Finding blob-like structures using diverging gradient field response
View Patent ↗A method for determining a blob-like structure in a target image includes receiving an input target image is received, determining a mean curvature of the input target image, smoothing the mean curvature is smoothed using a scale that determines a size of the blobs to be found, finding a blob-like structure in the input target image as a portion of the input target image where a smoothed curvature is higher than a threshold value, and outputting a result including an indication of the blob-like structure.
1. A non-transitory computer readable medium embodying instructions executable by a processor to perform a method for identifying a blob-like structure in a target image, the method comprising:
receiving an input target image;
determining a mean curvature of the input target image;
smoothing the mean curvature using a scale that determines a size of blob-like structures to be found;
finding a blob-like structure in the input target image as a portion of the input target image where a smoothed curvature is higher than a threshold value wherein finding a blob-like structure comprises computing a diverging gradient field response, wherein the diverging gradient field response is a scale-space of curvature of the input target image, wherein the scale-space of curvature is determined as
dgfr
σ
=
div
(
∇
→
f
∇
→
f
)
*
g
σ
where f is a scalar,
-
div
(
∇
→
f
∇
→
f
)
is the mean curvature, and
g σ is a Gaussian; and
outputting a result including an indication of the blob-like structure.
2. The method of claim 1 , wherein the result is an image of the blob-like structure.
3. The method of claim 1 , wherein the result is a set of coordinates of the blob-like structure.
4. The method of claim 1 , wherein the diverging gradient field response is used to determine a sphere corresponding to the blob-like structure directly in a gradient domain.
5. The method of claim 1 , wherein the input target image is N-dimensional, where N is a natural number.
6. A system for identifying a blob-like structure in a target image comprising:
a memory device storing an input target image and a plurality of instructions embodying a computer aided blob-like structure detection system; and
a processor for receiving the input target image and executing the plurality of instructions to perform a method comprising:
determining a mean curvature of the input target image;
smoothing the mean curvature using a scale that determines a size of the blob-like structures to be found;
finding a blob-like structure in the input target image as a portion of the input target image where a smoothed curvature is higher than a threshold value, wherein finding a blob-like structure comprises computing a diverging gradient field response, wherein the diverging gradient field response is a scale-space of curvature of the input target image, wherein the scale-space of curvature is determined as
dgfr
σ
=
div
(
∇
→
f
∇
→
f
)
*
g
σ
where f is a scalar,
-
div
(
∇
→
f
∇
→
f
)
is the mean curvature, and
g σ is a Gaussian; and
outputting a result including an indication of the blob-like structure.
7. The system of claim 6 , wherein the result is an image of the blob-like structure output to a display device in communication with the processor.
8. The system of claim 6 , wherein the result is a set of coordinates of the blob-like structure stored to the memory.
9. The system of claim 6 , wherein the diverging gradient field response is used to determine a sphere corresponding to the blob-like structure directly in a gradient domain.
10. The system of claim 6 , wherein the input target image is N-dimensional, where N is a natural number.
11. A non-transitory computer readable medium embodying instructions executable by a processor to perform a method for identifying a blob-like structure in a target image, the method comprising:
receiving the target image as input;
determining a diverging gradient field response function, wherein the diverging gradient field response includes a vector correlation of the field and a normalized gradient field of the target image wherein determining the diverging gradient field response function includes determining diverging gradient field components as separable Gaussian filters; and
identifying a blob-like structure in the target image according to the vector correlation and a threshold.
12. The method of claim 11 , wherein diverging gradient field components are normalized.
13. The method of claim 11 , wherein the vector correlation includes multiple convolutions, one per dimension of the target image.