Feature extracting method, subject recognizing method and image processing apparatus

• US 7,123,761 B2
• Filed: 11/07/2002
• Issued: 10/17/2006
• Est. Priority Date: 11/20/2001
• Status: Expired due to Fees

• Current Assignee: Konica Corporation (Konica Minolta Inc.)
• Original Assignee: Konica Corporation (Konica Minolta Inc.)
• Inventors: Kawano, Tsutomu
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): MACKOWEY, ANTHONY M
• Application Number:
  US10/289,981
• Publication Number:
  US 20030095698A1
• Time in Patent Office: 1,440 Days
• Field of Search: 382/128, 382/130, 382/132, 382/170, 382/171, 382/181, 382/190, 382/195, 382/199, 382/203, 382/224
• US Class Current: 382/132
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20068 Projection on vertical or horizontal image axis
• CPC Class Codes: G06T 2207/30012 Spine; Backbone
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 7/12 Edge-based segmentation

Abstract

A feature extracting method for a radiation image formed by radiation image signals each corresponding to an amount of radiation having passed through a radiographed subject, has plural different feature extracting steps, each of the plural different feature extracting steps having a respective feature extracting condition to extract a respective feature value; a feature value evaluating step of evaluating a combination of the plural different feature values; and a controlling step of selecting at least one feature extracting step from the plural different feature extracting steps based on an evaluation result by the feature value evaluating step, changing the feature extracting condition of the selected feature extracting step and conducting the selected feature extracting step so as to extract a feature value again based on the changed feature extracting condition from the radiation image.

Claims

1. A feature extracting method for a radiation image formed by radiation image signals each corresponding to an amount of radiation that has passed through a radiographed subject, said method comprising:

a plurality of different feature extracting steps, each of which has a respective feature extracting condition to extract a respective feature value so that a plurality of different feature values are extracted from the radiation image by the plurality of different feature extracting steps;

a feature value evaluating step of evaluating a combination of the plurality of different feature values extracted by the plurality of different feature extracting steps; and

a controlling step of;

selecting at least one feature extracting step from the plurality of different feature extracting steps based on an evaluation result by the feature value evaluating step, changing the feature extracting condition of the selected feature extracting step, and conducting the selected feature extracting step again so as to extract from the radiation image a feature value based on the changed feature extracting condition;

wherein the controlling step includes investigating the coordination of the combination of the plurality of different feature values, and when it is judged in the controlling step that the combination of the plurality of different feature values is not coordinated, the feature value to be extracted again is selected based on the uncoordinated combination of the plurality of different feature values and conducts the selected feature extracting step again.

2. The feature extracting method of claim 1, further comprising:

a subject region extracting step of extracting a subject region representing the radiographed subject from a radiographed image region,

wherein the plurality of different feature values are extracted from the subject region extracted by the subject region extracting step.

3. The feature extracting method of claim 2, wherein the plurality of different feature extracting steps comprises an outline recognizing step of extracting a feature value based on an outline of the subject region which is obtained by investigating an outline of the subject region.

4. The feature extracting method of claim 1, wherein the plurality of different feature extracting steps comprises an edge pattern recognizing step of investigating signal intensity variations among neighboring pixels around an arbitrarily selected pixel on the radiation image, determining an arbitrarily selected pixel having a large signal intensity variation as an edge pixel, and extracting a feature value based on a distribution of edge pixels.

5. The feature extracting method of claim 1, wherein the plurality of different feature extracting steps comprises a lung field region detecting step of investigating whether or not a lung field region exists on the radiation image and extracting a feature value based on a result of the investigation.

6. The feature extracting method of claim 1, wherein when it is recognized in the controlling step that there is a determinant feature to specify the radiographed subject in the combination of the plurality of different feature values, no feature is extracted again in the controlling step.

7. The feature extracting method of claim 1, wherein the controlling step includes consulting a list of uncoordinated combinations of a plurality of different feature values, and a feature value to be selected to be extracted again is predetermined for each of the uncoordinated combinations.

8. The feature extracting method of claim 1, wherein the feature extracting condition of each of the plurality of different feature extracting steps includes a plurality of threshold conditions to determine a feature value, and wherein the feature extracting condition changing in the controlling step comprises the threshold condition of the selected feature extracting step and the selected feature extracting step is conducted again so as to extract the feature value by using the changed threshold condition.

9. A method of recognizing a subject from a radiation image formed by radiation image signals each corresponding to an amount of radiation that has passed through a radiographed subject, said method comprising:

a plurality of different feature extracting steps, each of which has a respective feature extracting condition to extract a respective feature value so that a plurality of different feature values are extracted from the radiation image by the plurality of different feature extracting steps;

a feature value evaluating step of evaluating a combination of the plurality of different feature values extracted by the plurality of different feature extracting steps;

a controlling step of;

selecting at least one feature extracting step from the plurality of different feature extracting steps based on an evaluation result by the feature value evaluating step, changing the feature extracting condition of the selected feature extracting step, and conducting the selected feature extracting step again so as to extract from the radiation image a feature value based on the changed feature extracting condition; and

a recognizing step of recognizing a radiographing orientation and a radiographed part of the subject on the radiation image by using the plurality of different feature values;

wherein the controlling step includes investigating the coordination of the combination of the plurality of different feature values, and when it is judged in the controlling step that the combination of the plurality of different feature values is not coordinated, the feature value to be extracted again is selected based on the uncoordinated combination of the plurality of different feature values and conducts the selected feature extracting step again.

10. The method of claim 9, wherein the plurality of different feature extracting steps comprises an outline recognizing step, an edge pattern recognizing step and a lung field region detecting step.

11. The method of claim 9, wherein when it is recognized in the controlling step that there is a determinant feature to specify the radiographed subject in the combination of the plurality of different feature values, no feature is extracted again in the controlling step.

12. A computer system for conducting a feature extracting method for a radiation image formed by radiation image signals each corresponding to an amount of radiation that has passed through a radiographed subject, the computer system including a program that is executable by the computer system to cause the computer system to execute functions comprising:

a plurality of different feature extracting steps, each of which has a respective feature extracting condition to extract a respective feature value so that a plurality of different feature values are extracted from the radiation image by the plurality of different feature extracting steps;

a feature value evaluating step of evaluating a combination of the plurality of different feature values extracted by the plurality of different feature extracting steps; and

a controlling step of;

selecting at least one feature extracting step from the plurality of different feature extracting steps based on an evaluation result by the feature value evaluating step, changing the feature extracting condition of the selected feature extracting step, and conducting the selected feature extracting step again so as to extract from the radiation image a feature value based on the changed feature extracting condition;

wherein the controlling step includes investigating the coordination of the combination of the plurality of different feature values, and when it is judged in the controlling step that the combination of the plurality of different feature values is not coordinated, the feature value to be extracted again is selected based on the uncoordinated combination of the plurality of different feature values and conducts the selected feature extracting step again.

13. The computer system of claim 12, wherein the plurality of different feature extracting steps comprises an outline recognizing step, an edge pattern recognizing step and a lung field region detecting step.

14. The computer system of claim 12, wherein when it is recognized in the controlling step that there is a determinant feature to specify the radiographed subject in the combination of the plurality of different feature values, no feature is extracted again in the controlling step.

15. An image processing apparatus for processing a radiation image formed by radiation image signals each corresponding to an amount of radiation having that has passed through a radiographed subject, said apparatus comprising:

a plurality of different feature extracting sections, each of which has a respective feature extracting condition to extract a respective feature value so that a plurality of different feature values are extracted from the radiation image by the plurality of different feature extracting sections;

a feature value evaluating section to evaluate a combination of the plurality of different feature values extracted by the plurality of different feature extracting sections; and

a controlling section to select at least one feature extracting section from the plurality of different feature extracting sections based on an evaluation result by the feature value evaluating section, to change the feature extracting condition of the selected feature extracting section, and to control the selected feature extracting section so as to extract from the radiation image a feature value again based on the changed feature extracting condition;

wherein the controlling section investigates the coordination of the combination of the plurality of different feature values and when the controlling section judges that the combination of the plurality of different feature values is not coordinated, the controlling section selects the feature value to be extracted again based on the uncoordinated combination of the plurality of different feature values and controls the selected feature extracting section to select the selected feature value again.

16. The image processing apparatus of claim 15, further comprising:

a subject region extracting section of which extracts a subject region representing the radiographed subject from a radiographed image region,

wherein the plurality of different feature values are extracted from the subject region extracted by the subject region extracting section.

17. The image processing apparatus of claim 15, wherein the plurality of different feature extracting sections comprises an outline recognizing section which extracts a feature value based on an outline of the subject region which is obtained by investigating the outline of the subject region.

18. The image processing apparatus of claim 15, wherein the plurality of different feature extracting sections comprises an edge pattern recognizing section which investigates signal intensity variations among neighboring pixels around an arbitrarily selected pixel on the radiation image, determines an arbitrarily selected pixel having a large signal intensity variation as an edge pixel, and extracts a feature value based on a distribution of edge pixels.

19. The image processing apparatus of claim 15, wherein the plurality of different feature extracting sections comprises a lung field region detecting section which investigates whether or not a lung field region is present on the radiation image and extracts a feature value based on a result of the investigation.

20. The image processing apparatus of claim 15, wherein when the controlling section recognizes that there is a determinant feature to specify the radiographed subject in the combination of the plural different feature values, the controlling section a does not extract a feature value again.

21. The image processing apparatus of claim 15, wherein the controlling section includes a list of uncoordinated combinations of a plurality of different feature values, and a feature value to be selected to be extracted again is predetermined for each of the uncoordinated combinations.

22. The image processing apparatus of claim 15, wherein the feature extracting condition of each of the plurality of different feature extracting sections includes a plurality of threshold conditions to determine a feature value, and wherein the controlling section changes the threshold condition of the selected feature extracting section and controls the selected feature extracting section so as to extract the feature value again by using the changed threshold condition.