Page 50 - Video Coding for Mobile Communications Efficiency, Complexity, and Resilience
P. 50

Section 2.5.  Video Coding Basics                              27


            the double stimulus impairment scale (DSIS) and the double and single stim-
            ulus  continuous  quality  scales,  (DSCQS)  and  (SSCQS),  respectively.  For  a
            detailed description of  such  experiments the reader is referred to Ref. 25.
               Despite  their  reliability,  subjective  quality  experiments  are  expensive  and
            time  consuming.  Objective  measures  provide  cheaper  and  faster  alternatives.
            One  commonly  used  objective  measure  is  the  mean  squared  error  (MSE),
            which is de ned  as
                                     H
                                        V
                                1
                                                     ˆ
                        MSE =              [ f(x; y) − f(x; y)] ;       (2.18)
                                                          2
                              H × V
                                    x=1  y=1
            where  H  and  V  are  the  horizontal  and  vertical  dimensions  of  the  frame,  re-
                                     ˆ
            spectively,  and  f(x; y)  and  f(x; y)  are  the  pel  values  at  location  (x; y)of
            the  original  and  reconstructed  frames,  respectively.  Care  should  be  taken  to
            include  color  components  and  to  take  into  account  any  chroma  subsampling.
            For example, the MSE of a reconstructed 4:2:0 color frame can be calculated
            as
                                   
                                        V
                                     H
                               1
                                                    ˆ
                  MSE 4:2:0  =   3       [Y  (x; y) − Y (x; y)] 2


                            2  H × V  x=1  y=1
                                       H=2  V=2


                                                       ˆ
                                     +       [C (x; y) − C (x; y)] 2
                                               R
                                                        R
                                       x=1  y=1                        (2.19)
                                       H=2  V=2

                                                            ˆ

                                     +       [C (x; y) −  C (x; y)] 2  
                                               B
                                                        B
                                       x=1  y=1
                            2         1        1     � ):
                                            �
                          =  (MSE Y  �  +  MSE C +  MSE C
                            3         4     R   4    B
               A  more  common  form  of  the  MSE  measure  is  the  peak  signal-to-noise
            ratio  (PSNR),  which is  de ned  as
                                        
   2
                                          f
                                           max
                           PSNR = 10 log 10   MSE   (dB);               (2.20)
            where f max  is the maximum possible pel value (for example, 255 for an 8-bit
            resolution  component).  Although  this  measure  does  not  always  correlate  well
            with  perceived  video  quality,  its  relative  simplicity  makes  it  a  very  popular
            choice  in  the  video  coding  community.  Thus,  to  facilitate  comparisons  with
            other algorithms reported in the literature, this book adopts the PSNR measure.
            If  accuracy  is  a  major  concern,  then  more  sophisticated  objective  measures
            based on perceptual  models  can be used  [26].
               When testing a video coding algorithm, it is very important to subject it to
            a range of input video sequences with di6erent characteristics and a reasonable
   45   46   47   48   49   50   51   52   53   54   55