Method and apparatus for displaying multi-viewpoint video, displaying device, medium and program

A method for displaying a multi-viewpoint video, wherein the method comprising:

• acquiring different viewpoint-position combinations of at least two users located within a visible area of a screen;
• from a predetermined multi-viewpoint-image set, screening out multi-viewpoint images corresponding to each of viewpoint-position combinations;
• regarding visible viewpoint images of at least the corresponding viewpoint-position combinations among the multi-viewpoint images as target viewpoint images;
• combining the target viewpoint images of at least two viewpoint-position combinations as video frames, to obtain the multi-viewpoint video;
• and
• displaying the multi-viewpoint video by using the screen, wherein
• each of the viewpoint-position combinations comprises a left-eye viewpoint position and a right-eye viewpoint position,
• the step of combining the target viewpoint images of the at least two viewpoint-position combinations as the video frames, to obtain the multi-viewpoint video comprises:
• determining, among the multi-viewpoint images, a left-eye viewpoint image corresponding to the left-eye viewpoint position, and a right-eye viewpoint image corresponding to the right-eye viewpoint position;
• in the multi-viewpoint image, replacing a viewpoint image located at a viewpoint position on the left of the left-eye viewpoint position with the left-eye viewpoint image, and replacing a viewpoint image located at a viewpoint position on the right of the right-eye viewpoint position with the right-eye viewpoint image, to obtain a source-material multi-viewpoint image;
• and
• regarding visible viewpoint images of at least the corresponding viewpoint-position combinations in the source-material multi-viewpoint image as the target viewpoint images,
• after the step of obtaining the source-material multi-viewpoint image, the method further comprises replacing, in the source-material multi-viewpoint image, a viewpoint image located between the left-eye viewpoint position and the right-eye viewpoint position with the right-eye viewpoint image or the left-eye viewpoint image,
• the step of regarding the visible viewpoint images of at least the corresponding viewpoint-position combinations in the source-material multi-viewpoint image as the target viewpoint images comprises replacing a visible viewpoint image of a viewpoint-position combination other than the corresponding viewpoint-position combinations in the source-material multi-viewpoint image with a predetermined hidden image, to obtain the target viewpoint images,
• the step of acquiring the different viewpoint-position combinations of the at least two users located within the visible area of the screen comprises:
• by using an image collecting device, acquiring a first user image within the visible area of the screen;
• and
• identifying human faces in the first user image, to obtain viewpoint-position combinations corresponding to eyeballs of the at least two users, and
• after the step of obtaining the viewpoint-position combinations corresponding to the eyeballs of the at least two users, the method further comprises:
• by using the image collecting device, re-acquiring a second user image;
• and
• when a similarity between the second user image and the first user image is greater than a similarity threshold, performing human-face identification to the second user image, to update the viewpoint-position combinations.