[WM]: help abt embedding watremark in MPEG video sequence

[Anindya Sarkar <ani_great@yahoo.com>]

Hi

I was going through the following 2 papers:
1.   	Video watermark technique in motion vector
Jun Zhang; Jiegu Li; Ling Zhang;
Computer Graphics and Image Processing, 2001 Proceedings of XIV Brazilian Symposium on
15-18 Oct. 2001 Page(s):179 - 182
Digital Object Identifier 10.1109/SIBGRAPI.2001.963053


2. Embedding watermark in MPEG video sequence Jun Zhang; Maitre, H.; Jiegu Li; Ling Zhang; Multimedia Signal Processing, 2001 IEEE
Fourth Workshop on
3-5 Oct. 2001 Page(s):535 - 540
Digital Object Identifier 10.1109/MMSP.2001.962788 

In both the papers, the authors try to hide information in tne motion vectors. If you have used or implemented this scheme, then you
can help me out with some questions regarding the embedding scheme:

1) what is the threshold \eta that you set in step 3 of embedding scheme?

2) when we find angle \theta[j] in step 4, then do we have \theta[j] from -PI/2 to PI/2 or from -PI to PI..
By definition, an acute angle is a positive angle >0 & < PI/2.. and an obtuse angle is >PI/2 & <= PI.. Now if \theta[j] varies from
-PI to PI, then what is the relation between \theta[j] from -PI to 0, and their being acute or obtuse angles..

3) In step 5, we embed data depending on the values of q, T,  and n, where  T= 2*<Search Window used for Motion Estimation>;
d=(2n+l)/q, n is integer; q is used to specify the amplitude of the motion vector modification.

Now, I am using std MPEG 2 encoder & decoder..