
,
j
,
k
)
]
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
sin
[
ϕ
~
(
i
,
j
,
k
)
]
.
11. A time of flight system, comprising:
a light source;
an image sensor comprising a plurality of pixels;
memory comprising instructions stored thereon; and
a processor configured to execute the instructions to:
emit amplitude-modulated light from the light source toward an object;
acquire via the image sensor a plurality of image frames capturing light emitted from the light source that is reflected by the object, wherein the plurality of image frames are acquired at two or more different frequencies of the amplitude-modulated light and collectively form a multifrequency frame; and
for each pixel of the multifrequency frame,
determine a brightness level;
apply an adaptive denoising process by setting a kernel size based on the brightness level; and
perform a phase unwrapping process to determine a depth value for the pixel.
12. The time of flight system of claim 11 , wherein the instructions are executable to set a larger kernel for a lower brightness level, and set a smaller kernel for a higher brightness level.
13. The time of flight system of claim 12 , the instructions are executable to set a smaller kernel by adding zeroes to a kernel.
14. The time of flight system of claim 11 , wherein the instructions are executable to emit amplitude-modulated light at a plurality of different modulation frequencies.
15. The time of flight system of claim 11 , wherein the instructions are executable to apply the adaptive denoising process by applying:
S
(
m
,
n
,
k
)
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
S
~
(
m
,
n
,
k
)
with
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
=
1
wherein
{tilde over (S)}(m, n, k) is a signal corresponding to pixel (m,n) for a frequency k, of an array of size (M, N) with {m∈1, 2, . . . , M}, and {n∈1, 2, . . . , N}, for a frequency k∈1, 2, . . . , K,
S(m, n, k) is a signal with reduced noise, represented as a weighted sum of neighboring pixels, {i∈−I, −I+1, . . . , I} and {j∈−J, −J+1, . . . , J}, and
λ(i, j, k) are edge-preserving normalized weighting coefficients.
16. The time of flight system of claim 15 , wherein
λ
(
i
,
j
,
k
)
=
1
Norm
(
m
,
n
,
k
)
ξ
(
i
,
j
,
k
)
,
wherein
ξ( i,j,k )=γ( i,j,k ) I τ ( m,n,k ),
wherein
γ
(
i
,
j
,
k
)
→
1
-
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
σ
sh
2
∑
i
=
-
I
I
∑
j
=
-
J
J
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
,
and wherein
I
τ
(
m
,
n
,
k
)
=
{
0
if
γ
(
i
,
j
,
k
)
<
τ
(
m
,
n
,
k
)
1
if
γ
(
i
,
j
,
k
)
≥
τ
(
m
,
n
,
k
)
.
17. The time of flight system of claim 16 , wherein the instructions are executable to perform the phase unwrapping process by using an adaptive complex domain unwrapping formula comprising:
{
cos
[
ϕ
~
(
i
,
j
,
k
)
]
=
S
~
r
(
i
,
j
,
k
)
(
i
,
j
,
k
)
sin
[
ϕ
~
(
i
,
j
,
k
)
]
=
S
~
i
(
i
,
j
,
k
)
(
i
,
j
,
k
)
,
wherein
(
i
,
j
,
k
)
=
(
S
~
r
(
i
,
j
,
k
)
)
2
+
(
S
~
i
(
i
,
j
,
k
)
)
2
,
wherein cos [ϕ(i, j, k)] and sin [ϕ(i, j, k)] represent a signal with reduced noise and are determined by:
{
cos
[
ϕ
(
i
,
j
,
k
)
]
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
cos
[
ϕ
~
(
i
,
j
,
k
)
]
sin
[
ϕ
(
i
,
j
,
k
)
]
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
sin
[
ϕ
~
(
i
,
j
,
k
)
]
.
18. The time of flight system of claim 15 , wherein
λ
(
i
,
j
,
k
)
=
1
Norm
(
m
,
n
,
k
)
ξ
(
i
,
j
,
k
)
,
wherein
ξ
(
i
,
j
,
k
)
=
1
-
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
σ
sh
2
∑
i
=
-
I
I
∑
j
=
-
J
J
S
~
(
m
,
n
,
k
)
-
S
~
(
i
,
j
,
k
)
2
,
and
Norm
(
m
,
n
,
k
)
=
∑
i
=
-
I
I
∑
j
=
-
J
J
ξ
(
i
,
j
,
k
)
,
and wherein σ sh 2 is a parameter to adjust for sharpening.
19. A time of flight depth imaging system, comprising
a light source;
an image sensor;
memory comprising instructions stored thereon; and
a processor configured to execute the instructions to:
emit amplitude-modulated light from the light source toward an object,
acquire via the image sensor a plurality of image frames capturing light emitted from the light source that is reflected by the object, wherein the plurality of image frames are acquired at two or more different frequencies of the amplitude-modulated light and collectively form a multifrequency frame, and
for each pixel of the multifrequency frame,
determine a brightness level;
apply an adaptive denoising process by setting a kernel size based on the brightness level; and
perform an adaptive complex domain phase unwrapping based on the brightness level to determine a depth value for the pixel.
20. The time of flight system of claim 19 , wherein the instructions executable to perform the adaptive complex domain phase unwrapping comprise instructions executable to apply:
{
cos
[
ϕ
~
(
i
,
j
,
k
)
]
=
S
~
r
(
i
,
j
,
k
)
AB
~
(
i
,
j
,
k
)
sin
[
ϕ
~
(
i
,
j
,
k
)
]
=
S
~
i
(
i
,
j
,
k
)
AB
~
(
i
,
j
,
k
)
,
wherein (i,j,k)=√{square root over (({tilde over (S)} r (i, j, k)) 2 +({tilde over (S)} i (i, j, k)) 2 )}, and
wherein cos [ϕ(i, j, k)] and sin [ϕ(i, j, k)] represent a signal with reduced noise and are determined by:
{
cos
[
ϕ
(
i
,
j
,
k
)
]
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
cos
[
ϕ
~
(
i
,
j
,
k
)
]
sin
[
ϕ
(
i
,
j
,
k
)
]
=
∑
i
=
-
I
I
∑
j
=
-
J
J
λ
(
i
,
j
,
k
)
sin
[
ϕ
~
(
i
,
j
,
k
)
]
.