Wavelet transform apparatus and method, scalable video coding apparatus and method employing the same, and scalable video decoding apparatus and method thereof

A scalable video coding apparatus comprising:

• a motion estimation unit which performs motion estimation on a current frame of a video sequence and generates a motion vector;
• a temporal wavelet decomposition unit, executed by at least one processing device, which repeatedly performs a prediction step using the motion vector and a modified update step on the video sequence and forms video data with temporal scalability;
• a motion information coding unit which codes the motion vector provided by the motion estimation unit;
• and
• an entropy coding unit which entropy-codes the video data with temporal scalability provided by the temporal wavelet decomposition unit,
• wherein, for the modified update step, in the current frame, the temporal wavelet decomposition unit determines whether to update each target block at which a reference block indicated by the motion vector is placed and calculates a weight for each target sub-block included in a target block to be updated, and
• the weight is calculated using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block.

A scalable video coding method comprising:

• performing motion estimation on a current frame of a video sequence and generating a motion vector;
• repeatedly performing a prediction step using the motion vector and a modified update step on the video sequence and forming video data with temporal scalability;
• coding the motion vector;
• and
• entropy-coding the video data with temporal scalability,
• wherein, for the modified update step, in the current frame, the forming of the video data with temporal scalability comprises calculating a weight for each target sub-block included in a target block at which a reference block indicated by the motion vector is placed, and
• the weight is calculated using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block.

A scalable video coding method comprising:

• performing motion estimation on a current frame of a video sequence and generating a motion vector;
• repeatedly performing a prediction step using the motion vector and a modified update step on the video sequence and forming video data with temporal scalability;
• coding the motion vector;
• and
• entropy-coding the video data with temporal scalability,
• wherein, for the modified update step, the forming of the video data with temporal scalability comprises:
• in the current frame, determining whether to update each target block at which a reference block indicated by the motion vector is placed;
• and
• calculating the weight for each target sub-block included in a target block to be updated and
• the weight is calculated using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block.

A scalable video decoding apparatus comprising:

• an entropy-decoding unit which entropy-decodes video data with temporal scalability from a bitstream generated through a prediction step and a modified update step, wherein, for the modified update step, in the current frame, the temporal wavelet decomposition unit calculates a weight for each of a plurality of target sub-blocks included in a target block at which a reference block indicated by the motion vector is placed;
• a motion information decoding unit which decodes motion information from the bitstream and restores a motion vector;
• and
• a temporal wavelet synthesizing unit executed by at least one processing device, which performs a lifting-based inverse wavelet transform using the restored motion vector on the video data with temporal scalability and reconstructs a video sequence,
• wherein the weight is calculated using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block.

A scalable video decoding method comprising:

• entropy-decoding video data with temporal scalability from a bitstream generated through a prediction step and a modified update step, wherein, for the modified update step, in the current frame, the temporal wavelet decomposition unit calculates a weight for each of a plurality of target sub-blocks included in a target block at which a reference block indicated by the motion vector is placed;
• decoding motion information from the bitstream and restoring a motion vector;
• and
• performing a lifting-based inverse wavelet transform using the restored motion vector on the video data with temporal scalability and reconstructing a video sequence,
• wherein the weight is calculated using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block.

A lifting-based wavelet transform apparatus comprising:

• a prediction unit, executed by at least one processing device, which performs bi-directional motion compensation on odd-indexed frames based on even-indexed frames and generates high-frequency frames using differential frames from which most of energy is removed;
• an update determination unit which, in the even-indexed frame, determines whether to update each target block at which a reference block indicated by a motion vector is placed;
• and
• an update unit which updates a target block of the even-indexed frame which is determined to be updated by adding the differential frame to each target sub-block included in the target block and generates low-frequency frames,
• wherein the update determination unit comprises:
• a block update weight calculator which, in the even-indexed frame, calculates a block update weight using the number of entire pixels included in the target block and the number of connected pixels in the target block obtained from the reference block indicated by the motion vector;
• and
• a comparator which compares the block update weight of the target block with a predetermined threshold and determines whether to update the target block based on the comparison result, and
• the update unit comprises a sub-block weight setting unit which sets a weight for each target sub-block included in the target block using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block.

A lifting-based wavelet transform apparatus comprising:

• a prediction unit, executed by at least one processing device, which performs bi-directional motion compensation on odd-indexed frames based on even-indexed frames and generates high-frequency frames using differential frames from which most of energy is removed;
• and
• an update unit which updates a target block in the even-indexed frame by adding the differential frame to each target sub-block included in the target block and generates low-frequency frames,
• wherein the update unit comprises:
• a sub-block weight setting unit which sets a weight for each target sub-block included in the target block using the number of entire pixels included in each target sub-block and the number of connected pixels in each target sub-block;
• and
• a content-adaptive update unit which performs a content-adaptive update step on each pixel using the set weight for each target sub-block.

A lifting-based wavelet transform method comprising:

• performing bi-directional motion compensation on odd-indexed frames based on even-indexed frames and generating high-frequency frames using differential frames from which most of energy is removed;
• in the even-indexed frame, determining whether to update each target block in which a reference block indicated by a motion vector is placed;
• and
• updating a target block which is determined to be updated in the even-indexed frame by adding the differential frame to each target sub-block included in the target block and generating low-frequency frames,
• wherein the determining whether to update each target block comprises:
• in the even-indexed frame, calculating a block update weight using the number of entire pixels included in the target block and the number of connected pixels in the target block obtained from the reference block indicated by the motion vector;
• and
• comparing the block update weight of the target block with a predetermined threshold and determines whether to update the target block based on the comparison result.

A lifting-based wavelet transform method comprising:

• performing bi-directional motion compensation on odd-indexed frames based on even-indexed frames and generating high-frequency frames using differential frames from which most of energy is removed;
• setting a weight for each target sub-block included in a target block in the even-indexed frame using the number of pixels included in each target sub-block and the number of connected pixels in each target sub-block;
• and
• performing a content-adaptive update step on each pixel using the set weight for each target sub-block.