Analyzing pixel data by imprinting objects of a computer-implemented network structure into other objects

• US 20100265267A1
• Filed: 04/17/2009
• Published: 10/21/2010
• Est. Priority Date: 04/17/2009
• Status: Active Grant

• Current Assignee: Definiens AG (Astrazeneca PLC)
• Original Assignee: Definiens AG (Astrazeneca PLC)
• Inventors: Feehan, Owen, Schmidt, Guenter, Schaepe, Arno, Binnig, Gerd
• Granted Patent: US 8,319,793 B2
• Time in Patent Office: Days
• US Class Current: 345/619
• CPC Class Codes: G06V 10/426 Graphical representations
• CPC Class Codes: G06V 20/176 Urban or other man-made structures
• CPC Class Codes: G06V 20/188 Vegetation

Abstract

An analysis system analyzes digital images using a computer-implemented network structure that includes a process hierarchy, a class network and a data network. The data network includes image layers and object networks. Objects in a first object network are segmented into a first class, and objects in a second object network are segmented into a second class. One process step of the process hierarchy involves generating a third object network by imprinting objects of the first object network into the objects of the second object network such that pixel locations are unlinked from objects of the second object network to the extent that the pixel locations were also linked to objects of the first object network. The imprinting step allows object-oriented processing of digital images to be performed with fewer computations and less memory. Characteristics of an object of the third object network are then determined by measuring the object.

Claims

1. A method comprising:

(a) specifying a first object class;

(b) specifying a second object class;

(c) generating a first object network, wherein each object of the first object network belongs to the first object class, wherein the first object network includes a first object that is linked to a first plurality of pixel locations;

(d) generating a second object network, wherein each object of the second object network belongs to the second object class, wherein the second object network includes a second object that is linked to a second plurality of pixel locations, wherein all of the objects of the second object network are linked to a third plurality of pixel locations, and wherein the third plurality of pixel locations includes the second plurality of pixel locations;

(e) generating a third object network by imprinting each object of the first object network into the second object network, wherein the third object network includes a third object that is linked to a fourth plurality of pixel locations, wherein the fourth plurality of pixel locations is a subset of the third plurality of pixel locations and is calculated from the first plurality of pixel locations; and

(f) determining a characteristic of the third object by measuring the third object.

2. The method of claim 1, wherein the first plurality of pixel locations is in a first digital image, and the third plurality of pixel locations is in a second digital image.

3. The method of claim 2, wherein the calculation in (e) is based on both a first coordinate system of the first digital image and a second coordinate system of the second digital image.

4. The method of claim 1, wherein the calculation in (e) involves scaling, rotating and translating the first plurality of pixel locations.

5. The method of claim 2, wherein the first digital image has a first coordinate system and the second digital image has a second coordinate system, and wherein the calculation in (e) involves scaling, rotating and translating the first plurality of pixel locations using the second coordinate system.

6. The method of claim 5, further comprising:

determining a characteristic of the first object by measuring the first object, wherein the scaling, rotating and translating of the first plurality of pixel locations is performed based on the measured characteristic of the first object.

7. The method of claim 1, further comprising:

(g) generating a fourth object network by modifying the third object network such that each object of the third object network that is linked to pixel locations that are not contiguous to one another is divided into multiple objects of the fourth object network that are linked only to contiguous pixel locations.

8. The method of claim 1, wherein each object of the first object network fulfills a predefined condition.

9. The method of claim 8, wherein the condition is fulfilled by a minimum predetermined proportion of the first plurality of pixel locations satisfying the predefined condition.

10. The method of claim 8, wherein the condition is fulfilled by each object of the first object network having a higher probability of belonging to the first object class than belonging to the second object class.

11. The method of claim 1, wherein the generating the third object network in (e) is performed by imprinting a subset of the objects of the first object network into the second object network, and wherein objects in the subset are linked only to those pixel locations of a N-dimensional digital image that that fall within polyhedral region of the N-dimensional digital image.

12. The method of claim 1, further comprising:

(g) restoring the first object network and discarding the third object network if the characteristic of the third object determined in (f) does not satisfy a predetermined condition.

13. The method of claim 1, wherein the generating the first object network in (c) is performed using one processor of a computer at the same time that the generating the second object network in (d)is performed using another processor of the computer.

14. The method of claim 1, wherein the second object network is stored in a memory space, and wherein the second object network is overwritten by the third object network in the memory space.

15. The method of claim 1, wherein the first object class represents road segments, wherein each road segment has a length, and wherein the measuring the third object in (f) measures the length of the road segment.

16. The method of claim 1, wherein the first object class represents houses, wherein each house has a roof, and wherein the measuring in (f) measures the area of the roof.

17. The method of claim 1, wherein the first object class represents nuclei that are stained positive or negative, and wherein the measuring in (f) measures the number of nuclei stained positive and the number of nuclei stained negative.

18. A method comprising:

in specification mode specifying a first object class and a second object class;

selecting process steps of an image analysis script that define a segmentation method that generates a first object network belonging to the first object class and a second object network belonging to the second object class;

selecting an imprinting process step of the image analysis script by selecting the first object class and the second object class;

selecting a measurment process step of the image analysis script by specifying characteristics of a target object;

saving the image analysis script to a memory;

acquiring pixel values that are associated with pixel locations of a digital image; and

in an execution mode running the image analysis script, wherein the image analysis script;

executes the process steps of the segmentation method and generates the first object network and the second object network, wherein the first object network includes a first object, and wherein the second object network includes a second object;

imprints the first object network into the second object network such that a third object network is generated, wherein the third object network includes a third object; and

determines a characteristic of the third object.

19. The method of claim 18, wherein in the execution mode the image analysis script restores the second object network and discards the third object network if the characteristic of the third object does not satisfy a predetermined condition.

20. The method of claim 18, wherein the image analysis script generates the first object network by executing on a first computer processor and generates the second object network by executing at the same time on a second computer processor.

21. The method of claim 18, wherein the characteristic of the third object is displayed on a computer display.

22. The method of claim 18, wherein the characteristic of the third object is stored in a database.

23. A computer-readable medium having computer-executable instructions for analyzing a digital image by performing the steps of:

(a) generating a first plurality of objects by linking a first plurality of pixel locations on the digital image to the first plurality of objects, wherein the first plurality of objects forms a first object network;

(b) linking each of the first plurality of objects to a first object class;

(c) generating a second plurality of objects by linking a second plurality of the pixel locations to the second plurality of objects, wherein the second plurality of objects forms a second object network, wherein a third plurality of the pixel locations includes those pixel locations that belong both to the first plurality and to the second plurality of the pixel locations;

(d) linking each of the second plurality of objects to a second object class;

(e) generating a third object network by imprinting the first plurality of objects into the second plurality of objects such that the third plurality of pixel locations are relinked from objects of the second plurality of objects to objects of the third object network; and

(f) determining a characteristic of an object of the third object network by measuring the object.

24. A method comprising:

(a) specifying a first object class, wherein a first plurality of objects belongs to the first object class;

(b) specifying a second object class, wherein a second plurality of objects belongs to the second object class;

(c) acquiring pixel values that are associated with pixel locations;

(d) generating the first plurality of objects by linking a first plurality of the pixel locations to the first plurality of objects, wherein the first plurality of objects forms a first object network;

(e) linking each of the first plurality of objects to the first object class;

(f) generating the second plurality of objects by linking a second plurality of the pixel locations to the second plurality of objects, wherein the second plurality of objects forms a second object network, wherein a third plurality of the pixel locations includes those pixel locations that belong both to the first plurality and to the second plurality of the pixel locations;

(g) linking each of the second plurality of objects to the second object class;

(h) generating a third object network by imprinting the first plurality of objects into the second plurality of objects such that objects of the first plurality of objects that become part of the third object network are linked to the third plurality of pixel locations, and objects of the second plurality of objects that become part of the third object network are not linked to the third plurality of pixel locations; and

(i) determining a characteristic of an object of the third object network by measuring the object.

25. The method of claim 24, wherein the first plurality of pixel locations has a first resolution and the second plurality of pixel locations has a second resolution, and wherein the first resolution is different than the second resolution.

26. The method of claim 24, wherein the third plurality of pixel locations and the second plurality of pixel locations have equal resolutions, wherein the second plurality of pixel locations has a higher resolution than does the first plurality of pixel locations, and wherein those pixel locations that belong both to the first plurality and to the second plurality of the pixel locations include all those pixel locations of the third plurality of pixel locations that fall within any pixel location of the first plurality of pixel locations.

27. The method of claim 24, further comprising:

generating a fourth object network by modifying the third object network such that each object of the second plurality of objects that is linked to pixel locations that are not contiguous to one another is divided into objects of the fourth object network that are linked only to contiguous pixel locations.

28. The method of claim 24, wherein the generating the third object network in (h) is performed by imprinting a subset of the first plurality of objects into the second plurality of objects, wherein the subset includes only those objects of the first plurality of objects for which a minimum predetermined proportion of the pixel locations that are linked to those objects are also linked to objects of the second plurality of objects.

29. The method of claim 24, wherein the imprinting in (h) involves scaling, rotating and translating an object of the first object network before imprinting the first plurality of objects into the second plurality of objects.

30. The method of claim 24, wherein the generating the third object network in (h) is performed by imprinting a subset of the first plurality of objects into the second plurality of objects, and wherein the subset includes only those pixel locations of a three-dimensional digital image that are linked to objects of the first plurality of objects that fall within a polyhedral region of the three-dimensional digital image.

31. The method of claim 24, wherein the generating the first plurality of objects in (d) is performed using one processor of a computer at the same time that the generating the second plurality of object in (f) is performed using another processor of the computer.

32. The method of claim 24, wherein the second object network is stored in a memory space, and wherein the second object network is overwritten by the third object network in the memory space.

33. The method of claim 24, wherein the first object class represents road segments, wherein each road segment has a length, and wherein the measuring the object in (i) measures the length of the road segment.

34. The method of claim 24, wherein the first object class represents houses, wherein each house has a roof, and wherein the measuring in (i) measures the area of the roof.

35. A method comprising:

specifying a first object class, wherein a first plurality of objects belongs to the first object class;

specifying a second object class, wherein a second plurality of objects belongs to the second object class;

selecting a first process step to be added to an image analysis script, wherein the first process step performs a first segmentation of objects of an object network into the first object class, wherein the first process step is selected from a database of process steps;

selecting a second process step to be added to the image analysis script, wherein the second process step performs a second segmentation of the objects of the object network into the second object class;

saving the image analysis script to a memory;

acquiring pixel values that are associated with pixel locations of a digital image; and

running the image analysis script, wherein the image analysis script;

executes the first process step by linking a first plurality of the pixel locations to the first plurality of objects;

executes the second process step by linking a second plurality of the pixel locations to the second plurality of objects, wherein a third plurality of the pixel locations includes those pixel locations that belong both to the first plurality and to the second plurality of the pixel locations;

imprints the first plurality of objects into the second plurality of objects such that each of the third plurality of the pixel locations is unlinked from an object of the second plurality of objects and remains linked to an object of the first plurality of objects; and

determines a characteristic of an object linked to pixel locations of the third plurality of the pixel locations by measuring the object.

36. The method of claim 35, wherein the image analysis script imprints only a subset of the first plurality of objects into the second plurality of objects, and wherein the subset includes only those objects of the first plurality of objects that fulfill a condition.

37. The method of claim 35, wherein the first process step is executed on one processor of a computer at the same time that the second process step is executed on another processor of the computer.

38. A system comprising:

a display on which a digital image having pixel locations is displayed, wherein an object corresponds to a first plurality of the pixel locations on the digital image; and

means for measuring the object, wherein the object is part of a first object network, wherein the means generates a second object network by linking a second plurality of the pixel locations to objects of the second object network, wherein the means generates a third object network by linking a third plurality of the pixel locations to objects of the third object network, wherein a fourth plurality of the pixel locations includes those pixel locations that belong both to the second plurality and to the third plurality of the pixel locations, and wherein the means generates the first object network by imprinting the objects of the second object network into the objects of the third object network such that the fourth plurality of the pixel locations are unlinked from objects of the third object network and remain linked to objects of the second object network.

39. The system of claim 38, wherein the means displays an outline of the object in the digital image on the display.

40. The system of claim 38, wherein the means is executed as a script on a processor.