| 主权项 |
1. A method of encoding input video data into bitstreams, the input video data comprising a plurality of video data categories being two or more views or being two views and a depth map indicating multi-view depth information, the method comprising:
providing a first basic dynamic range (BDR) layer, the first BDR layer comprising a frame-compatible BDR representation of the plurality of video data categories from the input video data; and providing a layer grouping, the layer grouping comprising at least one second higher dynamic range (HDR) layer and one third HDR layer, the second HDR layer comprising a second frame-compatible HDR representation of the plurality of video data categories from the input video data, and the third HDR layer comprising a third frame-compatible HDR representation of the plurality of video data categories from the input video data, the third frame-compatible HDR representation being complementary with respect to the second frame-compatible HDR representation; providing a further layer grouping, the further layer grouping comprising at least one fourth BDR layer comprising a fourth frame-compatible BDR representation of the plurality of video data categories from the input video data, the fourth frame-compatible BDR representation being complementary with respect to the first frame-compatible BDR representation; encoding the first BDR layer to generate a first layer bitstream; generating a first layer filtered reconstructed image based on the first layer bitstream in a first layer reference picture buffer; encoding the fourth BDR layer to generate a fourth layer bitstream, wherein the fourth layer bitstream is generated by considering at least one selected from the group consisting of: a difference between the fourth frame-compatible BDR representation and the first layer filtered reconstructed image, inter prediction of temporally decoded pictures of the fourth BDR layer, and intra prediction of the fourth BDR layer; generating a fourth layer filtered reconstructed image based on the fourth layer bitstream in a fourth layer reference picture buffer encoding the second HDR layer to generate a second layer bitstream, wherein the second layer bitstream is generated by considering at least one selected from the group consisting of: a difference between the second frame-compatible HDR representation and the first layer filtered reconstructed image, inter prediction of temporally decoded pictures of the second HDR layer, and intra prediction of the second HDR layer; generating a second layer filtered reconstructed image based on the second layer bitstream in a second layer reference picture buffer; and encoding the third HDR layer to generate a third layer bitstream, wherein the third layer bitstream is generated by considering at least one selected from the group consisting of: a difference between the third frame-compatible HDR representation and the second layer filtered reconstructed image, inter prediction of temporally decoded pictures of the third HDR layer, intra prediction of the third HDR layer, and a difference between the third frame-compatible HDR representation, the second layer filtered reconstructed image, and the fourth layer filtered reconstructed image; wherein the first BDR layer is a base layer, and the fourth BDR layer, the second HDR layer, and the third HDR layer are enhancement layers; and wherein the method further comprises: reference processing one or more of the first layer filtered reconstructed image, the fourth layer filtered reconstructed image, and the second layer filtered reconstructed image with a plurality of reference processing units, the reference processing comprising at least one of de-multiplexing, up-sampling, de-interlacing, frequency filtering, and interpolating the filtered reconstructed images, wherein reference processing is performed on reference pictures at the base layer picture puffer for enhancing inter-layer prediction for one or more of the enhancement layers and on reference pictures in the reference picture buffers of higher-priority enhancement layers belonging to a same layer grouping having at least two enhancement layers for enhancing inter-layer prediction for one or more of the enhancement layers in that same layer grouping. |
