Luma-based chroma intra-prediction for video coding

1. A method for luma-based chroma intra-prediction in a video encoder or a video decoder, the method comprising:

filtering reconstructed neighboring samples of a reconstructed down sampled luma block;

computing parameters α and β of a linear model using the filtered, reconstructed neighboring samples of the reconstructed down sampled luma block and reconstructed neighboring samples of a corresponding chroma block, wherein the linear model is Pred C [x,y]=α·Rec L ′[x,y]+β, wherein x and y are sample coordinates, Pred C [x,y] are predicted chroma samples, and Rec L ′[x,y] are samples of the reconstructed down sampled luma block;

normalizing the parameter α by shifting the parameter α to reduce a precision of the parameter α; and

computing samples of a predicted chroma block from corresponding samples of the reconstructed down sampled luma block using the linear model and the parameters including the normalized parameter α.

2. The method of claim 1 , further comprising:

filtering the reconstructed neighboring samples of the corresponding chroma block prior to computing parameters α and β, and

wherein computing parameters α and β comprises computing the parameters using the filtered, reconstructed neighboring samples of the reconstructed down sampled luma block and the filtered, reconstructed neighboring samples of the corresponding chroma block.

3. The method of claim 1 , wherein normalizing the parameter α comprises reducing the precision of α to one selected from a group consisting of 6 bits, 7 bits, and 8 bits.

4. The method of claim 1 , further comprising:

down sampling the reconstructed filtered neighboring samples of the reconstructed down sampled luma block and the reconstructed neighboring samples of the corresponding chroma block prior to computing parameters α and β; and

wherein computing parameters α and β comprises computing the parameters using the down sampled reconstructed filtered neighboring samples of the reconstructed down sampled luma block and the down sampled reconstructed neighboring samples of the corresponding chroma block.

5. The method of claim 1 , wherein the reconstructed neighboring samples of the corresponding chroma block consist of reconstructed top neighboring samples and reconstructed left neighboring samples of the corresponding chroma block, and the reconstructed neighboring samples of the reconstructed down sampled luma block consist of reconstructed top neighboring samples and reconstructed left neighboring samples of the reconstructed down sampled luma block.

6. The method of claim 1 , wherein the reconstructed neighboring samples of the corresponding chroma block consist of reconstructed top neighboring samples of the corresponding chroma block and the reconstructed neighboring samples of the reconstructed down sampled luma block consist of reconstructed top neighboring samples of the reconstructed down sampled luma block.

7. The method of claim 1 , further comprising:

receiving predicted left neighboring samples of the corresponding chroma block and of the reconstructed down sampled luma block; and

wherein computing parameters α and β comprises computing the parameters using the predicted left neighboring samples of the corresponding chroma block and of the reconstructed down sampled luma block, the filtered, reconstructed neighboring samples of the reconstructed down sampled luma block, and the reconstructed neighboring samples of the corresponding chroma block, wherein the reconstructed neighboring samples of the reconstructed down sampled luma block consist of reconstructed top neighboring samples of the reconstructed down sampled luma block, and the reconstructed neighboring samples of the corresponding chroma block consist of reconstructed top neighboring samples of the corresponding chroma block.

8. A method for luma-based chroma intra-prediction in a video encoder or a video decoder, the method comprising:

computing parameters α and β of a linear model using reconstructed neighboring samples of a reconstructed down sampled luma block and of a corresponding chroma block, wherein the linear model is Pred C [x,y]=α·Rec L ′[x,y]+β, wherein x and y are sample coordinates, Pred C [x,y] are predicted chroma samples, and Rec L ′[x,y] are samples of the reconstructed down sampled luma block;

normalizing the parameter α by shifting the parameter α to reduce a precision of the parameter α; and

computing samples of a predicted chroma block from corresponding samples of the reconstructed down sampled luma block using the linear model and the parameters including the normalized parameter α.

9. The method of claim 8 , further comprising:

filtering the reconstructed neighboring samples of the reconstructed down sampled luma block prior to computing parameters α and β, wherein the parameters are computed using the filtered reconstructed neighboring samples of the reconstructed down sampled luma block and the reconstructed neighboring samples of the corresponding chroma block.

10. The method of claim 8 , wherein normalizing the parameter α comprises reducing the precision of a to one selected from a group consisting of 6 bits, 7 bits, and 8 bits.

11. The method of claim 8 , further comprising:

down sampling the reconstructed neighboring samples of the reconstructed down sampled luma block and the corresponding chroma block prior to computing parameters α and β, wherein the parameters are computed using the down sampled reconstructed neighboring samples of the reconstructed down sampled luma block and the corresponding chroma block.

12. The method of claim 8 , wherein the reconstructed neighboring samples of the corresponding chroma block consist of reconstructed top neighboring samples and reconstructed left neighboring samples of the corresponding chroma block, and the reconstructed neighboring sample of the reconstructed down sampled luma block consist of reconstructed top neighboring samples and reconstructed left neighboring samples of the reconstructed down sampled luma block.

13. The method of claim 8 , wherein the reconstructed neighboring samples of the corresponding chroma block consist of reconstructed top neighboring samples of the corresponding chroma block and the reconstructed neighboring samples of the reconstructed down sampled luma block consist of reconstructed top neighboring samples of the reconstructed down sampled luma block.

14. The method of claim 8 , further comprising:

receiving predicted left neighboring samples of the corresponding chroma block and of the reconstructed down sampled luma block; and

wherein computing parameters α and β comprises computing the parameters using the predicted left neighboring samples of the corresponding chroma block and of the reconstructed down sampled luma block, the filtered, reconstructed neighboring samples of the reconstructed down sampled luma block, and the reconstructed neighboring samples of the corresponding chroma block, wherein the reconstructed neighboring samples of the reconstructed down sampled luma block consist of reconstructed top neighboring samples of the reconstructed down sampled luma block, and the reconstructed neighboring samples of the corresponding chroma block consist of reconstructed top neighboring samples of the corresponding chroma block.

15. A method for luma-based chroma intra-prediction in a video encoder or a video decoder, the method comprising:

determining a type of luma-based chroma intra-prediction to be used, wherein the type is one of a plurality of types of luma-based chroma intra-prediction; and

performing luma-based chroma intra-prediction according to the determined type to generate a predicted chroma block, wherein performing luma-based chroma intra-prediction comprises:

computing parameters α and β of a linear model using neighboring samples of a reconstructed down sampled luma block and of a corresponding chroma block, wherein the linear model is Pred C [x,y]=α·Rec L ′[x,y]+β, wherein x and y are sample coordinates, Pred C [x,y] are predicted chroma samples, and Rec L ′[x,y] are samples of the reconstructed down sampled luma block;

normalizing the parameter α by shifting the parameter α to reduce a precision of the parameter α; and

computing samples of a predicted chroma block from corresponding samples of the reconstructed down sampled luma block using the linear model and the parameters including the normalized parameter α.

16. The method of claim 15 , wherein for a first type of luma-based chroma intra-prediction, the neighboring samples consist of reconstructed top neighboring samples and reconstructed left neighboring samples of the corresponding chroma block, reconstructed top neighboring samples and reconstructed left neighboring samples of the reconstructed down sampled luma block, for a second type of luma-based chroma intra-prediction, the neighboring samples consist of reconstructed top neighboring samples of the corresponding chroma block and reconstructed top neighboring samples of the reconstructed down sampled luma block, and for a third type of luma-based chroma intra-prediction, the neighboring samples consist of reconstructed top neighboring samples of the corresponding chroma block, reconstructed top neighboring samples of the reconstructed down sampled luma block, predicted left neighboring samples of the corresponding chroma block, and predicted left neighboring samples of the reconstructed down sampled luma block.

17. The method of claim 15 , further comprising:

filtering the neighboring samples prior to computing parameters α and β.