Orientation detection for chest radiographic images
AssigneeCarestream Health, Inc.
InventorsRuiping Li, Zhimin Huo
Life of this patent
Estimating the term…
Abstract
A method for identifying the orientation of a radiographic image of a patient generates a reduced-scale image having a smaller number of pixel elements than the original radiographic image and generates a reduced bit-depth image from the reduced-scale image. The method determines whether the image has portrait or landscape orientation by sectioning either the reduced-scale image or the reduced bit-depth image into at least two sections and calculating at least one value related to relative symmetry for each of the at least two sections.
Claims18
A method for identifying the orientation of a radiographic image of a patient, the method comprising:
- generating a reduced-scale image having a smaller number of pixel elements than the original radiographic image;
- generating a reduced bit-depth image from the reduced-scale image;
- and
- determining whether the image has portrait or landscape orientation by:
- sectioning either the reduced-scale image or the reduced bit-depth image into at least two sections;
- and
- calculating at least one value related to relative symmetry for each of the at least two sections.
A method for identifying the orientation of a radiographic image of a patient, the method comprising:
- generating a reduced-scale image having a smaller number of pixel elements than the original radiographic image by sub-sampling;
- and
- determining whether the image has portrait or landscape orientation by:
- sectioning the reduced-scale image into at least two sections;
- and
- calculating at least one value related to relative symmetry for each of the at least two sections of the reduced-scale image.
A method for identifying the orientation of a radiographic image of a patient, the method comprising:
- a) re-sampling the image data to provide a down-sampled image;
- b) performing a clustering operation on the down-sampled image to generate a binary image;
- c) determining whether the image orientation is portrait or landscape according to symmetry of the down-sampled image provided in step a);
- and
- d) determining an upright position of the patient according to symmetry of the binary image generated in step b).
Dependencies are inferred from the claim text.