Back to results
US 8,787,642 B2Granted
Method, device and computer-readable recording medium containing program for extracting object region of interest
AssigneeFUJIFILM Corporation
InventorYuanzhong Li
Life of this patent
Estimating the term…
Abstract
An object region extracting process for extracting an object region of interest from an image is automated to the maximum possible extent to improve user-friendliness. In this process, an arbitrary point is set in the object region of interest, and a presence area, which is likely to contain the entire object region of interest, is determined in the image using the set arbitrary point and a possible size of the object region of interest. Then, the object region of interest is extracted from the image based on the set arbitrary point and at least one point outside the determined presence area.
Claims8
A method for extracting a lesion region from an image, the method comprising:
- setting an arbitrary point in the lesion region;
- determining a presence area containing the set arbitrary point, the size of the presence area being larger than a reference size of the lesion region, said reference size being a maximum size of lesions appearing in recorded images that are of a type identical to the lesion region;
- and
- extracting the lesion region from the image based on the set arbitrary point and at least one point outside the determined presence area, wherein the set arbitrary point and the at least one point outside the determined presence area are within a single image from which the lesion region is being extracted, and wherein the lesion region is extracted based on a graph cut segmentation method using data of the arbitrary point as a first label, and data of the point outside the determined presence area as a second label, and wherein within the presence area, each pixel is assigned to one of the first label and the second label based on a probability of each pixel corresponding a group of the first label or the second label.
1 more element
A device for extracting a lesion region from an image, the device comprising:
- setting means to set an arbitrary point in the lesion region;
- area determining means to determine a presence area containing the set arbitrary point, the size of which is equal to or larger than a possible size of the lesion region said possible size being a maximum size of lesion regions, in recorded images, that are of a type identical to the lesion region;
- and
- object region extracting means to extract the lesion region from the image based on the set arbitrary point and at least one point outside the determined presence area, and wherein the object region extracting means extracts the lesion region based on a graph cut segmentation method using data of the arbitrary point as a first label, and data of the point outside the determined presence area as a second label, and wherein within the presence area, each pixel is assigned to one of the first label and the second label based on a probability of each pixel corresponding a group of the first label or the second label.
1 more element
A non-transitory computer-readable recording medium containing a program for causing a computer to carry out a process for extracting a lesion region from an image, the process comprising:
- setting an arbitrary point in the lesion region;
- determining a presence area containing the set arbitrary point, the size of which is equal to or larger than a possible size of the lesion region said possible size being a maximum size of lesion regions, in recorded images, that are of a type identical to the lesion region;
- and extracting the lesion region from the image based on the set arbitrary point and at least one point outside the determined presence area, wherein the lesion region is extracted based on a graph cut segmentation method using data of the arbitrary point as a first label, and data of the point outside the determined presence area as a second label, and wherein within the presence area, each pixel is assigned to one of the first label and the second label based on a probability of each pixel corresponding a group of the first label or the second label.
A method for extracting an organ region from an image, the method comprising:
- setting an arbitrary point in the organ region;
- determining a presence area containing the set arbitrary point, the size of which is equal to or larger than a possible size of the organ region said possible size being a maximum size of organ regions, in recorded images, that are of a type identical to the organ region;
- and
- extracting the organ region from the image based on the set arbitrary point and at least one point outside the determined presence area, and wherein the organ region is extracted based on a graph cut segmentation method using data of the arbitrary point as a first label, and data of the point outside the determined presence area as a second label, and wherein within the presence area, each pixel is assigned to one of the first label and the second label based on a probability of each pixel corresponding a group of the first label or the second label.
1 more element
A device for extracting an organ region from an image, the device comprising:
- setting means to set an arbitrary point in the organ region;
- area determining means to determine a presence area containing the set arbitrary point, the size of which is equal to or larger than a possible size of the organ region said possible size being a maximum size of organ regions, in recorded images, that are of a type identical to the organ region;
- and
- object region extracting means to extract the organ region from the image based on the set arbitrary point and at least one point outside the determined presence area, and wherein the object region extracting means extracts the organ region based on a graph cut segmentation method using data of the arbitrary point as a first label, and data of the point outside the determined presence area as a second label, and wherein within the presence area, each pixel is assigned to one of the first label and the second label based on a probability of each pixel corresponding a group of the first label or the second label.
1 more element
A non-transitory computer-readable recording medium containing a program for causing a computer to carry out a process for extracting an organ region from an image, the process comprising:
- setting an arbitrary point in the organ region;
- determining a presence area containing the set arbitrary point, the size of which is equal to or larger than a possible size of the organ region said possible size being a maximum size of organ regions, in recorded images, that are of a type identical to the organ region;
- and
- extracting the organ region from the image based on the set arbitrary point and at least one point outside the determined presence area, and wherein the organ region is extracted based on a graph cut segmentation method using data of the arbitrary point as a first label, and data of the point outside the determined presence area as a second label, and wherein within the presence area, each pixel is assigned to one of the first label and the second label based on a probability of each pixel corresponding a group of the first label or the second label.
1 more element
Dependencies are inferred from the claim text.