Object based boundary refinement method

• US 20060285743A1
• Filed: 06/20/2005
• Published: 12/21/2006
• Est. Priority Date: 06/20/2005
• Status: Active Grant

• Current Assignee: Leica Microsystems CMS GmbH (Danaher Corporation)
• Original Assignee: Shih-Jong J. Lee
• Inventors: Oh, Seho, Lee, Shih-Jong J.
• Granted Patent: US 7,466,872 B2
• Time in Patent Office: Days
• US Class Current: 382/170
• CPC Class Codes: G06T 2207/20104 Interactive definition of region of interest [ROI]
• CPC Class Codes: G06T 2207/20192 Edge enhancement; Edge preservation
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/155 involving morphological operators
• CPC Class Codes: G06V 10/26 Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation

Abstract

An object based boundary refinement method for object segmentation in digital images receives an image and a single initial object region of interest and performs refinement zone definition using the initial object regions of interest to generate refinement zones output. A directional edge enhancement is performed using the input image and the refinement zones to generate directional enhanced region of interest output. A radial detection is performed using the input image the refinement zones and the directional enhanced region of interest to generate radial detection mask output. In addition, a final shaping is performed using the radial detection mask having single object region output. A directional edge enhancement method determining pixel specific edge contrast enhancement direction according to the object structure direction near the pixel consists receives an image and refinement zones and performs 1D horizontal distance transform and 1D vertical distance transform using the refinement zones to generate horizontal distance map and vertical distance map outputs. A neighboring direction determination is performed using the horizontal distance map and the vertical distance map to generate neighboring image output. In addition, a directional edge contrast calculation using the neighboring image and input image having directional enhanced region of interest output.

Claims

1. An object based boundary refinement method for object segmentation in digital images comprising the steps of:

a) Input an image;

b) Input a single initial object regions of interest;

c) Perform refinement zone definition using the initial object regions of interest having refinement zones output;

d) Perform directional edge enhancement using the input image and the refinement zones having directional enhanced region of interest output;

e) Perform radial detection using the input image, the refinement zones and the directional enhanced region of interest having radial detection mask output.

2. The object based boundary refinement method of claim 1 further performs final shaping using the radial detection mask having single object region output.

3. The object based boundary refinement method of claim 1 wherein the refinement zones contain an outer-limit zone and an inner-limit zone.

4. The object based boundary refinement method of claim 1 wherein the directional edge enhancement method consists of the following steps:

a) Perform 1D horizontal distance transform using the refinement zones having horizontal distance map output;

b) Perform 1D vertical distance transform using the refinement zones having vertical distance map output;

c) Perform neighboring direction determination using the horizontal distance map and the vertical distance map having neighboring image output;

d) Perform directional edge contrast calculation using the neighboring image and input image having directional enhanced region of interest output.

5. The object based boundary refinement method of claim 1 wherein the radial detection method consists of the following steps:

a) Find object center using the initial object region of interest having object center output;

b) Perform rectangular to polar coordinate transformation using the initial object region of interest, the directional enhanced region of interest, the refinement zones and the input image having polar object region, polar enhanced region, polar refinement zones, and polar input image output;

c) Perform polar boundary detection using the polar object region, polar enhanced region, polar refinement zones, and polar input image having polar detection mask output.

6. The radial detection method of claim 5 further consists of a polar to rectangular coordinate transform step using polar detection mask having radial detection mask output.

7. The final shaping method of claim 2 further consists of the following steps:

a) Perform boundary constrain using the radial detection mask having constrained detection mask output;

b) Perform fine boundary detection using the radial detection mask and the directional enhanced region of interest having fine boundary mask output;

c) Combine the constrained detection mask the fine boundary mask and inner-limit zone having combined detection mask output;

d) Perform multi-directional conditional refinement using the combined detection mask and the outer-limit zone having single object region output.

8. A directional edge enhancement method determining pixel specific edge contrast enhancement direction according to the object structure direction near the pixel consists of the following steps:

a) Input refinement zones;

b) Input an image;

c) Perform 1D horizontal distance transform using the refinement zones having horizontal distance map output;

d) Perform 1D vertical distance transform using the refinement zones having vertical distance map output;

e) Perform neighboring direction determination using the horizontal distance map and the vertical distance map having neighboring image output;

f) Perform directional edge contrast calculation using the neighboring image and input image having directional enhanced region of interest output.

9. The 1D horizontal distance method of claim 8 performs first pass scans from left to right and the second pass scans in a reverse right to left order.

10. The neighboring direction determination method of claim 8 contains a δ

x determination method comprising the following steps;

a) Determines the left-right structure direction using the horizontal distance map having the first step δ

x value output;

b) Determine whether the pixel structure is dominated by the vertical direction having the final δ

x value output.

11. The neighboring direction determination method of claim 8 contains a δ

y determination method comprising the following steps;

a) Determines the top-bottom structure direction using the horizontal distance map having the first step δ

y value output;

b) Determine whether the pixel structure is dominated by the horizontal direction having the final δ

y value output.

12. The directional edge contrast calculation method of claim 8 performs subtraction of a pixel'"'"'s intensity from its adjacent neighboring pixels determined by δ

x and δ

y.

13. A radial detection method detects boundary along the radial direction in the polar coordinate transformed image domain consisting of the following steps:

a) Input an image;

b) Input an initial object region of interest;

c) Input directional enhanced region of interest;

d) Input refinement zones;

e) Find object center using the initial object region of interest having object center output;

f) Perform rectangular to polar coordinate transformation using the initial object region of interest, the directional enhanced region of interest, the refinement zones and the input image having polar object region, polar enhanced region, polar refinement zones, and polar input image output;

g) Perform polar boundary detection using the polar object region, polar enhanced region, polar refinement zones, and polar input image having polar detection mask output.

14. The radial detection method of claim 13 further consists of a polar to rectangular coordinate transform step using polar detection mask having radial detection mask output.

15. The radial detection method of claim 13 wherein the polar boundary detection is performed iteratively. In each iteration, the boundary points are determined for all angles, one radial location per angle.

16. The polar boundary detection method of claim 15 wherein the boundary points are determined using a dynamic cost function.

17. The dynamic cost function of claim 16 is determined by a combination of the polar enhanced region value, the radial edge contrast and a smooth factor that is dynamically updated in each iteration.