| 摘要 |
PROBLEM TO BE SOLVED: To reduce the amount of calculation for a conversion between a depth and an index number using a depth lookup table (DLT) in multi-view image encoding/decoding in which prediction is applied to the depth when an SIMD operation or the like of a CPU is utilized.SOLUTION: A conversion similar to that using a DLT is implemented by a limited number of operation instructions selected from the group consisting of addition, subtraction, multiplication, shift operation, and conditional branching. On this occasion, a depth value appearing in a multi-view image with a predetermined length is checked on the encoder side, information relating to a DLT is found and approximated by a line graph, and the limited number of operation instructions for implementing the conversion similar to that using the DLT are configured on the basis of a linear function constituting the graph. |
