Adaptive secondary prediction-based image encoding/decoding method, device and recording medium

A video decoding apparatus, comprising:

• a decoder configured to
• obtain a first motion vector of a current block from a bitstream, and
• obtain, from the bitstream, residual signals and a first information for indicating whether an inverse frequency transform is applied to the residual signals;
• a residual block reconstructer configured to reconstruct a residual block from the residual signals by
• inversely quantizing the obtained residual signals and skipping the inverse frequency transform for the inversely quantized residual signals, when the first information indicates skipping of the inverse frequency transform, and
• inversely quantizing the obtained residual signals and inversely frequency-transforming the inversely quantized residual signals, when the first information indicates non-skipping of the inverse frequency transform;
• a predictor configured to generate a predicted block by predicting the current block based on the obtained first motion vector;
• and
• an adder configured to add each pixel value of the reconstructed residual block to a corresponding pixel value of the predicted block to thereby reconstruct the current block.

A video encoding method, comprising:

• generating a predicted block of a current block by using a first motion vector of the current block;
• generating a first residual block by subtracting the predicted block from the current block;
• choosing to perform one of the following operations to generate a first quantized block of the current block:
• (i) performing a frequency transform of the first residual block into a frequency transformed first residual block and quantizing the frequency transformed first residual block to generate a frequency domain first quantized block, the first quantized block of the current block being set as the frequency domain first quantized block, and
• (ii) quantizing the first residual block to generate a spatial domain first quantized block of the current block, the first quantized block of the current block being set as the spatial domain first quantized block of the current block;
• choosing to perform one of the following operations to generate a quantized block to be encoded:
• (a) generating a second residual block by performing a secondary prediction on the spatial domain first quantized block of the current block by using a spatial domain first quantized block of a reference block, the reference block corresponding to a neighboring motion vector of the current block, the quantized block to be encoded being set as the second quantized block, and
• (b) skipping the performing of the secondary prediction, the quantized block to be encoded being set as the first quantized block;
• and
• encoding the quantized block to be encoded,
• encoding a first information indicating whether the frequency transform is performed on the first residual block based on the choosing and
• encoding a second information indicating whether the secondary prediction is performed on the spatial domain first quantized block, when the first residual block has been frequency-transformed into a frequency transformed first residual block.

A video decoding method, comprising:

• obtaining a first motion vector of a current block from a bitstream, and
• obtaining, from the bitstream, residual signals and a first information for indicating whether an inverse frequency transform is applied to the residual signals;
• reconstructing a residual block from the residual signals by
• inversely quantizing the obtained residual signals and skipping the inverse frequency transform for the inversely quantized residual signals, when the first information indicates skipping of the inverse frequency transform, and
• inversely quantizing the obtained residual signals and inversely frequency-transforming the inversely quantized residual signals, when the first information indicates non-skipping of the inverse frequency transform;
• generating a predicted block by predicting the current block based on the obtained first motion vector;
• and
• adding each pixel value of the reconstructed residual block to a corresponding pixel value of the predicted block to thereby reconstruct the current block.