Method and apparatus for video coding
A method of processing point cloud data at a decoder can include receiving three dimensional (3D) coordinates of a set of points of a point cloud including first points and a current point. Each of the first points can be associated with a reconstructed attribute value. A group of neighboring points of the current point can be determined from the first points. An attribute distance for each of the neighboring points can be determined based on the reconstructed attribute values of the neighboring points. An attribute prediction of the current point can be determined based on the attribute distances of the neighboring points.
1. A method of processing point cloud data at a decoder, comprising:
receiving three dimensional (3D) coordinates of a set of points of a point cloud including first points and a current point, each of the first points being associated with a reconstructed attribute value;
determining a group of neighboring points of the current point from the first points;
determining an attribute distance for each of the neighboring points based on the reconstructed attribute values of the neighboring points, the attribute distance of the respective neighboring point being a deviation of the reconstructed attribute value of the respective neighboring point from a median or mean value of the reconstructed attribute values of the neighboring points;
determining a weight for the reconstructed attribute value of each of the neighboring points based on the attribute distance of the respective neighboring point, the weight for the reconstructed attribute value of the respective neighboring point being inversely proportional to the attribute distance of the respective neighboring point;
determining an attribute prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points; and
combining the attribute prediction of the current point with a residual signal of the current point to obtain a reconstructed attribute value of the current point.
2. The method of claim 1 , wherein the group of determined neighboring points of the current point are a subset of the first points that are nearest to the current point in terms of geometric distances.
3. The method of claim 1 , wherein the attribute prediction of the current point is one of:
one of the reconstructed attribute values of the neighboring points with a highest weight among the weights of the reconstructed attribute values of the neighboring points,
a weighted sum of the reconstructed attribute values of the neighboring points based on the weights of the reconstructed attribute values of the neighboring points, and
a weighted median of the reconstructed attribute values of the neighboring points based on the weights of the reconstructed attribute values of the neighboring points.
4. The method of claim 1 , wherein the determining the weight for the reconstructed attribute value of the respective neighboring point includes:
determining the weight for the reconstructed attribute value of the respective neighboring point based on a geometric distance of the respective neighboring point, the weight for the reconstructed attribute value of the respective neighboring point being inversely proportional to the attribute distance and the geographic distance of the respective neighboring point.
5. The method of claim 1 , wherein the determining the attribute prediction of the current point further comprises:
determining one of the reconstructed attribute values of the neighboring points with a highest weight to be the prediction of the current point.
6. The method of claim 5 , wherein the determining the attribute prediction of the current point further comprises:
when a measure of variability of the reconstructed attribute values of the neighboring points is above a threshold, determining the one of the reconstructed attribute values of the neighboring points with the highest weight to be the prediction of the current point.
7. The method of claim 1 , wherein the determining the attribute prediction of the current point further comprises:
determining a weighted sum of the reconstructed attribute values of the neighboring points to form the prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points.
8. The method of claim 1 , wherein the determining the attribute prediction of the current point further comprises:
determining a weighted median of the reconstructed attribute values of the neighboring points to be the prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points.
9. The method of claim 1 , further comprising:
performing a pruning process to remove the point having an inconsistent reconstructed attribute value from the neighboring points of the current point, the point having the inconsistent reconstructed attribute value having a lowest attribute based or bilateral weight among the weights of the reconstructed attribute values of the neighboring points of the current point, wherein
the determining the attribute distance includes determining the attribute distance for each of the remaining neighboring points after the pruning process is performed.
10. The method of claim 1 , further comprising:
performing a pruning process to remove the point having an inconsistent reconstructed attribute value from the neighboring points of the current point, the point having the inconsistent reconstructed attribute value having an attribute based weight and a geometric distance based weight that have a difference above a threshold, wherein
the determining the attribute distance includes determining the attribute distance for each of the remaining neighboring points after the pruning process is performed.
11. The method of claim 1 , further comprising:
receiving a flag syntax element indicating a prediction method selected from multiple prediction methods; and
determining the attribute prediction of the current point based on the attribute distances of the neighboring points with the indicated prediction method.
12. An apparatus of processing point cloud data, comprising circuitry configured to:
receive three dimensional (3D) coordinates of a set of points of a point cloud including first points and a current point, each of the first points being associated with a reconstructed attribute value;
determine a group of neighboring points of the current point from the first points;
determine an attribute distance for each of the neighboring points based on the reconstructed attribute values of the neighboring points, the attribute distance of the respective neighboring point being a deviation of the reconstructed attribute value of the respective neighboring point from a median or mean value of the reconstructed attribute values of the neighboring points;
determine a weight for the reconstructed attribute value of each of the neighboring points based on the attribute distance of the respective neighboring point, the weight for the reconstructed attribute value of the respective neighboring point being inversely proportional to the attribute distance of the respective neighboring point;
determine an attribute prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points; and
combine the attribute prediction of the current point with a residual signal of the current point to obtain a reconstructed attribute value of the current point.
13. The apparatus of claim 12 , wherein the group of determined neighboring points of the current point are a subset of the first points that are nearest to the current point in terms of geometric distances.
14. The apparatus of claim 12 , wherein the attribute prediction of the current point is one of:
one of the reconstructed attribute values of the neighboring points with a highest weight among the weights of the reconstructed attribute values of the neighboring points,
a weighted sum of the reconstructed attribute values of the neighboring points based on the weights of the reconstructed attribute values of the neighboring points, and
a weighted median of the reconstructed attribute values of the neighboring points based on the weights of the reconstructed attribute values of the neighboring points.
15. The apparatus of claim 12 , wherein the circuitry is further configured to:
determine the weight for the reconstructed attribute value of the respective neighboring point based on a geometric distance of the respective neighboring point, the weight for the reconstructed attribute value of the respective neighboring point being inversely proportional to the attribute distance and the geographic distance of the respective neighboring point.
16. The apparatus of claim 12 , wherein the circuitry is further configured to:
determine one of the reconstructed attribute values of the neighboring points with a highest weight to be the prediction of the current point.
17. The apparatus of claim 16 , wherein the circuitry is further configured to:
when a measure of variability of the reconstructed attribute values of the neighboring points is above a threshold, determine the one of the reconstructed attribute values of the neighboring points with the highest weight to be the prediction of the current point.
18. The apparatus of claim 12 , wherein the circuitry is further configured to:
determine a weighted sum of the reconstructed attribute values of the neighboring points to form the prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points.
19. The apparatus of claim 12 , wherein the circuitry is further configured to:
determine a weighted median of the reconstructed attribute values of the neighboring points to be the prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points.
20. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform a method of processing point cloud data, the method comprising:
receiving three dimensional (3D) coordinates of a set of points of a point cloud including first points and a current point, each of the first points being associated with a reconstructed attribute value;
determining a group of neighboring points of the current point from the first points;
determining an attribute distance for each of the neighboring points based on the reconstructed attribute values of the neighboring points, the attribute distance of the respective neighboring point being a deviation of the reconstructed attribute value of the respective neighboring point from a median or mean value of the reconstructed attribute values of the neighboring points;
determining a weight for the reconstructed attribute value of each of the neighboring points based on the attribute distance of the respective neighboring point, the weight for the reconstructed attribute value of the respective neighboring point being inversely proportional to the attribute distance of the respective neighboring point;
determining an attribute prediction of the current point based on the weights of the reconstructed attribute values of the neighboring points; and
combining the attribute prediction of the current point with a residual signal of the current point to obtain a reconstructed attribute value of the current point.