Method and apparatus for adaptive progressive scan conversion

• US 5,386,237 A
• Filed: 11/19/1993
• Issued: 01/31/1995
• Est. Priority Date: 11/23/1992
• Status: Expired due to Term

• Current Assignee: Thomson Consumer Electronics SA (Vantiva SA)
• Original Assignee: Thomson Consumer Electronics SA (Vantiva SA)
• Inventors: Knee, Michael
• Primary Examiner(s): Kostak, Victor R.
• Application Number:
  US08/155,113
• Time in Patent Office: 438 Days
• Field of Search: 348/458, 348/448, 348/618-620, 348/625
• US Class Current: 348/458
• CPC Class Codes: H04N 7/012 Conversion between an interlaced and a progressive signal for graphics images G09G2310/0229

Abstract

Extra lines (Yout) for display be progressive scanning are obtained by adaptive interpolation from the original (interlaced) lines. The original lines are left untouched. Each interpolated pixel is the result of a soft switch (456,461,462) between pure interframe interpolation (452), which is the more suitable for areas without moving detail, and interpolation (451) using the Diag-3W algorithm, which is the more suitable for moving areas. The soft switch coefficient (S) is calculated using a normalized comparison between two estimates (E_v,E_t) of the likely error resulting from each of the two kinds of interpolation. To limit the occurrence of wrong decisions due to noise or to certain structures in the picture, the decision is spread horizontally, vertically and temporally, but only over a very small area so that the decision remains local to the pixel (X) being interpolated.

Claims

1. A method for calculating sample values of interpolated lines for each sample for adaptive proscan conversion of interlaced lines, comprising:

calculating a spatial error estimate from spatially and temporally adjacent pixels of adjacent fields;

calculating a temporal error estimate from spatially and temporally adjacent pixels of adjacent fields;

calculating a normalized error estimate from the spatial error estimate and the temporal error estimate;

controlling a soft switch in accordance with the normalized error estimate for mixing a spatially interpolated estimate value and a temporally interpolated estimate value for said sample to generate the final value for said sample;

calculating said spatially interpolated estimate value from spatially adjacent pixels of the current field; and

calculating said temporally interpolated estimate value from temporally adjacent pixels of adjacent fields.

2. A method according to claim 1, further comprising forming respective pixel differences within each of the adjacent fields for calculating said spatial error estimate.

3. A method according to claim 1 further comprising forming respective pixel pair differences between the adjacent fields for calculating said temporal error estimate.

4. A method according to claim 1 further comprising forming pixel differences corresponding to a detected picture structure direction for calculating said spatially interpolated estimate value.

5. A method according to claim 2 further comprising the step of selecting the lower, especially the minimum, of said pixel differences for calculating said spatial error estimate.

6. A method according to claim 3 further comprising the step of selecting the higher, especially the maximum, of said pixel pair differences for calculating said temporal error estimate.

7. A method according to claim 1 further comprising

forming respective pixel differences within each of the adjacent fields and selecting the second minimum of said pixel differences for calculating said spatial error estimate;

and

forming respective pixel pair differences between adjacent fields and selecting the second maximum of said pixel pair differences for calculating said temporal error estimate.

8. A method according to claim 1 wherein the step of calculating said normalized error estimate, En, is calculated according to the formula:

##EQU6## where Et is the temporal error estimate, Ev is the spatial error estimate, P is a pre-selected value and E_n is then limited to;

##EQU7## to form a limited normalized error estimate, E'"'"'n, where ε

represents one least significant bit in the representation of E_n.

9. A method according to claim 8 further comprising filtering said limited normalized error estimate horizontally for controlling said soft switch.

10. A method according to claim 8 wherein said sample is of a given word length and further comprising selecting a word length less than the word length of said sample for calculation of said spatial error estimate (E_v), said temporal error estimate (E_t), said normalized error estimate (E_n) and/or said limited normalized error estimate (E'"'"'_n).

11. Apparatus for adaptive proscan conversion of interlaced lines by calculating sample values of the missing lines for each sample, comprising:

a vertical error circuit in which from spatially and temporally adjacent pixels of adjacent fields a spatial error estimate is calculated;

a temporal error circuit in which from spatially and temporally adjacent pixels of adjacent fields a temporal error estimate is calculated;

a soft switch processor in which the spatial error estimate and the temporal error estimate are combined to form a normalized error estimate;

a spatial interpolation circuit in which a spatially interpolated estimate value is calculated from spatially adjacent pixels of the current field;

a temporal interpolation circuit in which a temporally interpolated estimate value is calculated from temporally adjacent pixels of adjacent fields;

a soft switch, controlled by the normalized error estimate and which mixes accordingly the spatially interpolated estimate value and the temporally interpolated estimate value for said sample to generate the final value for said sample; and

a line speedup circuit for generating the proscan output signal from said interlace lines and said sample values.

12. Apparatus according to claim 11, characterized in that said normalized error estimate (E_n) is calculated according to the formula:

##EQU8## where En is said normalized error estimate, Et is said temporal error estimate, Ev is said spatial error estimate, P is a pre-selected value and E_n is then limited to;

##EQU9## to form a limited normalized error estimate (E'"'"'_n), where ε

represents one least significant bit in the representation of E_n.

13. Apparatus according to claim 11 wherein said limited normalized error estimate (E'"'"'_n) is filtered in a horizontal filter, the output signal of which controls said soft switch.

14. Apparatus according to claim 12 wherein said limited normalized error estimate is filtered in a horizontal filter, the output signal of which controls said soft switch.

15. Apparatus according to claim 11 wherein said spatial error estimate, said temporal error estimate, said normalized error estimate and/or said limited normalized error estimate are calculated in the related circuits with a word length less than the word length of said sample.

16. Apparatus according to claim 12 wherein said spatial error estimate, said temporal error estimate, said normalized error estimate and/or said limited normalized error estimate are calculated in the related circuits with a word length less than the word length of said sample.

17. Apparatus according to claim 13 wherein said spatial error estimate, said temporal error estimate, said normalized error estimate and/or said limited normalized error estimate are calculated in the related circuits with a word length less than the word length of said sample.

18. Apparatus according to claim 14 wherein said spatial error estimate, said temporal error estimate, said normalized error estimate and/or said limited normalized error estimate are calculated in the related circuits with a word length less than the word length of said sample.