Automated inverse telecine conversion
View Patent ↗The present invention is generally directed to automated methods and systems for converting image streams having a first frame rate to a second frame rate without the need for user intervention. Embodiments of the present invention obviate the effects of processing of a telecine process. In one embodiment, where frames are encoded by a single video field, a statistical analysis of the differences between adjacent frames reveals a telecine pattern, thereby identifying which frames to remove. In another embodiment, where frames are encoded by even and odd video fields, which are interleaved to produce the frame, a statistical analysis of the differences between adjacent fields reveals the telecine pattern, identifies which frames to remove, and identifies frames that are candidates for re-interleaving.
1. A method of detecting a telecine pattern embedded within a first sequence of video frames, the method comprising:
receiving at least a portion of the frames from the first sequence of video frames;
measuring a plurality of differences between frames and their adjacent frames in the sequence and storing the measure differences;
arranging the measured differences into a plurality of data groups for analysis wherein each data group of the plurality comprises an “in-group” and one or more “out-groups,” wherein each in-group comprises two or more selected measured differences between adjacent frames that are spaced 5 frame intervals apart, wherein each corresponding out-group comprises other measured differences between adjacent frames;
performing a statistical analysis of the in-groups and the out-groups of the plurality of data groups to determine which data group has the in-group that corresponds to duplicate frames of the telecine pattern; and
generating a second sequence of video frames from the first sequence by removing one or more duplicate frames from the first sequence.
2. The method as defined in claim 1 , wherein the measurement of the differences between frames includes comparing every fourth pixel of a first frame to every fourth pixel of a second frame.
3. The method as defined in claim 1 , wherein the measurement of the differences between frames includes computing a summation of an absolute sum of differences between pixels.
4. The method as defined in claim 1 , wherein the measurement of the differences between frames includes computing a summation of the square of the differences between pixels.
5. The method as defined in claim 1 , wherein the measurement of the differences between frames is further normalized according to a number of pixels per frame that are compared.
6. The method as defined in claim 1 , wherein the measurement of the differences between frames is further saturated to a predetermined level.
7. The method as defined in claim 1 , further comprising:
maintaining the stored differences in a collection;
shifting the stored differences in the collection in response to a new measurement of differences between frames; and
adding the new measurement of differences to the collection.
8. The method as defined in claim 1 , wherein the statistical analysis includes a computation of a mean.
9. The method as defined in claim 1 , wherein the statistical analysis includes a computation of a variance.
10. The method as defined in claim 1 , wherein the statistical analysis includes a computation of a standard deviation.
11. The method as defined in claim 1 , further comprising varying a sample size of the measured differences analyzed in the plurality of data groups in response to a failure to detect the telecine pattern in a larger portion.
12. The method as defined in claim 1 , further comprising varying a threshold used for detection of the telecine pattern at least in part based on a selected sample size.
13. The method as defined in claim 1 , wherein arranging comprises arranging a sample size of 20 measured differences of adjacent frames.
14. The method as defined in claim 1 , wherein the plurality of data groups comprises 5 data groups, wherein each of the in-groups of the 5 data groups comprises a difference frame position within 5 consecutive frame positions.
15. The method as defined in claim 1 , wherein each data group comprises the in-group and one out-group, wherein the one out-group comprises the measured differences for a remainder of the frame differences not included in the in-group.
16. The method as defined in claim 1 , wherein arranging comprises initially arranging a sample size of 20 measured differences of adjacent frames, further comprising decreasing the sample size to 15 measured differences in response to a failure to detect the telecine pattern in the sample size of 20.
17. A system that detects a telecine pattern embedded within a first sequence of video frames, the system comprising:
a server adapted to receive at least a portion of the frames from the first sequence of video frames;
means for measuring a plurality of differences between frames and their adjacent frames in the sequence and storing the measure differences;
means for arranging the measured differences into a plurality of data groups for analysis wherein each data group of the plurality comprises an “in-group” and one or more “out-groups,” wherein each in-group comprises two or more selected measured differences between adjacent frames that are spaced 5 frame intervals apart, wherein each corresponding out-group comprises other measured differences between adjacent frames; and
means for performing a statistical analysis of the plurality of data groups to determine which data group has the in-group that corresponds to duplicate frames of the telecine pattern.
18. The system as defined in claim 17 , wherein the means for measuring a plurality of differences between frames includes means for comparing every fourth pixel of a first frame to every fourth pixel of a second frame.
19. The system as defined in claim 17 , wherein the measurement of the differences between frames includes means for computing a summation of an absolute sum of differences between pixels.
20. The system as defined in claim 17 , wherein the statistical analysis includes a computation of a variance.
21. A computer-readable medium encoded with computer executable instructions for detecting a telecine pattern embedded within a sequence of video frames comprising:
a module with instructions configured to receive at least a portion of the frames from the sequence of video frames;
a module with instructions configured to measure a plurality of differences between frames and their adjacent frames in the sequence and storing the measure differences;
a module with instructions configured to arrange the measured differences into a plurality of data groups wherein each data group of the plurality comprises an “in-group” and one or more “out-groups,” wherein each in-group comprises two or more selected measured differences between adjacent frames that are spaced 5 frame intervals apart, wherein each corresponding out-group comprises other measured differences between adjacent frames;
a module with instructions configured to perform a statistical analysis of the in-groups and the out-groups of the plurality of data groups to determine which data group has the in-group that corresponds to duplicate frames of the telecine pattern.
22. The computer-readable medium as defined in claim 21 , wherein the module with instructions configured to measure the plurality of differences between frames includes instructions for computing a summation of an absolute sum of differences between pixels.
23. The computer-readable medium as defined in claim 21 , wherein the statistical analysis includes a computation of a variance.
24. The computer-readable medium as defined in claim 21 , wherein the statistical analysis includes a computation of a standard deviation.