System and method with advance predicted bit-plane coding for progressive fine-granularity scalable (PFGS) video coding

Claim 1

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: encoding a first bitstream representing a base layer; encoding a second bitstream representing one or more low quality enhancement layers, the encoding comprising: motion compensating an original image using the base layer as a reference to form a low quality predicted image, transforming the low quality predicted image to form low quality predicted coefficients, quantizing the low quality predicted coefficients to form quantized coefficients, dequantizing the quantized coefficients to form dequantized coefficients, deriving low quality residues as a difference between the low quality predicted coefficients and the dequantized coefficients, bit plane coding the low quality residues to form coded low quality residues, and compressing the coded low quality residues using variable length coding; and encoding a third bitstream representing one or more high quality enhancement layers based in part on values predicted from the base layer and the low quality enhancement layer.

Claim 2

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: encoding a first bitstream representing a base layer; encoding a second bitstream representing one or more low quality enhancement layers, the encoding comprising: motion compensating an original image using the base layer as a reference to form a low quality predicted image, transforming the low quality predicted image to form low quality predicted coefficients, quantizing the low quality predicted coefficients to form quantized coefficients, dequantizing the quantized coefficients to form dequantized coefficients, deriving low quality residues as a difference between the low quality predicted coefficients and the dequantized coefficients, bit plane coding the low quality residues to form coded low quality residues, and compressing the coded low quality residues using variable length coding; and encoding a third bitstream representing one or more high quality enhancement layers based in part on values predicted from the base layer and the low quality enhancement layer.

Claim 7

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: encoding a first bitstream representing a base layer; encoding a second bitstream representing one or more low quality enhancement layers; and encoding a third bitstream representing one or more high quality enhancement layers based in part on values predicted from the base layer and the low quality enhancement layer; wherein the encoding a second bitstream representing one or more low quality enhancement layers and the encoding a third bitstream representing one or more high quality enhancement layers comprises: motion compensating an original image using the base layer as a reference to form a low quality predicted image and using an enhancement layer as a reference to form a high quality predicted image, transforming the low quality predicted image to form low quality predicted coefficients and transforming the high quality predicted image to form high quality predicted coefficients, quantizing the low quality predicted coefficients to form quantized coefficients, dequantizing the quantized coefficients to form dequantized coefficients, deriving low quality residues as a difference between the high quality predicted coefficients and the dequantized coefficients, bit plane coding the low quality residues to form coded low quality residues, deriving high quality residues as a difference between the high quality predicted coefficients and the coded low quality residues, bit plane coding the high quality residues to form coded high quality residues, and compressing the coded low quality residues and the coded high quality residues using variable length coding.

Claim 8

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: encoding a first bitstream representing a base layer; encoding a second bitstream representing one or more low quality enhancement layers; and encoding a third bitstream representing one or more high quality enhancement layers based in part on values predicted from the base layer and the low quality enhancement layer; wherein the encoding a second bitstream representing one or more low quality enhancement layers and the encoding a third bitstream representing one or more high quality enhancement layers comprises: motion compensating an original image using the base layer as a reference to form a low quality predicted image and using an enhancement layer as a reference to form a high quality predicted image, transforming the low quality predicted image to form low quality predicted coefficients and transforming the high quality predicted image to form high quality predicted coefficients, quantizing the low quality predicted coefficients to form quantized coefficients, dequantizing the quantized coefficients to form dequantized coefficients, deriving low quality residues as a difference between the high quality predicted coefficients and the dequantized coefficients, bit plane coding the low quality residues to form coded low quality residues, deriving high quality residues as a difference between the high quality predicted coefficients and the coded low quality residues, bit plane coding the high quality residues to form coded high quality residues, and compressing the coded low quality residues and the coded high quality residues using variable length coding.

Claim 12

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from low quality to increasingly higher quality, the method comprising: predicting layers in a current frame using multiple layers of varying quality in a previous frame; deriving low quality predicted residues from a low quality layer in the current frame and a low quality layer in the previous frame; encoding the low quality predicted residues; and predicting, from the encoded low quality residues, high quality predicted residues representative of differences between a high quality layer in the current frame and a high quality layer in the previous frame.

Claim 13

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from low quality to increasingly higher quality, the method comprising: predicting layers in a current frame using multiple layers of varying quality in a previous frame; deriving low quality predicted residues from a low quality layer in the current frame and a low quality layer in the previous frame; encoding the low quality predicted residues; and predicting, from the encoded low quality residues, high quality predicted residues representative of differences between a high quality layer in the current frame and a high quality layer in the previous frame.

Claim 20

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: forming a base layer and multiple enhancement layers in a first frame representing an original image; and predicting, from the base layer and at least one enhancement layer in the first frame, a base layer and multiple enhancement layers in a second frame representing a predicted image; and deriving low quality residues resulting from the predicting of the base layer and low quality enhancement layers from the original image; encoding the low quality residues; and predicting, from the encoded low quality residues, high quality residues indicative of results from predicting high quality enhancement layers from the original image, wherein the multiple enhancement layers include at least first, second, and third layers, and the forming of the multiple enhancement layers comprises predicting even frames from the base and second layers of preceding frames and predicting odd frames from the base and third layers of preceding frames.

Claim 21

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: forming a base layer and multiple enhancement layers in a first frame representing an original image; and predicting, from the base layer and at least one enhancement layer in the first frame, a base layer and multiple enhancement layers in a second frame representing a predicted image; and deriving low quality residues resulting from the predicting of the base layer and low quality enhancement layers from the original image; encoding the low quality residues; and predicting, from the encoded low quality residues, high quality residues indicative of results from predicting high quality enhancement layers from the original image, wherein the multiple enhancement layers include at least first, second, and third layers, and the forming of the multiple enhancement layers comprises predicting even frames from the base and second layers of preceding frames and predicting odd frames from the base and third layers of preceding frames.

Claim 25

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: forming a base layer and multiple enhancement layers in a first frame representing an original image; and predicting, from the base layer and at least one enhancement layer in the first frame, a base layer and multiple enhancement layers in a second frame representing a predicted image; and deriving low quality residues resulting from the predicting of the base layer and low quality enhancement layers from the original image; encoding the low quality residues; and predicting, from the encoded low quality residues, high quality residues indicative of results from predicting high quality enhancement layers from the original image, wherein the forming of the multiple enhancement layers comprises predicting each of the enhancement layers in a current frame from a lower quality reference layer of a reference frame if a relationship holds true, the relationship being: L mod N=i mod M where L is a variable designating a particular enhancement layer, i is a variable designating a particular frame, N is a variable representing a layer group depth of how many layers are predicted from a common reference layer, and M is a variable representing a frame group depth of how many frames are grouped together for prediction purposes.

Claim 26

A method for coding video data according to layered coding techniques in which the video data is represented as multi-layered frames, each frame having multiple layers ranging from a base layer of low quality to enhancement layers of increasingly higher quality, the method comprising: forming a base layer and multiple enhancement layers in a first frame representing an original image; and predicting, from the base layer and at least one enhancement layer in the first frame, a base layer and multiple enhancement layers in a second frame representing a predicted image; and deriving low quality residues resulting from the predicting of the base layer and low quality enhancement layers from the original image; encoding the low quality residues; and predicting, from the encoded low quality residues, high quality residues indicative of results from predicting high quality enhancement layers from the original image, wherein the forming of the multiple enhancement layers comprises predicting each of the enhancement layers in a current frame from a lower quality reference layer of a reference frame if a relationship holds true, the relationship being: L mod N=i mod M where L is a variable designating a particular enhancement layer, i is a variable designating a particular frame, N is a variable representing a layer group depth of how many layers are predicted from a common reference layer, and M is a variable representing a frame group depth of how many frames are grouped together for prediction purposes.