Autostereo tapestry representation

A computer-based method for generating at least two views of a scene from a tapestry image, each of the at least two views being associated with one desired viewpoint of the scene, the method comprising:

• providing a leftmost view image of a scene;
• providing a rightmost view image of the scene;
• forming, by a computer, a disparity map, wherein the disparity map comprises distance information between the leftmost view image and the rightmost view image;
• forming, by a computer, an occlusion map, wherein the occlusion map comprises information on pixels visible in the rightmost view image but not visible in the leftmost view image, wherein disoccluded regions are not filled;
• forming, by a computer, an autostereo tapestry image by inserting in the leftmost view image, based on the occlusion map, wherein disoccluded regions are not filled, the pixels visible in the rightmost view image but not visible in the leftmost view image;
• forming, by a computer, a left-shift displacement map, wherein the left-shift displacement map comprises distance information between the leftmost view image and the autostereo tapestry image;
• forming, by a computer, a right-shift displacement map, wherein the right-shift displacement map comprises distance information between the rightmost view image and the autostereo tapestry image;
• providing the autostereo tapestry image;
• providing the left-shift displacement map;
• providing a display parameter F, wherein the display parameter F determines a target view position between a leftmost view position and a rightmost view position;
• and
• rendering, by a computer, the target view position, wherein forming the autostereo tapestry image comprises configuring the autostereo tapestry image to allow derivation of the target view position.

An encoding system, the system comprising:

• a tapestry generator adapted to generate a tapestry image, a left-displacement map and a right-displacement map based on a disparity map, an occlusion map and a plurality of views comprising at least a leftmost and a rightmost input views, according to steps comprising:
• providing a leftmost view image of a scene;
• providing a rightmost view image of the scene;
• forming, by a computer, the disparity map, wherein the disparity map comprises distance information between the leftmost view image and the rightmost view image;
• forming, by a computer, the occlusion map, wherein the occlusion map comprises information on pixels visible in the rightmost view image but not visible in the leftmost view image, wherein disoccluded regions are not filled;
• forming, by a computer, the tapestry image by inserting in the leftmost view image, based on the occlusion map, wherein disoccluded regions are not filled, the pixels visible in the rightmost view image but not visible in the leftmost view image;
• forming, by a computer, the left-shift displacement map, wherein the left-shift displacement map comprises distance information between the leftmost view image and the tapestry image;
• and
• forming, by a computer, the right-shift displacement map, wherein the right-shift displacement map comprises distance information between the rightmost view image and the tapestry image;
• and
• an encoder adapted to encode the tapestry image, the left-displacement map and the right-displacement map into a bitstream,
• wherein forming the autostereo tapestry image comprises configuring the autostereo tapestry image to allow derivation of at least one view image other than the leftmost view image, the at least one view image being the rightmost view image or an intermediate view image being intermediate between the leftmost and the rightmost view images.

A decoding system, the system comprising:

• a decoder adapted to decode a tapestry image, a left-shift displacement map and a right-shift displacement map, wherein the tapestry image, the left-shift displacement map and the right-shift displacement map are based on one or more input views, according to steps comprising:
• providing a leftmost view image of a scene;
• providing a rightmost view image of the scene;
• forming, by a computer, a disparity map, wherein the disparity map comprises distance information between the leftmost view image and the rightmost view image;
• forming, by a computer, an occlusion map, wherein the occlusion map comprises information on pixels visible in the rightmost view image but not visible in the leftmost view image, wherein disoccluded regions are not filled;
• forming, by a computer, the tapestry image by inserting in the leftmost view image, based on the occlusion map and wherein disoccluded regions are not filled, the pixels visible in the rightmost view image but not visible in the leftmost view image;
• forming, by a computer, the left-shift displacement map, wherein the left-shift displacement map comprises distance information between the leftmost view image and the tapestry image;
• and
• forming, by a computer, the right-shift displacement map, wherein the right-shift displacement map comprises distance information between the rightmost view image and the tapestry image;
• and
• a view generation unit adapted to derive one or more output images of a scene based on the tapestry image, the left-shift displacement map and the right-shift displacement map.