Automated landmark extraction from three-dimensional whole body scanned data

• US 20060171590A1
• Filed: 08/05/2005
• Published: 08/03/2006
• Est. Priority Date: 12/09/2004
• Status: Active Grant

• Current Assignee: National Tsing Hua University
• Original Assignee: National Tsing Hua University
• Inventors: Wang, Mao-Jun, Lu, Jun-Ming
• Granted Patent: US 7,561,726 B2
• Time in Patent Office: Days
• US Class Current: 382/190
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

A method for automated landmark extraction from three-dimensional whole body scanned data extracts feature points as landmarks from a plurality of point clouds obtained from a three-dimensional whole body scanner, without pre-making the landmarks on a human body before scanning. The system includes: (1) Silhouette analysis, which projects three-dimensional body scanned data onto a two-dimensional surface and observes the variations in curvature and depth; (2) Minimum circumference, which uses the variations of the circumference of body parts to define the locations of the landmarks and feature lines; (3) Gray-scale detection, which converts the color information of human body from RGB values into gray-scale values to locate the landmarks with greater variations in brightness; and (4) Human-body contour plot, which simulates the sense of touch to locate landmarks by finding the prominent and concave parts on the human body.

Claims

1. A method for automated landmark extraction from three-dimensional whole body scanned data, said method comprising the steps of:

extracting feature points as landmarks on a human body; and

without pre-making landmarks before scanning, directly extracting crotch, intersection of neck and shoulder, acromion, chest line, hip line, and rearmost of hip based on three-dimensional changes in shape of the human body.

2. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 1, further comprising:

utilizing silhouette analysis wherein three-dimensional body scanned data is projected onto the coronal plane and the landmarks including crotch and intersection of neck and shoulder are extracted by locating the locally minimum points on the silhouette curves.

3. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 2, wherein, when utilizing to locate the intersection of neck and shoulder in which all points between neck-top and acromion are taken into consideration, the point with greatest vertical distance to a line connected by neck-top and acromion being extracted as intersection of neck and shoulder.

4. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 1, further comprising:

utilizing silhouette analysis, wherein the three-dimensional body scanned data is projected onto the coronal plane and acromion is extracted by locating the locally maximum point on the silhouette curves of the shoulder section.

5. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 1, further comprising:

utilizing silhouette analysis, wherein the three-dimensional body scanned data is projected onto the sagittal plane and the rearmost of hip is extracted by locating the locally maximum point on the silhouette curves of the hip section.

6. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 1, further comprising:

utilizing silhouette analysis, wherein the three-dimensional body scanned data is projected onto the sagittal plane and the chest line and hip line are extracted by locating the layers with the greatest depths on the silhouette curves.

7. A method for automated landmark extraction from three-dimensional whole body scanned data, said method comprising the steps of:

extracting feature points as landmarks on a human body; and

without pre-making the landmarks before scanning, directly extracting wrist and waist line based on three-dimensional changes in shape of the human body.

8. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 7, further comprising:

using a minimum circumference on the human body for extracting the wrist and waist line.

9. A method for automated landmark extraction from three-dimensional whole body scanned data, said method comprising the steps of:

extracting feature points as landmarks on a human body; and

without pre-making the landmarks before scanning, directly extracting bust points based on three-dimensional changes in shape of the human body.

10. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 9, wherein, when utilizing to extract bust points for male subjects, color information of the human body is converted from RGB values into gray-scale values and “

visually darker parts”

are located to find parts with noticeable variations in brightness to extract bust points for male subjects.

11. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 9, wherein, when utilizing for extraction of bust points for female subjects by using human-body contour plots, sense of touch is simulated by locating prominent parts in a chest section of the human body to extract bust points for female subjects.

12. A method for automated landmark extraction from three-dimensional whole body scanned data, said method comprising the steps of:

extracting feature points as landmarks on a human body; and

without pre-making the landmarks before scanning, directly extracting an armpit based on three-dimensional changes in shape of the human body.

13. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 12, further comprising:

utilizing silhouette analysis, wherein the three-dimensional body scanned data is projected onto the coronal plane and the approximate locations of the armpits are extracted by locating locally maximum points on the silhouette curves of the intersection of torso and arms.

14. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 13, further comprising:

utilizing gray-scale detection, wherein color information of the human body data is converted from RGB values into gray-scale values and “

visually darker parts”

is located to find parts with noticeable variations in brightness;

integrating with the silhouette analysis; and

further searching near approximate locations of the armpits and then extracting a highest point in neighboring areas with noticeable variations in brightness as the armpits.

15. A method for automated landmark extraction from three-dimensional whole body scanned data, said method comprising the steps of:

extracting feature points as landmarks on a human body; and

without pre-making the landmarks before scanning, directly extracting a seventh cervical point, inferior angle of scapula, patella, lateral malleolus, elbow, suprasternale, and navel based on three-dimensional changes in shape of the human body.

16. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 15, wherein, wherein utilizing for extraction of various landmarks by using human-body contour plots, a sense of touch is simulated and prominent parts of a human body are located to extract locations of the seventh cervical point, inferior angle of scapula, patella, lateral malleolus, and elbow.

17. The method for automated landmark extraction from three-dimensional whole body scanned data as defined in claim 15, further comprising:

utilizing human-body contour plots for extraction, wherein sense of touch is simulated and concave parts of the human body are located to extract the locations of the suprasternale and navel.