Method and apparatus of encoding/decoding point cloud geometry data using azimuthal coding mode
A method of encoding a point cloud into a bitstream of encoded point cloud data representing a physical object is provided. The method includes an azimuthal coding mode providing a series of bits for encoding a coordinate of a point of the point cloud. The method includes: dividing an interval, to which the point coordinate belongs to, into a left half interval and a right half interval; selecting a context based on an apparent angle (AA d ) representing an interval angle seen from a sensor that captured the point; and context-adaptive binary entropy encoding a bit (b d ) of the series of bits, into the bitstream, based on the selected context, the encoded bit (b d ) indicating which of the two half intervals the point coordinate belongs to.
1 . A method of encoding a point cloud into a bitstream of encoded point cloud data representing a physical object, the method comprising an azimuthal coding mode providing a series of bits for encoding a coordinate of a point of the point cloud, wherein the method comprises:
dividing an interval, to which the point coordinate belongs to, into a left half interval and a right half interval;
selecting a context based on an apparent angle (AA d ) representing an interval angle seen from a sensor that captured the point; and
context-adaptive binary entropy encoding a bit (b d ) of the series of bits, into the bitstream, based on the selected context, the encoded bit (b d ) indicating which of the two half intervals the point coordinate belongs to;
wherein the apparent angle (AA d ) is estimated based on at least one of a first angle (φ (node, d)) associated with a lower bound of the interval, a second angle (φ (top, d)) associated with an upper bound of the interval and a third angle (φ (middle, d)) associated with a middle point of the interval;
wherein the apparent angle (AA d ) is estimated based on the first angle (φ (node, d)) and the second angle (φ (top, d));
wherein the context is selected based on relative magnitudes of an elementary azimuthal angle (Δφ) and the apparent angle (AA d ).
2 . The method of claim 1 , wherein the apparent angle (AA d ) is estimated based on an interval size (s d ) and a distance (r) between the point and the sensor that captured the point.
3 . The method of claim 1 , wherein the context is selected based on a ratio between the elementary azimuthal angle (Δφ) over the apparent angle (AA d ).
4 . The method of claim 3 , wherein contexts are grouped into at least two context subsets based on a range of particular values of the ratio, and selecting the context comprises:
selecting a context subset based on the range of particular values and the ratio, and
selecting a context from the contexts of the selected context subset;
wherein each context subset corresponds to a different prediction quality level of the predicted azimuthal angle φ pred , where the corresponding prediction quality level is low when the ratio is much larger than 1 or much smaller than 1, and the corresponding prediction quality level is high when the ratio is close to 1;
wherein selecting a context from the contexts of the selected context subset comprise:
selecting a context subset from the at least two context subsets based on a comparison of the ratio with one or more thresholds; and
selecting a context from the selected context subset based on a comparison of the predicted azimuthal angle with a left angle (φ left,d ) and a right angle (φ right,d ).
5 . The method of claim 4 , wherein selecting the context from the contexts of the selected context subset depends on a predicted azimuthal angle (φ_pred) associated with the point and a left angle (φ left,d ) associated with the left half interval and a right angle (φ right,d ) associated with the right half interval.
6 . The method of claim 4 , wherein the context is selected from a context table (ctxTable T ) having as first entry a context subset index (ctxSetIdx) indicating a selected context subset, and as a second entry a context index (cxtIdx) indicating a selected context in the selected context subset.
7 . A method of decoding a point cloud from a bitstream of encoded point cloud data representing a physical object, the method comprising an azimuthal coding mode providing a series of bits for decoding a coordinate of a point of the point cloud, wherein the method comprises:
dividing an interval, to which the point coordinate belongs to, into a left half interval and a right half interval;
selecting a context based on an apparent angle (AA d ) representing an interval angle seen from a sensor that captured the point; and
context-adaptive binary entropy decoding a bit (b d ), from the bitstream, based on the selected context, the decoded bit (b d ) indicating which of the two half intervals the point coordinate belongs to;
wherein the apparent angle (AA d ) is estimated based on at least one of a first angle (φ (node, d)) associated with a lower bound of the interval, a second angle (φ (top, d)) associated with an upper bound of the interval and a third angle (φ (middle, d)) associated with a middle point of the interval;
wherein the apparent angle (AA d ) is estimated based on the first angle (φ (node, d)) and the second angle (φ (top, d));
wherein the context is selected based on relative magnitudes of an elementary azimuthal angle (Δφ) and the apparent angle (AA d ).
8 . A non-transitory computer-readable storage medium having stored thereon instructions that, when executed by one or more processors, cause the one or more processors to carry out a method of encoding a point cloud from a bitstream of encoded point cloud data representing a physical object, the method comprising an azimuthal coding mode providing a series of bits for encoding a coordinate of a point of the point cloud, wherein the method comprises:
dividing an interval, to which the point coordinate belongs to, into a left half interval and a right half interval;
selecting a context based on an apparent angle representing an interval angle seen from a sensor that captured the point; and
context-adaptive binary entropy encoding a bit of the series of bits, into the bitstream, based on the selected context, said coded bit indicating which of the two half intervals the point coordinate belongs to;
wherein the apparent angle (AA d ) is estimated based on at least one of a first angle (φ (node, d)) associated with a lower bound of the interval, a second angle (φ (top, d)) associated with an upper bound of the interval and a third angle (φ (middle, d)) associated with a middle point of the interval;
wherein the apparent angle (AA d ) is estimated based on the first angle (φ (node, d)) and the second angle (φ (top, d));
wherein the context is selected based on relative magnitudes of an elementary azimuthal angle (Δφ) and the apparent angle (AA d ).
9 . The non-transitory computer-readable storage medium of claim 8 , wherein the apparent angle (AA d ) is estimated based on an interval size (s d ) and a distance (r) between the point and the sensor that captured the point.
10 . The non-transitory computer-readable storage medium of claim 8 , wherein the context is selected based on a ratio between the elementary azimuthal angle (Δφ) over the apparent angle (AA d ).
11 . The non-transitory computer-readable storage medium of claim 10 , wherein contexts are grouped into at least two context subsets based on a range of particular values of the ratio, and selecting the context comprises:
selecting a context subset based on the range of particular values and the ratio; and
selecting a context from the contexts of the selected context subset;
wherein each context subset corresponds to a different prediction quality level of the predicted azimuthal angle (φ pred ), where the corresponding prediction quality level is low when the ratio is much larger than 1 or much smaller than 1, and the corresponding prediction quality level is high when the ratio is close to 1;
wherein selecting a context from the contexts of the selected context subset comprise:
selecting a context subset from the at least two context subsets based on a comparison of the ratio with one or more thresholds; and
selecting a context from the selected context subset based on a comparison of the predicted azimuthal angle with a left angle (φ left,d ) and a right angle (φ right,d ).
12 . The method non-transitory computer-readable storage medium of claim 11 , wherein selecting the context from the contexts of the selected context subset depends on a predicted azimuthal angle (φ pred ) associated with the point and a left angle (φ left,d ) associated with the left half interval and a right angle (φ right,d ) associated with the right half interval.
13 . The method non-transitory computer-readable storage medium of claim 11 , wherein the context is selected from a context table (ctxTable T ) having as first entry a context subset index (ctxSetIdx) indicating a selected context subset, and as a second entry a context index (cxtIdx) indicating a selected context in the selected context subset.