Composed prediction and restricted merge

A video encoding apparatus for encoding a video into a data stream using block-based predictive coding, the apparatus comprising:

• a processor;
• and
• a memory storing instructions that, when executed by the processor, cause the processor to:
• derive a collection of one or more primitive predictions for a block using a set of primitive prediction modes adapted for predicting blocks of a picture of the video, wherein each primitive prediction in the collection of primitive predictions is derived using a respective primitive prediction mode from the set of primitive prediction modes;
• compose a composite prediction signal for the block by combining primitive predictions from the collection of one or more primitive predictions;
• and
• predict the block using the composite prediction signal, wherein the memory stores instructions that, when executed by the processor, cause the processor to compose the composite prediction signal using a recurrence calculation:
• q1=p1
• qn+1=ƒn(qn,pνn+1, . . . ,Vn+1), for n=1 . . . K,
• wherein:
• q=qK+1 is the composite prediction signal;
• p1, . . . , pνk+1 is the collection of one or more primitive predictions derived using the set of primitive prediction modes;
• ν1=1 and νn+1=vn+mn, mn is the number of primitive prediction that ƒn depends on, m1+ . . . +mk=vk+1−1 and νk+1 is the cardinality of the collection of one or more primitive predictions;
• and
• ƒn is a function that maps an intermediate composite prediction signal qn and one or more primitive prediction signals pνn+1, . . . , pνn+1 to a new intermediate prediction signal qn+1.

A video decoding method for decoding a video from a data stream using block-based predictive coding, the method comprising:

• deriving a collection of one or more primitive predictions for a block using a set of primitive prediction modes adapted for predicting blocks of a picture of the video, wherein each primitive prediction in the collection of primitive predictions is derived using a respective primitive prediction mode from the set of primitive prediction modes;
• composing a composite prediction signal for the block by combining primitive predictions from the collection of one or more primitive predictions;
• and
• predicting the block using the composite prediction signal, further comprising composing the composite prediction signal using a recurrence calculation:
• q1=p1
• qn+1=ƒn(qn,pνn+1, . . . ,Vn+1), for n=1 . . . K,
• wherein:
• q=qK+1 is the composite prediction signal;
• p1, . . . , pνk+1 is the collection of one or more primitive predictions derived using the set of primitive prediction modes;
• ν1=1 and νn+1=vn+mn, mn is the number of primitive prediction that ƒn depends on, m1+ . . . +mk=vk+1−1 and νk+1 is the cardinality of the collection of one or more primitive predictions;
• and
• ƒn is a function that maps an intermediate composite prediction signal qn and one or more primitive prediction signals pνn+1, . . . , pνn+1 to a new intermediate prediction signal qn+1.

A video encoding method for encoding a video into a data stream using block-based predictive coding, the method comprising:

• deriving a collection of one or more primitive predictions for a block using a set of primitive prediction modes adapted for predicting blocks of a picture of the video, wherein each primitive prediction in the collection of primitive predictions is derived using a respective primitive prediction mode from the set of primitive prediction modes;
• composing a composite prediction signal for the block by combining primitive predictions from the collection of one or more primitive predictions;
• and
• predicting the block using the composite prediction signal, further comprising composing the composite prediction signal using a recurrence calculation:
• q1=p1
• qn+1=ƒn(qn,pνn+1, . . . ,Vn+1), for n=1 . . . K,
• wherein:
• q=qK+1 is the composite prediction signal;
• p1, . . . , pνk+1 is the collection of one or more primitive predictions derived using the set of primitive prediction modes;
• ν1=1 and νn+1=vn+mn, mn is the number of primitive prediction that ƒn depends on, m1+ . . . +mk=vk+1−1 and νk+1 is the cardinality of the collection of one or more primitive predictions;
• and
• ƒn is a function that maps an intermediate composite prediction signal qn and one or more primitive prediction signals pνn+1, . . . , pνn+1 to a new intermediate prediction signal qn+1.