[M4IF Discuss] RE: RE: [M4IF Technotes] Re: motion estimation, please advise

[Mithun N]

Hi..
Motion estimation isnt performed on Cr and Cb
components.
As such I have no idea about the PMVFAST method.
Guess its based on the diamond search
method...with a
large diamond and small diamond pattern.
Ya but since all the methods are based on
correlation between the
2 frames, they essentially need to find at what
point in the reference
frame, the correlation is maximum.(to achieve as
small VLC for the
error block as possible)
The only difference between the methods is in the
points at which
the search is done in the reference frame.
regards
Mithun Nayak
-----Original Message-----
From: cloverleafland   mail.hongkong.com
[mailto:cloverleafland   mail.hongkong.com]
Sent: Friday, November 08, 2023 8:43 AM
To: <mithun.nayak   patni.com>;
Cc: discuss   lists.m4if.org
Subject: Re: RE: [M4IF Technotes] Re: motion
estimation, please advise
Hi..
Thanks for your reply....
actually I would like to use PMVFAST method...
but I don't know would there be help if using a
combination of Y, Cr, Cb elements during the
estimation in PMVFAST method.
Besides, I am confused that, isn't that the same
between different methods?as the criteria for
finding a motion vector is by the minimum SAD
value?
regards,
Clover
> hi
> i didnt understand your first question
properly..
> Could you clarify it a bit further??
> As for the Motion estimation, there are papers
> available of the type of method to be used ..
> A combination of three step search and four step
> search is useful in case the motion isnt too
much
> between 2 time instants.
> This method is generally called Majority Voting
> algorithm where the encoder decides between
> the 2 methods by using the surrounding
macroblock
> motion vectors.
> This method is more useful if u plan to build a
> real time
> encoder.
> Else there are other methods such as Full
Search..
> which take a lot of processing time.
> regards
> Mithun Nayak
>
>
> -----Original Message-----
> From: technotes-admin   lists.m4if.org
> [mailto:technotes-admin   lists.m4if.org]On Behalf
> Of
> cloverleafland   mail.hongkong.com
> Sent: Thursday, November 07, 2023 3:20 PM
> To: technotes   lists.m4if.org
> Cc: discuss   lists.m4if.org
> Subject: [M4IF Technotes] Re: motion estimation,
> please advise
>
>
> Dear all,
>
> >From previous questions on motion estimation,
it
> can be done on Y component in each macroblock or
> combining a fraction of Cr and Cb together with
Y
> into a 16x16 blocks for estimation. I would like
> to ask is there any further information on which
> method is better based on quality, speed and
other
> performance? any source or papers comparing the
> performance of these two methods that I would
> find?
> Thanks very much.
>
> Regards,
> Clover
> ---------------------------------------------
>  Åwªï¨Ï¥ÎHongKong.com¶l¥ó¨t²Î
>  Thank you for using hongkong.com Email system
>
> _______________________________________________
> Technotes mailing list
> Technotes   lists.m4if.org
> http://lists.m4if.org/mailman/listinfo/technotes
>
----------------------------------------------
 Åwªï¨Ï¥ÎHongKong.com¶l¥ó¨t²Î
 Thank you for using hongkong.com Email system