Inter-layer prediction for scalable video coding

Abstract

Systems, methods, and instrumentalities are disclosed for increasing the efficiency of inter-layer prediction using an enhanced inter-layer reference (EILR) picture as a reference picture for inter-layer prediction for encoding an enhancement layer. A luminance component and chrominance components of an inter-layer reference (ILR) picture may be enhanced. High frequency information may be obtained by processing an inter-layer motion compensated (ILMC) picture with a high pass filter. Low frequency information may be obtained by processing an ILR picture with a low pass filter. The EILR picture may be generated as a function of the high frequency information, the low frequency information, and/or the ILR picture.

1 - 26 . (canceled) 
     
     
         27 . A video coding method comprising:
 receiving an inter-layer reference (ILR) picture, motion compensated pixels of an enhancement layer (EL) reference picture, and a current EL picture;   deriving a plurality of filter coefficients for inter-layer prediction based on at least the ILR picture and the current EL picture such that a distortion between the current EL picture and a predictor generated using the plurality of filter coefficients is minimized; and   signaling the plurality of filter coefficients in a bitstream.   
     
     
         28 . A video coding method of  claim 27 , wherein the plurality of filter coefficients comprises a plurality of low pass filter coefficients, and deriving the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of low pass filter coefficients such that, the ILR picture after being filtered using the plurality of low pass filter coefficients, corresponds to a minimum distortion from the current EL picture.   
     
     
         29 . A video coding method of  claim 28 , wherein deriving the plurality of filter coefficients for the inter-layer prediction further comprises determining the plurality of low pass filter coefficients with a constraint that a summation of the plurality of low pass filter coefficients is equal to one. 
     
     
         30 . A video coding method of  claim 27 , wherein the plurality of filter coefficients comprises a plurality of high pass filter coefficients, and deriving the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of high pass filter coefficients such that, the ILR picture after adding high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the plurality of high pass filter coefficients, corresponds to a minimum distortion from the current EL picture.   
     
     
         31 . A video coding method of  claim 30 , wherein deriving the plurality of filter coefficients for the inter-layer prediction further comprises determining the plurality of high pass filter coefficients with a constraint that a summation of the plurality of high pass filter coefficients is equal to zero. 
     
     
         32 . A video coding method of  claim 27 , wherein the plurality of filter coefficients comprises a plurality of low pass filter coefficients and a plurality of high pass filter coefficients, and deriving the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of low pass filter coefficients and the plurality of high pass filter coefficients, such that, a combination of low frequency information generated by filtering the ILR picture using the plurality of low pass filter coefficients and high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the plurality of high pass filter coefficients, corresponds to a minimum distortion from the current EL picture.   
     
     
         33 . A video coding method of  claim 32 , wherein deriving the plurality of filter coefficients for the inter-layer prediction further comprises determining the plurality of filter coefficients with a constraint that a summation of the plurality of high pass filter coefficients and the plurality of the low pass filter coefficients is equal to one. 
     
     
         34 . The video coding method of  claim 27 , wherein the plurality of filter coefficients comprises a plurality of low pass filter coefficients and a plurality of high pass filter coefficients, and determining the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of high pass filter coefficients such that, the ILR picture after adding high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the plurality of high pass filter coefficients, corresponds to a first minimum distortion from the current EL picture; and   determining the plurality of low pass filter coefficients such that, the ILR picture after filtering using the plurality of low pass filter coefficients and adding high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the determined plurality of high pass filter coefficients, corresponds to a second minimum distortion from the current EL picture.   
     
     
         35 . The video coding method of  claim 27 , further comprising:
 encoding the current EL picture using the predictor generated based on the plurality of filter coefficients;   identifying sample positions that the encoded current EL picture refers to on the predictor; and   deriving a plurality of second-pass filter coefficients based on the identified sample positions, wherein the plurality of filter coefficients signaled in the bitstream includes the plurality of second-pass filter coefficients.   
     
     
         36 . The video coding method of  claim 27 , further comprising:
 determining a quantization factor such that a distortion energy between the current EL picture and the predictor is minimized; and   converting the plurality of filter coefficients to integer values based on the quantization factor.   
     
     
         37 . A video coding system comprising:
 a processor programmed to:   receive an inter-layer reference (ILR) picture, motion compensated pixels of an enhancement layer (EL) reference picture, and a current EL picture;   derive a plurality of filter coefficients for inter-layer prediction based on at least the ILR picture and the current EL picture such that a distortion between the current EL picture and a predictor generated using the plurality of filter coefficients is minimized; and   signal the plurality of filter coefficients in a bitstream.   
     
     
         38 . A video coding system of  claim 37 , wherein the plurality of filter coefficients comprises a plurality of low pass filter coefficients, and deriving the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of low pass filter coefficients such that, the ILR picture after being filtered using the plurality of low pass filter coefficients, corresponds to a minimum distortion from the current EL picture.   
     
     
         39 . A video coding system of  claim 38 , wherein deriving the plurality of filter coefficients for the inter-layer prediction further comprises determining the plurality of low pass filter coefficients with a constraint that a summation of the plurality of low pass filter coefficients is equal to one. 
     
     
         40 . A video coding system of  claim 37 , wherein the plurality of filter coefficients comprises a plurality of high pass filter coefficients, and deriving the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of high pass filter coefficients such that, the ILR picture after adding high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the plurality of high pass filter coefficients, corresponds to a minimum distortion from the current EL picture.   
     
     
         41 . A video coding system of  claim 40 , wherein deriving the plurality of filter coefficients for the inter-layer prediction further comprises determining the plurality of high pass filter coefficients with a constraint that a summation of the plurality of high pass filter coefficients is equal to zero. 
     
     
         42 . A video coding system of  claim 37 , wherein the plurality of filter coefficients comprises a plurality of low pass filter coefficients and a plurality of high pass filter coefficients, and deriving the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of low pass filter coefficients and the plurality of high pass filter coefficients, such that, a combination of low frequency information generated by filtering the ILR picture using the plurality of low pass filter coefficients and high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the plurality of high pass filter coefficients, corresponds to a minimum distortion from the current EL picture.   
     
     
         43 . A video coding system of  claim 42 , wherein deriving the plurality of filter coefficients for the inter-layer prediction further comprises determining the plurality of filter coefficients with a constraint that a summation of the plurality of high pass filter coefficients and the plurality of the low pass filter coefficients is equal to one. 
     
     
         44 . The video coding system of  claim 37 , wherein the plurality of filter coefficients comprises a plurality of low pass filter coefficients and a plurality of high pass filter coefficients, and determining the plurality of filter coefficients for the inter-layer prediction further comprises:
 determining the plurality of high pass filter coefficients such that, the ILR picture after adding high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the plurality of high pass filter coefficients, corresponds to a first minimum distortion from the current EL picture; and   determining the plurality of low pass filter coefficients such that, the ILR picture after filtering using the plurality of low pass filter coefficients and adding high frequency information generated by filtering the motion compensated pixels of the EL reference picture using the determined plurality of high pass filter coefficients, corresponds to a second minimum distortion from the current EL picture.   
     
     
         45 . The video coding system of  claim 37 , wherein the processor is further configured to:
 encode the current EL picture using the predictor generated based on the plurality of filter coefficients;   identify sample positions that the encoded current EL picture refers to on the predictor; and   derive a plurality of second-pass filter coefficients based on the identified sample positions, wherein the plurality of filter coefficients signaled in the bitstream includes the plurality of second-pass filter coefficients.   
     
     
         46 . The video coding system of  claim 37 , wherein the processor is further configured to:
 determine a quantization factor such that a distortion energy between the current EL picture and the predictor is minimized; and   convert the plurality of filter coefficients to integer values based on the quantization factor.