Image processing apparatus, image processing method, and image processing program

An image processing method comprising:

• acquiring a plurality of images of a subject captured by a camera with overlapping imaging ranges from a plurality of viewpoints;
• detecting damage to a surface of the subject from the acquired plurality of images;
• measuring a size of the detected damage;
• analyzing each of the acquired plurality of images to estimate a position and a posture of the camera at which each image has been captured, and to generate three-dimensional point group data of feature points;
• generating a three-dimensional patch model of the subject on the basis of the generated point group data;
• selecting an image corresponding to each patch of the three-dimensional patch model from the acquired plurality of images;
• extracting an image applied to each patch from the selected image corresponding to each patch;
• calculating, on the basis of the estimated position and posture of the camera at which the extracted image has been captured, a correction value for correcting a measurement result of the size of the detected damage in the extracted image;
• and
• correcting the measurement result of the size of the detected damage in the extracted image by the calculated correction value.

A non-transitory, computer-readable tangible recording medium which records thereon a program for causing, when read by a computer, the computer to execute:

• acquiring a plurality of images of a subject captured by a camera with overlapping imaging ranges from a plurality of viewpoints;
• detecting damage to a surface of the subject from the acquired plurality of images;
• measuring a size of the detected damage;
• analyzing each of the acquired plurality of images to estimate a position and a posture of the camera at which each image has been captured, and to generate three-dimensional point group data of feature points;
• generating a three-dimensional patch model of the subject on the basis of the generated point group data;
• selecting an image corresponding to each patch of the three-dimensional patch model from the acquired plurality of images;
• extracting an image applied to each patch from the selected image corresponding to each patch;
• calculating, on the basis of the estimated position and posture of the camera at which the extracted image has been captured, a correction value for correcting a measurement result of the size of the detected damage in the extracted image;
• and
• correcting the measurement result of the size of the detected damage in the extracted image by the calculated correction value.