Calibration of radiographic images
AssigneeBrainlab AG
InventorsZeev Russak, Zeev Glozman
Life of this patent
Estimating the term…
Abstract
Apparatus for calibrating an image, the image comprising a spherical marker, the apparatus comprising: an image receiver, configured to receive the image, and a marker finder, associated with the image receiver and configured to automatically find an image of the marker in the received image.
Claims27
Apparatus for calibration of radiographic images, the images comprising a spherical calibration marker appearing therein, the apparatus implemented on an electronic processor and comprising:
- a) an image receiver, configured to receive a radiographic image with said spherical calibration marker appearing therein;
- and
- b) a marker finder, associated with said image receiver and configured to automatically find an image of the spherical calibration marker in the received image by:
- edge detecting the image,
- forming a binary image from the edge detection, the binary image being formed by defining pixels therein either as a first value indicating belonging to an edge or to a second value indicating not belonging to an edge;
- pixels in said binary image corresponding to pixels detected as edges being set to said first value, and
- testing respective pixels of said binary image to determine whether a circle of said pixels of said first value belonging to an edge is described around a respective pixel in said binary image, and if so, setting said edge as a detected circle for calibration, wherein said testing comprises:
- stepwise around said pixels belonging to said edge selecting pairs of pixels and a presumed center pixel and following said selecting, pairwise testing said pairs of pixels for having a radius about said presumed center pixel, setting a key and a counter for each combination of center and radius and applying a respective counter for pixel pairs that share a same radius and center, and then, using said keys and counters, selecting as the detected circle that circle on which lies the largest number of said pairs of pixels which share a same center and radius;
- and
- c) a calibrator associated with the marker finder, for using said detected described circle and respective pixel as a centered circle and calibrating said image.
Method for calibrating radiographic images, the images comprising a spherical calibration marker appearing therein, the method comprising:
- a) receiving the image;
- and
- b) automatically finding an image of the spherical calibration marker in the received image, said finding comprising
- edge detecting the image,
- forming a binary image from the edge detection, by setting each pixel corresponding to a detected edge to a first value indicating an edge, all other pixels being set to a second value indicating non-edge,
- testing respective pixels of said binary image to determine a circle that is described by set pixels set as edge pixels, around said respective pixel in said binary image, wherein said testing comprises:
- stepwise around said edge pixels selecting pairs of said edge pixels and a presumed center pixel;
- subsequent to said selecting, pairwise testing said selected pixel pairs for sharing a radius about said presumed center pixel;
- from the tested pairs, setting keys and using counters respectively for each instance of a shared center and radius, and selecting a largest number of said pairs which share a same center and radius;
- and
- setting the center and radius shared by said largest number of pixels as said described circle, and
- c) calibrating said image based on said described circle.
Method for calibrating an image, the image comprising a spherical marker, the method comprising:
- a) taking a picture of a person with the spherical marker positioned in proximity of said person, thereby generating the image;
- b) automatically finding an imprint of the spherical marker in the image;
- and
- c) calibrating the image utilizing said found image of the spherical marker, wherein said automatically finding comprises:
- edge detecting the image,
- forming a binary image from the edge detection by setting each pixel corresponding to a detected edge to a first value indicating an edge, all other pixels being set to a second value indicating non-edge,
- pairwise testing pixels of said detected edge to determine whether a circle is described by respective pixels of a given pair set to the value of edge, around a presumed center pixel in said binary image, wherein said testing comprises:
- stepwise around said edge pixels selecting pairs of said detected edge pixels and said presumed center pixel;
- and
- subsequent to said selecting, carrying out said pairwise testing of said pairs of edge pixels and said presumed center for sharing a radius;
- setting a key for each presumed center and radius;
- using counters per key to count each pair sharing a radius and a center;
- from said counters, finding a single circle shared by a largest number of said pairs of edge pixels and said presumed center;
- selecting said single circle as said described circle, with said presumed center pixel as a circle center, and
- calibrating said image using said described circle.
Dependencies are inferred from the claim text.