Apparatus for intra predicting a block, apparatus for reconstructing a block of a picture, apparatus for reconstructing a block of a picture by intra prediction
View Patent ↗An apparatus for intra predicting a block to be predicted of a picture from neighboring image samples sequentially extending along two rays starting at a common intersection point and pointing into different directions, the neighboring image samples including edge neighboring samples bordering the block to be predicted, includes a smoothing filter and a determiner. The smoothing filter is configured to one-dimensionally filter the neighboring image samples, which are sequentially arranged into a sequence by anti-parallelly traveling along a first one of the two rays and parallelly traveling along a second one of the two rays to obtain a filtered image sample sequence. The determiner is configured to determine a prediction value for a predetermined image sample of the block to be predicted by using the filtered image sample sequence.
1. An apparatus for reconstructing a block of a picture defined in a first bit-depth precision by prediction, comprising:
a reconstructor configured to reconstruct reference image samples from a data stream at a second bit-depth precision being equal to or greater than the first bit-depth precision to achieve reconstructed reference image samples;
an extractor configured to extract a transform coefficient block from the data stream;
a predictor configured to determine predicted image samples for the block from the reconstructed reference image samples at a third bit-depth precision higher than the second bit-depth precision;
an inverse transformer configured to apply an inverse spectral decomposition transform to the transform coefficient block to achieve reconstructed residual sample values for the block at a fourth precision greater than the first precision; and
wherein the reconstructor is further configured to reconstruct the block by summation of the predicted image samples and the reconstructed residual sample values at an increased bit-depth precision equal to or greater than at least one of the third and fourth bit-depth precisions and reducing a result of the summation comprising the increased bit-depth precision down to the second bit-depth precision.
2. The apparatus according to claim 1 ,
wherein the reconstructor is further configured such that the reconstructed reference image samples are neighboring image samples, the neighboring image samples comprising edge-neighboring samples bordering the block to be reconstructed.
3. The apparatus according to claim 1 ,
wherein the predictor is further configured such that prediction values for the predicted image samples for the block are determined by interpolating the reconstructed reference image samples.
4. The apparatus according to claim 3 , wherein the predictor is further configured to increase the bit-depth precision of the reconstructed reference image samples before interpolating.
5. The apparatus according to claim 3 , wherein the predictor is further configured to interpolate the reconstructed reference image samples by using a FIR filter; and wherein coefficients of the FIR filter are adapted such that denominators of the coefficients are equal and wherein the third bit-depth precision of the predicted image samples is dependent on the denominators of the coefficients.
6. The apparatus according to claim 5 , wherein the predictor is further configured such that the third bit-depth precision is by, a number of bits, equal to log 2 of the denominators of the coefficients higher than the bit-depth second precision.
7. The apparatus according to claim 1 , further comprising a reproduction output, the reproduction output configured to provide a reconstruction of the block of the picture in the first bit-depth precision.
8. An apparatus for encoding a block of a picture defined in a first bit-depth precision, comprising:
a reconstructor configured to reconstruct reference image samples at a second bit-depth precision being equal to or greater than the first bit-depth precision to achieve reconstructed reference image samples;
a predictor configured to determine predicted image samples for the block from the reconstructed reference image samples at a third bit-depth precision higher than the second bit-depth precision;
a residual sample value determiner configured to compare the predicted image samples and original image samples of the block of the picture defined in the first bit-depth precision to determine residual sample values for the block;
a transformer configured to apply a spectral decomposition transform to the achieved residual sample values and use a quantization to achieve a transform coefficient block;
an inverse transformer configured to apply an inverse spectral decomposition transform to the transform coefficient block to achieve reconstructed residual sample values for the block at a fourth bit-depth precision greater than the first bit-depth precision;
wherein the reconstructor is further configured to achieve reconstructed image samples of the block by summation of the predicted image samples and the reconstructed residual sample values at an increased bit-depth precision equal to or greater than at least one of the third and fourth bit-depth precisions and reducing a result of the summation comprising the increased bit-depth precision down to the second bit-depth precision; and
a data stream inserter configured to insert the transform coefficient block into a data stream.
9. A method for reconstructing a block of a picture defined in a first bit-depth precision by prediction, comprising:
reconstructing a reference image sample from a data stream at a second bit-depth precision being equal to or greater than the first bit-depth precision;
extracting a transform coefficient block for the block from the data stream;
determining predicted image samples for the block from the reconstructed reference image samples at a third bit-depth precision higher than the second bit-depth precision;
applying an inverse spectral decomposition transform to the transform coefficient block to achieve reconstructed residual sample values for the block at a fourth bit-depth precision greater than the first bit-depth precision; and
reconstructing the block by summation of the predicted image samples and the reconstructed residual sample values at an increased precision equal to or greater than at least one of the third and fourth bit-depth precisions and reducing a result of the summation comprising the increased precision down to the second bit-depth precision.
10. A method for encoding a block of a picture defined in a first bit-depth precision, comprising:
reconstructing reference image samples at a second bit-depth precision being equal or greater than the first bit-depth precision;
determining predicted image samples for the block from the reconstructed reference image samples at a third bit-depth precision higher than the second bit-depth precision;
comparing the predicted image samples and original image samples of the block of the picture defined in the first bit-depth precision to determine residual sample values for the block;
applying a spectral decomposition transform to the achieved residual sample values and using a quantization to achieve a transform coefficient block;
applying an inverse spectral decomposition transform to the transform coefficient block to achieve the reconstructed residual sample values for the block at a fourth bit-depth precision greater than the first precision;
achieving reconstructed image samples of the block by summation of the predicted image samples and the reconstructed residual sample values at an increased precision equal to or greater than at least one of the third and fourth bit-depth precisions and reducing a result of the summation comprising the increased bit-depth precision down to the second bit-depth precision; and
inserting the transform coefficient block into a data stream.
11. A non-transitory computer readable digital storage medium having stored thereon a computer program comprising a program code for performing, when running on a computer, a method for reconstructing a block of a picture defined in a first bit-depth precision by prediction, said method comprising:
reconstructing a reference image sample from a data stream at a second bit-depth precision being equal to or greater than the first bit-depth precision;
extracting a transform coefficient block for the block from the data stream;
determining predicted image samples for the block from the reconstructed reference image samples at a third bit-depth precision higher than the second bit-depth precision;
applying an inverse spectral decomposition transform to the transform coefficient block to achieve reconstructed residual sample values for the block at a fourth bit-depth precision greater than the first bit-depth precision; and
reconstructing the block by summation of the predicted image samples and the reconstructed residual sample values at an increased bit-depth precision equal to or greater than at least one of the third and fourth bit-depth precisions and reducing a result of the summation comprising the increased bit-depth precision down to the second bit-depth precision.
12. A non-transitory computer readable digital storage medium having stored thereon a computer program comprising a program code for performing, when running on a computer, a method for encoding a block of a picture defined in a first bit-depth precision, said method comprising:
reconstructing reference image samples at a second bit-depth precision being equal or greater than the first bit-depth precision;
determining predicted image samples for the block from the reconstructed reference image samples at a third bit-depth precision higher than the second bit-depth precision;
comparing the predicted image samples and original image samples of the block of the picture defined in the first bit-depth precision to determine residual sample values for the block;
applying a spectral decomposition transform to the achieved residual sample values and using a quantization to achieve a transform coefficient block;
applying an inverse spectral decomposition transform to the transform coefficient block to achieve the reconstructed residual sample values for the block at a fourth bit-depth precision greater than the first bit-depth precision;
achieving reconstructed image samples of the block by summation of the predicted image samples and the reconstructed residual sample values at an increased bit-depth precision equal to or greater than at least one of the third and fourth bit-depth precisions and reducing a result of the summation comprising the increased bit-depth precision down to the second bit-depth precision; and
inserting the transform coefficient block into a data stream.