3:2 pull-down film mode detection using fuzzy logic

1. A method for 3:2 pull-down film mode detection of a stream of NTSC video fields {F[j]}={ . . . F[n−2], F[n−1], F[n], . . . }, said method comprising the steps of:

a) generating from said {F[j]} a sequence {S[j]}, wherein a S[n] in said sequence denotes an accumulated sum of absolute value of luma difference between two fields F[n] and F[n−2] over a common set of pixel positions;

b) generating from said {S[j]} a sequence {P[j]} whose elements are in one-to-one correspondence with the elements of said sequence {S[j]}, wherein a P[n] is assigned a symbol from a two-symbol set by applying a set of fuzzy inference rules; and

c) in response to said {P[j]} having a pattern that characterized a stream of 3:2 pull-downed fields, designating said {F[j]} as 3:2 pull-downed fields from film source.

2. The method of claim 1 , wherein said two-symbol set is {L,H} and wherein said set of fuzzy inference rules are:

with X[n−1]=|S[n]−S[n−1]|, X[n−2]=|S[n]−S[n−2]|, X[n−3]=|S[n]−S[n−3]|, X[n−4]=|S[n]−S[n−4]|, and with fuzzy sets of BIG and SMALL defined to characterize sized of X[n−1], X[n−2], X[n−3] and X[n−4],

if X[n−1] is BIG, and X[n−2] is BIG, and X[n−3] is BIG, and X[n−4] is BIG,

then P[n] is assigned symbol ‘L’;

otherwise, P[n] is symbol ‘H’.

3. The method of claim 2 , further comprising the steps of:

performing symbol assignment to said P[n] by performing fuzzy logic inference algorithm to implement said set of fuzzy logic inference rules.

4. The method of claim 3 , wherein said fuzzy logic inference algorithm comprises the step of:

defining a fuzzy subset per each fuzzy logic inference rule of said set of fuzzy logic inference rules, thereby resulting in a plurality of fuzzy subsets;

forming a fuzzy union set of said plurality of fuzzy subsets; and

defuzzifying said fuzzy union set by using a center of gravity defuzzification method to generate a value g adapted to determine symbol assignment of said P[n].

5. The method of claim 2 , wherein said pattern is selectable from a group consisted of:

LHHHH repeated m times in said {P[j]},

HHHHL repeated m times in said {P[j]},

HHHLH repeated m times in said {P[j]},

HHLHH repeated m times in said {P[j]}, and

HLHHH repeated m times in said {P[j]}, wherein said m is a positive integer.

6. The method of claim 1 , wherein in said step c), the signs of Z[n−1]=(S[n]−S[n−1]), Z[n−2]=(S[n]−S[n−2]), Z[n−3]=(S[n]−S[n−3]) and Z[n−4]=(S[n]−S[n−4]) are adapted as a criteria to rule out erroneously designating said {F[j]} as 3:2 pull-downed fields form film source.

7. A method for 3:2 pull-down film mode detection of a stream of NTSC video fields {F[j]}={ . . . F[n−2], F[n−1], F[n], . . . }, said method comprising the steps of:

a) generating from said {F[j]} a sequence {S[j]}, wherein a S[n] in said sequence denotes an accumulated sum of absolute value of luma difference between two fields F[n] and F[n−2] over a common set of pixel positions:

b) for said S[n], characterizing the sizes of X[n−1]=|S[n]−S[n−1]|, X[n−2]=S[n]−S[n−2]|, X[n−3]=|S[n]−S[n−3]|, X[n−4]=|S[n]−S[n−4]| with their fuzzy memberships in fuzzy logic sets;

c) generating from said {S[j]} a reference sequence {P[j] whose elements correspond one-to-one to elements of {S[j]}, wherein a P[n] corresponding to said S[n] is assigned a symbol from a set having two symbols by applying fuzzy inference rules, and wherein said P[n] is adapted to summarize the fuzzy set memberships of said X[n−1], X[n−2], X[n−3] and X[n−4] in said fuzzy sets; and

in response to said {P[j]} containing a pattern that characterizes a stream of 3:2 pull-downed fields, designating said {F[j]} as set of 3:2 pull-downed fields from film source.

8. The method of claim 7 , wherein said set of two symbols is {L, H}, wherein said fuzzy sets are defined as BIG and SMALL, and wherein said set of fuzzy inference rules are:

if X[n−1] is BIG, and X[n−2] is BIG, and X[n−3] is BIG, and X[n−4] is BIG then

P[n] is assigned symbol ‘L’;

otherwise, P[n] is symbol ‘H’.

9. The method of claim 8 , further comprising the steps of:

performing symbol assignment to said P[n] by performing fuzzy logic inference algorithm to implement said set of fuzzy logic inference rules.

10. The method of claim 9 , wherein said fuzzy logic inference algorithm comprises the step of:

defining a fuzzy subset per each fuzzy logic inference rule of said set of fuzzy logic inference rules, thereby resulting in a plurality of fuzzy subsets;

forming a fuzzy union set of said plurality of fuzzy subsets; and

defuzzifying said fuzzy union set by using a center of gravity defuzzification method to generate a value g adapted to determine symbol assignment of said P[n].

11. The method of claim 8 , wherein said pattern is selectable from a group consisted of:

LHHHH repeated m times in said {P[j]},

HHHHL repeated m times in said {P[j]},

HHHLH repeated m times in said {P[j]},

HHLHH repeated m times in said {P[j]}, and

HLHHH repeated m times in said {P[j]}, wherein said m is a positive integer.

12. The method of claim 7 , wherein in said step c), the signs of Z[n−1]=(S[n]−S[n−1]), Z[n−2]=(S[n]−S[n−2], Z[n−3]=(S[n]−S[n−3]) and Z[n−4]=(S[n]−S[n−4]) are adapted as a criteria to rule our erroneously designating said {F[j]} as 3:2 pull-downed fields from film source.

13. A system for performing 3:2 pull-down film mode detection on a NTSC stream of video fields F[j]}={ . . . F[n−2], F[n−1], F[n], . . . }, said system comprising:

a difference accumulate unit adapted for generating from said {F[n]} a sequence {S[j]}, wherein a S[n] denotes an accumulation sum of absolute value of luma difference between two fields F[n] and F[n−2] over a common set of pixel positions;

a fuzzy logic decision unit adapted for generating from said {S[j]} a sequence {P[j]} whose elements are in one-to-one correspondence with the elements of said sequence {S[j]}, wherein a P[n] corresponding to said S[n] is assigned a symbol from a set of two symbols by applying a set of fuzzy inference rules; and

a decision unit adapted for designating said {F[j]} by said {P[j]} as 3:2 pull-downed fields in response to said {P[j]} having a pattern that characterizes a stream of 3:2 pull-downed fields from film source.

14. The system of claim 13 , wherein said set of two symbols is {L,H}, and wherein said set of fuzzy inference rules are:

with fuzzy sets of BIG and SMALL defined to characterize sizes of X[n−1], X[n−2], X[n−3] and X[n−4] by fuzzy memberships of X[n−1], X[n−2], X[n−3] and X[n−4} in said fuzzy sets BIG and SMALL, wherein X[n−1]=|S[n]−S[n−1]|, X[n−2]=|S[n]−S[n−2]|, X[n−3]=|S[n]−S[n−3]|, X[n−4]=|S[n]−S[n−4]|,

if X[n−1] is BIG and X[n−2] is BIG, and X[n−3] is BIG, and X[n−4] is BIG,

then P[n] is assigned symbol ‘L’;

otherwise, P[n] is symbol ‘H’.

15. The system of claim 14 , wherein said fuzzy logic decision unit is adapted to perform symbol assignment to said P[n] by performing fuzzy logic inference algorithm to implement said set of fuzzy logic inference rules.

16. The system of claim 13 , wherein said fuzzy logic inference algorithm performed by said fuzzy decision unit comprises the step of:

defining a fuzzy subset per each fuzzy logic inference rule of said set of fuzzy logic inference rules, thereby resulting in a plurality of fuzzy subsets;

forming a fuzzy union set of said plurality of fuzzy subsets; and

defuzzifying said fuzzy union set by using a center of gravity defuzzification method to generate a value g adapted to determine symbol assignment of said P[n].

17. The system of claim 14 , wherein said pattern is selectable from a group consisted of:

LHHHH repeated m times in said {P[j]},

HHHHL repeated m times in said {P[j]},

HHHLH repeated m times in said {P[j]},

HHLHH repeated m times in said {P[j]}, and

HLHHH repeated m times in said {P[j]}, wherein said m is a positive integer.

18. The method of claim 13 , wherein said decision unit uses the signs of Z[n−1]=(S[n]−S[n−1]), Z[n−2]=(S[n]−S[n−2]), Z[n−3]=(S[n]−S[n−3]) and Z[n−4]=(S[n]−S[n−4]) as a criteria to rule our erroneously designating said {F[j]} as 3:2 pull-downed fields from film source.