Method and device for acquiring clothing image attribute points

A device for acquiring clothing image attribute points, characterized by comprising:

• a processor configured to
• save five attribute points manually calibrated, including, shoulder angle point at a left side and a shoulder angle point at right side, a midpoint of a lower hem, a midpoint of sleeve edges of a left sleeve and a midpoint of sleeve edges of a right sleeve, for a front presented clothing image;
• and
• determine first to eleventh points below in sequence by taking a horizontal direction as a horizontal coordinate and a vertical direction as a vertical coordinate, for a first side of two bilaterally symmetric sides of the clothing image:
• a first point being an upper end point at the first side;
• a second point being located on a shoulder edge at the first side, a horizontal coordinate of the second point being a midpoint of the first point and the shoulder angle point at the first side;
• a third point being an outside point of the sleeve edge at the first side;
• a fourth point being an inside point of the sleeve edge at the first side;
• a fifth point and a sixth point being respectively located on an outside edge in a sleeve at the first side and an inside edge in the sleeve at the first side, a connecting line of the fifth point and the sixth point being perpendicular to a connecting line of the shoulder angle point at the first side and the third point, a vertical coordinate of a perpendicular foot being located at a preset position between the shoulder angle point at the first side and the third point;
• a seventh point being an angle point of the lower hem at the first side;
• and
• eighth to eleventh points being located on a connecting line from a midpoint of the second point and the shoulder angle point at the first side to the seventh point, and being distributed at preset positions;
• and
• determining points symmetric with the first to eleventh points at a second side of the two bilaterally symmetric sides of the clothing image.