Computers & Electronics

A Novel Digital Watermarking Technique Based on ISB (Intermediate Significant Bit)

Description
Least Significant Bit (LSB) technique is the earliest developed technique in watermarking and it is also the most simple, direct and common technique. It essentially involves embedding the watermark by replacing the least significant bit of the image
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
     Abstract—  Le ast  Signifi ca n t  Bi t  (LSB) t e c hnique i s   t he e ar  lie st  developed t e c hnique in w at e r  m ar  king a nd i t  i s   a l s o t he mo st   s imple, di r  e ct   a nd c ommon t e c hnique. I t  e ss en t i a lly involve s  em  b edding t he w at e r  m ar  k  b y r  epl ac ing t he le ast   s ignifi ca n t    b i t  of t he im a ge d ata  wi t h a    b i t  of t he w at e r  m ar  k d ata . The di sa dv a n ta ge of LSB i s   t h at  i t  i s  no t   r  o  b u st   a g a in st   attac k  s . In t hi s   st udy in t e r  medi at e s ignifi ca n t    b i t  (ISB) h as    b een u s ed in o r  de r    t o imp r  ove t he r  o  b u st ne ss  of t he w at e r  m ar  king s y st em. The a im of t hi s  model i s   t o r  epl ac e t he w at e r  m ar  ked im a ge  pixel s    b y new pixel s   t h at   ca n p r  o t e ct   t he w at e r  m ar  k d ata   a g a in st   attac k  s   a nd at   t he sa me t ime keeping t he new pixel s  ve r  y c lo s e t o t he o r  igin a l pixel s  in o r  de r    t o p r  o t e ct   t he qu a li t y of w at e r  m ar  ked im a ge. The t e c hnique i s    bas ed on t e st ing t he v a lue of t he w at e r  m ar  k pixel acc o r  ding t o t he ra nge of e ac h  b i t -pl a ne.  K  eywo r  d  s—   Wat  e r  m ar  king,    LSB,    ISB,  R o b u st  ne ss. I.   I  NT R  ODUCTION  n a  digi ta l im a ge w at e r  m ar  king s y st em, info r  m at ion carr  ying w at e r  m ar  k i s  em  b edded in im a ge s . Ide a lly, i t   s hould  b e no  pe rc ep t i  b le diffe r  en c e  b e t ween t he w at e r  m ar  ked a nd o r  igin a l im a ge, a nd t he w at e r  m ar  k s hould  b e diffi c ul t   t o r  emove o r    a l t e r   wi t hou t   t he deg ra d at ion of t he ho st  im a ge [1] [2]. A  w at e r  m ar  k u s u a lly i s   a    b in ar  y s equen c e r  ep r  e s en t ing a   s e r  i a l num  b e r   o r    cr  edi t   car  d num  b e r  , a  logo, a  pi ct u r  e o r    a   s ign at u r  e. I t  i s  u s ed t o  p r  ove t he c opy r  igh t  o r   owne rs hip. Digi ta l w at e r  m ar  king h as    b e c ome a   s ignifi ca n t   t opi c  of c ompu t e r    sc ien c e due t o t he in cr  e as ing popul ar  i t y of t he In t e r  ne t   a nd t he e ss en t i a l need of d ata   s e c u r  i t y [3]. In gene ra l, w at e r  m ar  king sc heme c on s i sts  of: w at e r  m ar  k em  b edding a nd w at e r  m ar  k ex tract ing. Em  b edding w at e r  m ar  king in t o t he im a ge i s  pe r  fo r  med u s u a lly  b y modifying t he im a ge c h aract e r  i st i cs   s u c h as  lumin a n c e v a lue s  o r    tra n s fo r  m dom a in c oeffi c ien ts . Sele ct ion of t he c oeffi c ien ts  depend s  on  pe rc ep t u a l cr  i t e r  i a   as  well as  on a  key in str  umen t ed pe r  mu tat ion t o in cr  e as e t he s e c u r  i t y a nd r  o  b u st ne ss  of t he s y st em. Em  b edding ca n  b e done in a n im a ge dependen t  / independen t   a ddi t ive m a nne r   o r     b y s ome s u  bst i t u t ion me c h a ni s m s . I t  i s  of  t en ne c e ssar  y t o u t ilize Hum a n Vi s u a l Sy st em (HVS) model s  fo r    a d a  p t ively em  b edding t he w at e r  m ar  k. Thi s   ca n r  edu c e t he imp acts  of t he modifi cat ion s  on im a ge qu a li t y o r   fo r    t he sa me vi s u a l qu a li t y a  mu c h str  onge r   w at e r  m ar  k ca n  b e em  b edded [4]. II.   B IT -P L A  NE M ETHOD   A    b i t -pl a ne of digi ta l im a ge s  i s   a   s e t  of  b i ts  h a ving t he sa me  po s i t ion in t he r  e s  pe ct ive  b in ar  y num  b e rs . In g r  ey sca le im a ge A k  ra m M. Zeki i s  wi t h t he F ac ul t y of Compu t e r   S c ien c e a nd Info r  m at ion Sy st em, Unive rs i t y Te c hnology M a l a y s i a , UTM Skud a i, 81310, Joho r  , M a l a y s i a  ( a k  ra mzeki@y a hoo. c om). A ziz a h A . M a n a f i s  wi t h t he College of S c ien c e a nd Te c hnology, Unive rs i t y Te c hnology M a l a y s i a , Ci t y C a mpu s , J a l a n Sem ara k, 54100 K  u a l a  Lumpu r  , M a l a y s i a  ( a ziz a h07@ c i t y ca mpu s .u t m.my). r  ep r  e s en tat ion, t he r  e ar  e 8  b i t -pl a ne s : t he fi rst    b i t -pl a ne c on ta in s   t he s e t  of t he mo st   s ignifi ca n t    b i ts  MSB a nd t he 8 t h    b i t -pl a ne c on ta in s   t he le ast   s ignifi ca n t    b i ts  LSB. The s e t  in  b e t ween i.e. f  r  om 2 nd   t o 7 t h    b i t -pl a ne s   ar  e in t e r  medi at e s ignifi ca n t    b i ts  ISB, as   s hown in Fig. 1. 1   s  t     Bi    t     - pl    a  n e  2  n d  Bi    t     - pl    a  n e   3  r    d  Bi    t     - pl    a  n e  4   t     h  Bi    t     - pl    a  n e   5   t     h  Bi    t     - pl    a  n e   6   t     h  Bi    t     - pl    a  n e   7   t     h  Bi    t     - pl    a  n e   8   t     h  Bi    t     - pl    a  n e   1 st  pixel 2 nd  Pixel 3 r  d  pixel 4 t h  pixel 5 t h  pixel . . . . . . . . . . . . . . . . . . . . . . . . . . . Fig. 1 A    b i t -pl a ne of digi ta l im a ge s  The v a lue of e ac h  b i t  of t he 8  b i t -pl a ne ca n  b e  p r  e s en t ed  b y 2 n-1 , whe r  e n i s  o r  de r   of t he pl a ne start ing f  r  om 1 t o 8. i.e: (2 0  + 2 1  + 2 2  + 2 3  + 2 4  + 2 5  + 2 6  + 2 7 ) = (1 + 2 + 4 + 8 + 16 + 32 + 64 + 128) = 255. The m a ximum v a lue t h at   ca n fi t  in 8  b i ts  i s  255 a nd t he minimum v a lue i s  0. A ny modifi cat ion t o t he 8 t h    b i t -pl a ne will c h a nge t he  pixel v a lue  b y ±1, t he 7 t h    b i t -pl a ne  b y ±2, t he 6 t h    b i t -pl a ne  b y ±4, t he 5 t h    b i t -pl a ne  b y ±8, t he 4 t h    b i t -pl a ne  b y ±16, t he 3 r  d    b i t -pl a ne  b y ±32, t he 2 nd    b i t -pl a ne  b y ±64, a nd t he 1 st    b i t -pl a ne  b y ±128. As   a   r  e s ul t , if t he c h a nged v a lue i s   s m a ll ( s u c h as  in 8 t h    b i t -pl a ne), t he im a ge qu a li t y i s  kep t  high. W hile a    b ig c h a nged v a lue ( s u c h as  1 st    b i t -pl a ne) ca u s e s   t he im a ge qu a li t y t o  b e highly deg ra ded. III.   The R  e s e arc h A  pp r  o ac h In t hi s   st udy, w at e r  m ar  king t e c hnique  bas ed on in t e r  medi at e s ignifi ca n t    b i t  (ISB) h as    b een developed in o r  de r    t o imp r  ove t he r  o  b u st ne ss   a nd t he qu a li t y of w at e r  m ar  ked im a ge. Thi s   st udy u s e s  one  b i t -pl a ne t o A k  ra m M. Zeki a nd A ziz a h A . M a n a f A  Novel Digi ta l Wat e r  m ar  king Te c hnique B as ed on ISB (In t e r  medi at e Signifi ca n t  Bi t ) I Mo st  Signifi ca n t  Bi t  (MSB) Le ast  Signifi ca n t  Bi t  (LSB) World Academy of Science, Engineering and Technology 50 2009989  em  b ed t he w at e r  m ar  k o  b  je ct  in t o a   s ele ct ed  b i t -pl a ne. The nex t   st ep a f  t e r    s ele ct ing one  b i t -pl a ne fo r   em  b edding i s  finding ra nge s  of t he c ho s en  b i t -pl a ne. The leng t h of t he ra nge L i s  2 n-1  (L i s   t he m a ximum v a lue of e ac h ra nge – t he minimum v a lue of t he ra nge + 1) a nd t he num  b e r   of ra nge s  in e ac h  b i t -pl a ne ar  e 256 / L. W e ca n no t i c e t h at  in e ac h ra nge t he  b i t   c h a nge s    b e t ween 0 a nd 1. The num  b e r   of ra nge s  fo r    t he fi rst    b i t -pl a ne ar  e 2 only, as  follow s  [0:127] a nd [128:255]. In o t he r   wo r  d s , t he  b i t  in t he fi rst   ra nge i s  0, while t he  b i t  in t he s e c ond ra nge i s  1 a nd t he leng t h of e ac h ra nge of t he fi rst    b i t -pl a ne i s  128. Fo r    t he s e c ond  b i t -pl a ne t he r  e ar  e 4 ra nge s   as  follow s : [0:63] [64:127] [128:191] [192:255] a nd t he leng t h of t he ra nge s  i s  64, a nd s o on, as   s hown in T ab le I. T A BLE   I R  A  NGES OF E A CH BIT - PL A  NE W ITH THE LENGTH   Bit   -Plane   Length   of    the   ranges   Number   of    ranges   Ranges  1 128 2 [0:127] [128:255] 2 64 4 [0:63] [64:127] [128:191] [192:255] 3 32 8 [0:31] [32:63] … [192:223] [224:255] 4 16 16 [0:15] [16:31] … [224:239] [240:255] 5 8 32 [0:7] [8:15] … [240:247] [248:255] 6 4 64 [0:3] [4:7] … [248:251] [252:255] 7 2 128 [0:1] [2:3] … [252:253] [254:255] 8 1 256 [0] [1] … [254] [255] Du r  ing t he em  b edding t he r  e i s  no c h a nging t o t he pixel if t he sa me  b i t  will  b e em  b edded. In o t he r   wo r  d s , t he sa me ra nge will  b e s ele c t ed t o lo c at e t he w at e r  m ar  ked pixel (i.e. em  b edded  b i t  i s  1 a nd t he s ele c t ed  b i t  of t he o r  igin a l pixel v a lue w as  1 t oo, o r   em  b edding 0 if t he o r  igin a l pixel v a lue w as  0). Bu t  if t he s ele c t ed  b i t  of t he o r  igin a l  b in ar  y pixel i s  no t   t he sa me as   t he em  b edded one, t he p r  eviou s  o r   nex t   ra nge will  b e s ele c t ed. The new ra nge will  b e de t e r  mined depending on i ts  di sta n c e t o t he o r  igin a l pixel v a lue. Howeve r  , s ele c t ing ne ar  e st   ra nge t o t he o r  igin a l pixel will imp r  ove t he qu a li t y of w at e r  m ar  ked im a ge. The  b e st  w at e r  m ar  king r  o  b u st ne ss   c a n  b e o  bta ined if t he  pixel i s  in t he middle of ra nge s , s o t h at   a ny c h a nge on t he pixel  b y atta c k  s  will no t   a ffe c t   t he s ele c t ed  b i t . W hile if t he pixel v a lue i s  lo c at ed in t he edge s  of ra nge s , a ny s m a ll c h a nge  b y atta c k  s  will move t he pixel f  r  om a   ra nge t o t he nex t  o r   p r  eviou s   ra nge. In t hi s   c as e, t he s ele c t ed  b i t  will  b e c h a nged f  r  om 0 t o 1 o r   vi c e ve rsa . Due t o t hi s   c h a nge, t he w at e r  m ar  k c a nno t    b e ex tra c t ed. Me a nwhile t he  b e st  w at e r  m ar  ked im a ge qu a li t y c a n  b e o  bta ined if t he w at e r  m ar  ked pixel ( a f  t e r   em  b edding) i s  ve r  y c lo s e t o t he o r  igin a l pixel. So in c as e t he em  b edded  b i t  i s  no t   t he sa me as   t he o r  igin a l  b i t , t he  b e st  w at e r  m ar  ked im a ge qu a li t y c a n  b e o  bta ined  b y moving t he pixel t o t he lo c at ion in t he edge of ra nge s   t ow ar  d s   t he o r  igin a l pixel P. In t hi s   st udy we will tr  y t o find  b e st  pixel v a lue in  b e t ween t he middle a nd t he edge of t he ra nge t h at   c a n s u r  vive a g a in st  diffe r  en t   t ype s  of atta c k  s   a nd at   t he sa me t ime keeping minimum im a ge di st o rt ion. Thi s  i s  done  b y po s i t ioning t he w at e r  m ar  ked pixel a w a y f  r  om t he edge of t he ra nge. A ss ume t h at   t he  b i as  v a lue (X) i s   at  le ast   t he di sta n c e f  r  om t he po s i t ion of t he w at e r  m ar  ked pixel P` t o t he edge of t he ra nge (whi c h i s  mo r  e c lo s e t o t he o r  igin a l pixel). Th at  me a n s  if t he di sta n c e f  r  om t he pixel t o t he edge of t he ra nge i s  g r  e at e r    t h a n t he  b i as  v a lue, t hen t he po s i t ion of t he pixel will no t   c h a nge. W hile if t he di sta n c e f  r  om t he pixel t o t he edge of t he ra nge i s   s m a lle r    t h a n t he  b i as  v a lue, t hen t he po s i t ion of t he  pixel will c h a nge t o  b e as  f  ar    as   t he  b i as  v a lue. So, t he  b i as  v a lue (X) ∈  [0, L/2 – 1]. In o t he r   wo r  d s , t he minimum of t he  b i as  v a lue i s  0 a nd t he m a ximum of t he  b i as  v a lue in e a c h  b i t -pl a ne i s  L/2 – 1. If t he  b i as  v a lue (X) = (L/2 – 1), t he w at e r  m ar  ked pixel P` will  b e lo c at ed in t he middle of t he ra nge. W hile if t he  b i as  v a lue (X) = 0, t he w at e r  m ar  ked pixel P` will  b e t he sa me as   t he o r  igin a l  pixel, t hi s  i s  in c as e o r  igin a l  b i t  i s   t he sa me as   t he em  b edded one. A nd P` will  b e lo c at ed in t he edge of t he ra nge if t he o r  igin a l  b i t  i s  no t   t he sa me as   t he em  b edded one. T ab le II s how s  diffe r  en t    b i t -pl a ne s  wi t h diffe r  en t    b i as  v a lue s  (X). T A BLE   II T HE R  A  NGE OF THE BI A S V A LUE (X)  FO R   DIFFE R  ENT BIT - PL A  NES . Bit   -Plane   L   X  1 128 [0 - 63] 2 64 [0 - 31] 3 32 [0 - 15] 4 16 [0 - 7] 5 8 [0 - 3] 6 4 [0 - 1] 7 2 [0] 8 1 [0] F r  om t he ab ove tab le, we c a n no t i c e t h at   t he r  o  b u st ne ss  will no t    b e imp r  oved in 7 t h   a nd 8 t h    b i t -pl a ne  b y u s ing t hi s  me t hod,  b e c a u s e t he r  e ar  e no ra nge middle v a lue s . So t he w at e r  m ar  ked pixel will  b e lo c at ed in t he edge s  of t he ra nge. Fin a lly we s hould no t i c e t h at   t he ex tra c t ing sta ge of t he p r  opo s ed me t hod i s  ve r  y e as y, di r  e c t  ex tra c t ing f  r  om t he c ho s en  b i t -pl a ne will give t he w at e r  m ar  k o  b  je c t . IV.   Em  b edding P r  o c e ss   In t hi s   st udy w at e r  m ar  k em  b edding in a ll  b i t -pl a ne s  will  b e done wi t h a ll po ss i  b le  b i as  v a lue s  (X) (fo r   eve r  y em  b edding t he r  o  b u st ne ss   a nd t he qu a li t y of w at e r  m ar  ked im a ge will  b e me as u r  ed). To imp r  ove t he s e c u r  i t y of t he s y st em, t he w at e r  m ar  k o  b  je c t  i s  en c r  yp t ed u s ing t he s h ar  ed key u s ing t he D ata  En c r  yp t ion S ta nd ar  d (DES) a lgo r  i t hm. Fo r    t hi s  pu r   po s e, t he w at e r  m ar  k d ata  i s   s c ra m  b led in o r  de r    t o en s u r  e a ddi t ion a l s e c u r  i t y. One of t he  b e st  me t hod s  h as    b een u s ed he r  e t o en c r  yp t   t he em  b edding po s i t ion a nd de t e r  mine t he pixel  b i t , fo r   em  b edding in t he ho st  im a ge, i s   c a lled t he Ra ndom Pixel M a nipul at ion Te c hnique [5]. To ve r  ify t he r  o  b u st ne ss  of t he p r  opo s ed me t hod, no r  m a lized c r  o ss   c o rr  el at ion NCC will  b e u s ed fo r    t hi s   pu r   po s e [6] [7]. NCC i s   a n impo rta n t  pe r  fo r  m a n c e  p ara me t e r   in a ny ex tra c t ing module. NCC i s  defined in (1). W he r  e W (x,y) i s   t he o r  igin a l w at e r  m ar  k im a ge a nd W ’(x,y) i s   t he ex tra c t ed w at e r  m ar  k im a ge. ∑∑∑∑ =  x y x y  y x W  y xW  y xW   NCC  2 )],([ ),('),(  (1) World Academy of Science, Engineering and Technology 50 2009990  To st udy t he qu a li t y of w at e r  m ar  ked im a ge, t he pe a k s ign a l t o noi s e rat io PSN R   will  b e u s ed fo r    t hi s  pu r   po s e. In gene ra l, a   p r  o c e ss ed im a ge i s   acc ep tab le t o t he hum a n eye s  if i ts  PSN R   i s  g r  e at e r    t h a n 30 dB [8]. The l ar  ge r    t he PSN R  , t he  b e tt e r   i s   t he im a ge qu a li t y. The PSN R   i s  defined as  given  b y (2). dBijijnm PSN   R min j ))(1255(log10 21010210 ∑∑ −=−= −×=  β α   (2)   αij i s   t he pixel of t he c ove r   im a ge in whi c h t he c oo r  din at e s   ar  e (i, j) a nd βij i s   t he pixel of t he w at e r  m ar  ked im a ge in whi c h c oo r  din at e s   ar  e (i, j). (m, n) i s   t he s ize of t he c ove r   im a ge a nd w at e r  m ar  ked im a ge. I t   s h a ll  b e ass umed t h at   t he op t imum v a lue s hould in a ny cas e h a ve PSN R   no t  le ss   t h a n 30. A ny em  b edded s e t  whi c h will h a ve PSN R   i s  le ss   t h a n 30 will  b e igno r  ed. The  b e st  NCC v a lue will  b e c ho s en as   a    b e st  em  b edding stat u s . The p r  opo s ed me t hod ca n  b e p r  e s en t ed  b y few st ep s   as  follow s : S t ep 1: Sele ct   t he  b i t -pl a ne one  b y one, f  r  om 0 t o 8. S t ep 2: Find t he leng t h of t he ra nge of t he s ele ct ed  b i t -pl a ne  b y L = 2 n-1  (n fo r   1 st    b i t -pl a ne i s  8 while fo r   8 t h    b i t - pl a ne i s  1). S t ep 3: C r  e at e tab le ra nge s  of t he s ele ct ed  b i t -pl a ne (Num  b e r   of ra nge s  i s  256 / L). S t ep 4: Fo r   e ac h ra nge, t wo pixel s   ar  e t o  b e found: t he m a ximum v a lue of t he ra nge a nd t he minimum v a lue of t he ra nge, s o t h at  L = m a ximum v a lue – minimum v a lue +1. S t ep 5: E ac h ra nge i s  divided in t o t wo equ a l g r  oup s ; t he leng t h of e ac h g r  oup i s  (L/2). •   In cas e t he o r  igin a l  b i t  i s  equ a l t o t he em  b edded  b i t , t he di sta n c e d,  b e t ween t he o r  igin a l pixel a nd t he ne ar  e st  edge of t he ra nge, i s   t o  b e found a nd t he following st ep s   ar  e done: o   If t he o r  igin a l pixel i s  in t he lef  t -h a nd g r  oup, d = O r  igin a l Pixel P – minimum pixel v a lue of t he ra nge.    If (X ≤ d), t he w at e r  m ar  ked pixel P` = P.    O t he r  wi s e, t he w at e r  m ar  ked pixel P` = minimum pixel v a lue of t he ra nge + X. o   If t he o r  igin a l pixel i s  in t he r  igh t -h a nd g r  oup, d = m a ximum pixel v a lue of t he ra nge  – O r  igin a l pixel P.    If (X ≤ d), t he w at e r  m ar  ked pixel P` = P.    O t he r  wi s e, t he w at e r  m ar  ked pixel P` = m a ximum pixel v a lue of t he ra nge – X. •   In cas e t he o r  igin a l  b i t  i s  diffe r  en t  f  r  om t he em  b edded  b i t , t he following st ep s   ar  e t o  b e done: o   If t he o r  igin a l pixel i s  in t he lef  t -h a nd g r  oup, o r   in l ast   ra nge, t he w at e r  m ar  ked pixel P` = m a ximum pixel v a lue of t he p r  eviou s   ra nge – X. o   If t he o r  igin a l pixel i s  in t he r  igh t -h a nd g r  oup, o r   in t he fi rst   ra nge, t he w at e r  m ar  ked  pixel P`= minimum pixel v a lue of t he nex t   ra nge + X. S t ep 6: C a l c ul at e t he PSN R    a nd NCC fo r   e ac h em  b edding. S t ep 7: If PSN R   ≥ 30 d  b  find t he highe st  NCC a nd sa ve t he stat u s  whi c h i s   c on s ide r  ed t he  b e st   stat u s  fo r   em  b edding. V.   R  e s ul ts   a nd A n a ly s i s   The w at e r  m ar  k o  b  je ct  (logo im a ge) whi c h i s  in g r  ey sca le level im a ge i s   s hown in Fig. 2. I t   c on ta in s  90 × 90  pixel s   a nd will  b e em  b edded wi t hin t he ho st  g r  ey sca le level im a ge whi c h i s   s hown in Fig. 3, a nd c on ta ining 256 × 256 pixel s . The em  b edding done wi t h a ll  b i t -pl a ne s   a nd few diffe r  en t   t ype s  of attac k  s  h a ve  b een a  pplied t o t he w at e r  m ar  ked im a ge in o r  de r    t o t e st   t he r  o  b u st ne ss  of t he  p r  opo s ed t e c hnique. Fig. 2 G r  ey sca le logo wi t h 90 × 90 pixel s  Fig. 3 G r  ey sca le ho st  im a ge wi t h 256 × 256 pixel s  The fi rst   attac k u s ed he r  e i s  JPEG, whi c h i s   t ype of c omp r  e ss ion t h at  minimize s   t he a moun t  of c olou r   info r  m at ion in t he im a ge t h at  i s   act u a lly no t  di st ingui s h ab le  b y t he hum a n eye. The qu a li t y of t he c omp r  e ss ion u s ed i s  85. The s e c ond attac k i s    b lu rr  ing t he im a ge, t hi s   t ype of fil t e r    t h at   s of  t en s   a n im a ge a nd m a ke s  i t  look  s    b lu rr  ed. Thi s  fil t e r    r  e t u r  n s   a   c i rc ul ar    a ve ra ging fil t e r   (pill  b ox) wi t hin t he s qu ar  e m atr  ix of s ide 2 × ra diu s +1 [9]. The ra diu s  u s ed he r  e i s  1. The t hi r  d attac k i s  G a u ss i a n fil t e r   [10], whi c h i s  de s igned t o p ass   a   st ep fun ct ion wi t h ze r  o ove rs hoo t   a nd minimum r  i s e t ime; t hi s  fil t e r    r  e t u r  n s   a   r  o tat ion a lly s ymme tr  i c  G a u ss i a n low p ass  fil t e r   of s ize [3 3]. I t  h as    b een u s ed he r  e wi t h sta nd ar  d devi at ion s igm a  (0.5). The fo rt h attac k i s   W iene r   fil t e r  , t he go a l of t hi s  fil t e r   i s   t o fil t e r   ou t  noi s e t h at  h as   c o rr  up t ed a   s ign a l. W iene r   me t hod i s    bas ed on stat i st i cs  e st im at ed f  r  om a  lo ca l neigh  b ou r  hood of e ac h pixel of s ize m-  b y-n. To e st im at e t he lo ca l im a ge me a n, sta nd ar  d devi at ion in t hi s   st udy (m a nd n) i s  3. W iene r   fil t e r   e st im at e s   t he a ddi t ive noi s e  powe r     b efo r  e doing t he fil t e r  ing [11]; t he noi s e in t hi s   st udy i s  0.001. The fif  t h attac k u s ed he r  e i s  Spe c kle noi s e; t hi s  me t hod a dd s  mul t ipli cat ive noi s e t o t he im a ge wi t h me a n 0 a nd v ar  i a n c e -in t hi s   st udy- 0.01 [11]. Fin a lly geome tr  i c   tra n s fo r  m attac k  s   s u c h as   R  o tat ion a nd S ca ling h a ve  b een t e st ed he r  e, r  o tat ion wi t h a ngle 1° h as    b een a  pplied a nd sca ling wi t h 50% f  r  om t he o r  igin a l im a ge i s   a  pplied. O  b viou s ly t he r  o  b u st ne ss   t o r  o tat ion a nd sca ling ca n only  b e ac hieved a f  t e r    r  e- r  o tat ing o r    r  e- sca ling t he im a ge t o i ts  o r  igin a l s ize. The pe a k s ign a l t o noi s e rat io (PSN R  ) a nd t he no r  m a lized cr  o ss   c o rr  el at ion (NCC) fo r    t he p r  opo s ed World Academy of Science, Engineering and Technology 50 2009991  me t hod fo r    a ll po ss i  b le  b i as  v a lue s  h a ve  b een ca l c ul at ed a f  t e r    a  pplying c ho s en attac k  s   as   s hown in T ab le III, IV, V, VI, VII, VIII, IX a nd X fo r    a ll  b i t -pl a ne s  1 st , 2 nd , 3 r  d , 4 t h , 5 t h , 6 t h , 7 t h , a nd 8 t h   r  e s  pe ct ively.  No t i c e t h at   t he PSN R   h as    b een ca l c ul at ed a f  t e r    t he em  b edding t he d ata   a nd  b efo r  e a  pplying a ny attac k  s . A l t hough s ome attac k  s  imp r  ove t he qu a li t y of t he im a ge s u c h as : fil t e r  ing a nd c omp r  e ss ion, s ome o t he rs  de str  oy t he im a ge s u c h as :  b lu rr  ing a nd noi s e. T A BLE III   T HE PSN R    A  ND  NCC  FO R   THE P R  OPOSED METHOD FO R   THE 1 ST BIT - PL A  NE   X   PSN R    NCC   (JPEG)   NCC   (Blu rrin g)   NCC   (G a u ssian )   NCC   ( Winner)   NCC   (Speckle)  0 16.13647 0.795478 0.744931 0.7323 0.7056 0.779114 1 16.00184 0.842855 0.749759 0.7484 0.8473 0.805124 2 15.86849 0.880796 0.758538 0.7586 0.9298 0.844149 3 15.73641 0.90722 0.77309 0.7755 0.9675 0.872815 4 15.6056 0.928738 0.786227 0.7887 0.9843 0.906456 5 15.47602 0.944807 0.799774 0.803 0.989 0.938923 6 15.34768 0.961884 0.815191 0.8168 0.9921 0.970004 7 15.22055 0.972511 0.82719 0.8324 0.993 1 8 15.09462 0.978956 0.84503 0.8522 0.9942 1 9 14.96986 0.985035 0.865521 0.873 0.9959 1 10 14.84627 0.988114 0.879178 0.8858 1 1 11 14.72382 0.99132 0.889336 0.8977 1 1 12 14.6025 0.994532 0.902111 0.91 1 1 13 14.4823 0.996114 0.916023 0.9248 1 1 14 14.36318 0.997196 0.930629 0.935 1 1 15 14.24515 0.997998 0.937078 0.939 1 1 16 14.12817 0.998393 0.940395 0.9438 1 1 17 14.01224 0.99891 0.944081 0.9504 1 1 18 13.89733 0.998948 0.946434 0.9551 1 1 19 13.78344 0.999326 0.953067 0.9614 1 1 20 13.67054 0.999553 0.959391 0.9656 1 1 21 13.55862 0.999828 0.96494 0.9707 1 1 22 13.44765 0.999859 0.969446 0.9739 1 1 23 13.33764 0.999903 0.972865 0.9759 1 1 24 13.22855 0.999896 0.974027 0.9784 1 1 25 13.12037 0.999888 0.975839 0.9806 1 1 26 13.0131 0.999992 0.978075 0.9834 1 1 27 12.9067 0.999992 0.979552 0.9857 1 1 28 12.80117 1 0.981525 0.9879 1 1 29 12.69649 1 0.983731 0.9897 1 1 30 12.59264 1 0.985815 0.9913 1 1 31 12.48962 1 0.988688 0.9934 1 1 32 12.38739 1 0.990442 0.9945 1 1 33 12.28595 1 0.992826 0.9959 1 1 34 12.18528 1 0.994241 0.9963 1 1 35 12.08536 1 0.994502 0.9967 1 1 36 11.98619 1 0.994972 0.9973 1 1 37 11.88775 1 0.995754 0.9975 1 1 38 11.79002 1 0.995827 0.998 1 1 39 11.693 1 0.995935 0.9979 1 1 40 11.59667 1 0.99628 0.9982 1 1 41 11.50102 1 0.996645 0.9984 1 1 42 11.40605 1 0.997328 0.9985 1 1 43 11.31174 1 0.997388 0.9986 1 1 44 11.21808 1 0.997833 0.9987 1 1 45 11.12507 1 0.997954 0.9989 1 1 46 11.03269 1 0.997904 0.9993 1 1 47 10.94094 1 0.997902 0.9995 1 1 48 10.84981 1 0.998106 0.9996 1 1 49 10.75928 1 0.998338 0.9997 1 1 50 10.66936 1 0.998509 0.9999 1 1 51 10.58002 1 0.998673 1 1 1 52 10.49128 1 0.998808 1 1 1 53 10.40311 1 0.999242 1 1 1 54 10.31551 1 0.999349 1 1 1 55 10.22848 1 0.99942 1 1 1 56 10.14201 1 0.999882 1 1 1 57 10.0561 1 0.999947 1 1 1 58 9.97073 1 0.999955 1 1 1 59 9.885903 1 0.99995 1 1 1 60 9.80161 1 0.999948 1 1 1 61 9.717845 1 0.999954 1 1 1 62 9.634601 1 0.999954 1 1 1 63 9.551874 1 0.999954 1 1 1 World Academy of Science, Engineering and Technology 50 2009992  T A BLE IV   T HE PSN R    A  ND  NCC  FO R   THE P R  OPOSED METHOD FO R   THE 2  ND BIT - PL A  NE   X   PSN R    NCC   (JPEG)   NCC   (Blurring)   NCC   (Gaussian)   NCC   (Winner)   NCC   (Speckle)  0 24.06413 0.711555 0.645592 0.635 0.5851 0.697784 1 23.75336 0.785238 0.702612 0.707 0.6442 0.75555 2 23.44878 0.842088 0.755879 0.7629 0.7319 0.810577 3 23.15018 0.876803 0.799364 0.8048 0.7955 0.864797 4 22.85737 0.909485 0.832591 0.835 0.8536 0.896109 5 22.57019 0.931049 0.853794 0.8608 0.8958 0.92045 6 22.28847 0.946609 0.874192 0.8846 0.9197 0.941476 7 22.01203 0.959485 0.889573 0.8969 0.9359 0.955118 8 21.74073 0.969397 0.899595 0.9102 0.9515 0.969498 9 21.47442 0.977538 0.91829 0.918 0.9654 0.979595 10 21.21296 0.983219 0.922482 0.9241 0.9943 0.992713 11 20.95622 0.98888 0.925283 0.9319 0.9952 1 12 20.70405 0.992618 0.932704 0.9406 1 1 13 20.45631 0.993836 0.938843 0.948 1 1 14 20.21286 0.996574 0.945634 0.953 1 1 15 19.97358 0.99799 0.953053 0.9564 1 1 16 19.73833 0.998516 0.958725 0.96 1 1 17 19.50699 0.998988 0.960425 0.9617 1 1 18 19.27942 0.999589 0.962984 0.9669 1 1 19 19.05551 0.999911 0.963277 0.9698 1 1 20 18.83511 0.999923 0.964398 0.9745 1 1 21 18.61813 0.999984 0.965228 0.9754 1 1 22 18.40444 1 0.967169 0.9765 1 1 23 18.19394 1 0.968555 0.9773 1 1 24 17.98652 1 0.973079 0.9783 1 1 25 17.78207 1 0.982227 0.9793 1 1 26 17.58049 1 0.983041 0.9798 1 1 27 17.38168 1 0.983431 0.9823 1 1 28 17.18553 1 0.983483 0.984 1 1 29 16.99196 1 0.983458 0.9845 1 1 30 16.80087 1 0.983453 0.9872 1 1 31 16.61218 1 0.983453 0.9885 1 1 T A BLE V   T HE PSN R    A  ND  NCC  FO R   THE P R  OPOSED METHOD FO R   THE 3 R  D BIT - PL A  NE   X   PSN R    NCC   (JPEG)   NCC   (Blurring)   NCC   (Gaussian)   NCC   (Winner)   NCC   (Speckle)  0 31.0378 0.748387 0.754281 0.7545 0.7202 0.760441 1 30.3148 0.800633 0.795909 0.8016 0.7515 0.815772 2 29.62676 0.841893 0.824179 0.8319 0.7935 0.850114 3 28.97074 0.877778 0.844959 0.8578 0.8291 0.887326 4 28.34418 0.902601 0.880603 0.8828 0.865 0.917557 5 27.74483 0.922822 0.888579 0.8966 0.8902 0.937661 6 27.17071 0.9387 0.901801 0.911 0.903 0.956092 7 26.62007 0.954276 0.910795 0.9215 0.9162 0.968579 8 26.09127 0.964553 0.91888 0.9296 0.9269 0.984737 9 25.58283 0.972665 0.92555 0.9411 0.9471 0.996588 10 25.09338 0.980038 0.931065 0.9532 0.9903 0.9991 11 24.62168 0.986049 0.934879 0.9601 0.9933 0.99957 12 24.16658 0.989564 0.960694 0.9634 1 0.999761 13 23.72701 0.991286 0.963537 0.9674 1 1 14 23.30198 0.993667 0.963251 0.9717 1 1 15 22.89059 0.995542 0.963181 0.9759 1 1 T A BLE VI T HE PSN R    A  ND  NCC  FO R   THE P R  OPOSED METHOD FO R   THE 4 TH BIT - PL A  NE   X   PSN R    NCC   (JPEG)   NCC   (Blurring)   NCC   (Gaussian)   NCC   (Winner)   NCC   (Speckle)  0 0 37.23247 0.697749 0.726566 0.7358 0.7012 1 1 35.77832 0.752892 0.773111 0.7791 0.7374 2 2 34.46706 0.799402 0.821054 0.8221 0.7715 3 3 33.27344 0.844238 0.844937 0.8493 0.7952 4 4 32.17838 0.875453 0.865212 0.874 0.8295 5 5 31.16731 0.902997 0.876462 0.9005 0.8579 6 6 30.229 0.9203 0.9108 0.9115 0.8885 7 7 29.35279 0.922757 0.916648 0.9259 0.9062 T A BLE VII T HE PSN R    A  ND  NCC  FO R   THE P R  OPOSED METHOD FO R   THE 5 TH BIT - PL A  NE . X   PSN R    NCC   (JPEG)   NCC   (Blurring)   NCC   (Gaussian)   NCC   (Winner)   NCC   (Speckle)  0 42.41361 0.597743 0.654393 0.6589 0.6177 0.620673 1 39.78235 0.666284 0.721876 0.7269 0.6631 0.659024 2 37.5967 0.726554 0.774336 0.7856 0.6895 0.686901 3 35.72715 0.766078 0.837519 0.8244 0.7232 0.708491 World Academy of Science, Engineering and Technology 50 2009993
Search
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks