Video decoder architecture and method for using same

Abstract

A decoder and method for using a new picture or frame type is provided. This type is referred to a an SP-picture. The temporal redundancies are not exploited in I-frames, compression efficiency of I-frame coding is significantly lower than the predictive coding. A method allows use of motion compensated predictive coding to exploit temporal redundancy in the sequence while still allowing perfect reconstruction of the frame using different reference frames. Methods using this new picture type provide for error resilience/recovery, bandwidth scalability, bitstream switching, processing scalability, random access and other functions. The SP-type picture provides for, among other functions, switching between different bitstreams, random access, fast forward and fast error-recovery by replacing I-pictures to increase the coding efficiency. As will be demonstrated, SP-pictures have the property that identical SP-frames may be obtained even when they are predicted using different reference frames.

1  to  18  cancelled.  
     
     
         19 . A method of coding a switching picture S.sub. 12  for switching from a first bitstream having a SP-picture S.sub. 1  to a second bitstream having a second SP-picture S.sub. 2 , said method comprising the steps of; copying the bitstream of the intra macroblocks in second SP-picture S.sub. 2  to switching picture S.sub. 12 ; and encoding said copied macroblocks comprising the steps of: forming a predicted frame for S.sub. 12  by performing motion estimation with a plurality of reference pictures which are preceding S.sub. 1  in said first bitstream; calculating a set of transform coefficients for predicted image c.sub.pred by performing a forward transform; Quantizing the obtained transform coefficients to resulting in quantized coefficient levels; and subtracting said quantized coefficient levels from a set of corresponding coefficient levels of said second SP-picture S.sub. 2 .  
     
     
         20 . A method of reconstructing blocks of encoded data comprising the steps of: forming a prediction P(x,y) of a current block using a plurality of received motion vectors and a reference frame; performing a forward transform on P(x,y) to obtain a set of transform coefficients c.sub.pred for P(x,y); quantizing said set of transform coefficients c.sub.pred, resulting in quantized prediction image coefficients I.sub.pred. adding said received quantized coefficients for the prediction error I.sub.err to I.sub.pred resulting in quantized reconstruction image coefficients I.sub.rec using: I.sub.rec=I.sub.pred+(beta(QP 2 ).times.I.sub.err+0.5.time-s.beta(QP 1 ))//beta(OP 1 ). wherein beta(QP) is a parameter dependent on method of quantization and used QP value; dequantizing I.sub.rec resulting in dequantized coefficients d.sub.rec; and performing inverse transform for d.sub.rec.  
     
     
         21 . The method of  claim 20 , further comprising the step of normalizing said plurality of inverse transformed dequantized coefficients.  
     
     
         22 . The method of  claim 21 , further comprising the step of filtering said plurality of normalized inverse transformed dequantized coefficients.