Method of image segmentation

1. An image segmentation method comprising:

selecting a region of interest including an object whose image is to be segmented;

detecting the edge of the object;

using a processor for repeatedly performing, until a specified precision is obtained, a process in which by using a layered mesh structure based on the detected edge, a mesh boundary of the object is detected, a mesh size is reduced by one layer, and by using areas adjacent to the detected mesh boundary, correcting the mesh boundary; and

generating a mask to fill an inside area of the correcting mesh boundary and masking the object.

2. The method of claim 1 , wherein the region of interest is selected by combining at least one region distinguished by a circle, an oval, a polygon, or a hand-drawn line by a user, and including the object.

3. The method of claim 1 , wherein the detecting of mesh boundary of the object includes:

specifying an initial mesh size;

forming a mesh based on the edge according to the specified initial mesh size; and

detecting the mesh boundary from the formed mesh.

4. The method of claim 3 , wherein the correcting of the mesh boundary includes:

detecting a candidate node containing the mesh boundary based on the mesh size reduced by one layer; and

expanding a region to a specified area adjacent to the mesh boundary by taking the detected candidate node as a seed.

5. The method of claim 3 , wherein, in the detecting of the mesh boundary, the mesh boundary is detected until all of the boundary of the object is visited according to a priority of detecting a mesh block containing a next edge from a current mesh block, and the priority is defined based on a left-hand method and a right-hand method.

6. The method of claim 5 , referring to a motion direction from a previous mesh block to a current mesh block, the priority is defined according to the left-hand method and the right-hand method on the basis of a virtual line perpendicular to a motion direction.

7. The method of claim 6 , wherein, in the detecting of the mesh boundary, a chain coding method is used.

8. The method of claim 4 , wherein, in the expanding of the region, a local region growing method taking a candidate node detected in an immediately previous layer as a seed is used.

9. The method of claim 1 , wherein the specified precision indicates a precision that as the result of repetitive reduction of the mesh size from an upper layer to a lower layer, the mesh size of a current layer is the same as a threshold.

10. The method of claim 9 , wherein the threshold indicates the size of one pixel.

11. computer-readable storage medium encoded with processing instructions for causing a processor to execute an image segmentation method, the method comprising:

selecting a region of interest including an object whose image is to be segmented;

detecting an edge of the object;

repeatedly performing, until a specified precision is obtained, a process in which by using a layered mesh structure based on the detected edge, a mesh boundary of the object is detected, a mesh size is reduced by one layer, and correcting the mesh boundary using areas adjacent to the detected mesh boundary; and

generating a mask to fill an inside area of the correcting mesh boundary and masking the object.

12. An image segmenting method, comprising:

selecting a region of interest including an object whose image is to be segmented;

detecting an edge of the object;

specifying a mesh size;

forming a mesh using a layered mesh structure based on the detected edge;

detecting a mesh boundary of the object;

determining whether the mesh size is 1;

reducing the mesh size by one layer when the mesh size is not 1;

using a processor for correcting the mesh boundary using areas adjacent to the detected mesh boundary;

generating a mask filling an inside area of the corrected mesh boundary; and

image masking the object.

13. The method of claim 12 , wherein the region of interest is specified by a user.

14. The method of claim 12 , wherein the edge of the object is detected using a Sobel algorithm.

15. The method of claim 12 , wherein the initial mesh size is a 32×32 pixel unit.

16. The method of claim 12 , wherein the mesh is formed using Delaunay triangulation.

17. An image segmenting unit, comprising:

a selecting section selecting a region of interest including an object whose image is to be segmented;

an edge detecting section detecting an edge of the object;

a specifying section specifying a mesh size;

a forming section forming a mesh using a layered mesh structure based on the detected edge;

a boundary detecting section detecting a mesh boundary of the object;

a size determining section determining whether the mesh size is 1;

a size reducing section reducing the mesh size by one layer when the mesh size is not 1;

a boundary correcting section correcting the mesh boundary using areas adjacent to the detected mesh boundary;

a mask generator generating a mask filling an inside area of the corrected mesh boundary; and

an image masking section performing image masking of the object.