Interlace motion artifact detection using vertical frequency detection and analysis
View Patent ↗An interlace motion artifact detector which identifies video image spatial frequencies characteristic of motion artifacts. The detected frequency is the maximum which can be represented by the vertical sampling rate of the video format (i.e., the Nyquist frequency). This frequency is detected by a pair of partial Discrete Fourier Transforms (DFT) which each calculate only the frequency component of interest. Additional vertical frequency components at one half and one quarter the interlace motion artifact frequency are also detected via a partial DFT. The presence of these lower frequencies acts as an indication of an erroneous motion artifact detection. Additionally, the dynamic range and maximum level of the video data is used as an indication of when to boost the frequency detection levels in areas of low brightness and/or contrast.
1. A method for detecting interlace motion artifacts comprising:
a) detecting a presence of multiple vertical frequencies in an image;
b) analyzing relative levels of the presence of multiple vertical frequencies wherein the relative levels are determined, at least in part, based on a plurality of varying sample set sizes of multiple vertical frequencies; and
c) detecting interlace motion artifacts based on the analyzed relative levels.
2. A method for detecting interlace motion artifacts comprising:
a) detecting a presence of multiple vertical frequencies in an image;
b) analyzing relative levels of the presence of multiple vertical frequencies;
c) detecting interlace motion artifacts based on the analyzed relative levels;
d) determining an overall measure of image intensity and dynamic range; and
e) compensating for the detection of interlace motion artifacts in areas of at least one of low luminosity and contrast.
3. A method for the detection of interlaced motion artifacts comprising:
a) obtaining eight vertically aligned luma data samples;
b) calculating a partial discrete fourier transform for a f max value;
c) calculating a partial discrete fourier transform for a f max /2 value; and
d) calculating a partial discrete fourier transform for a f max /4 value.
4. The method of claim 3 further comprising:
a) obtaining four vertically aligned luma data samples;
b) calculating a second f max value; and
c) passing the f max value, the f max /2 value, the f max /4 value and the second f max value though a filter resulting in a filtered f max value, a filtered f max /2 value, a filtered f max /4 value and a filtered second f max value.
5. The method of claim 4 wherein the filtered values are obtained by:
a) obtaining a first and second previous f max values, a current f max value and a next and second next f max values;
b) doubling the first previous, current and next f max values;
c) summing the doubled first previous, current and next f max values with the second previous and second next f max value; and
d) dividing the sum by 8.
6. A method for the prevention of false detection of interlace motion artifacts comprising:
a) obtaining a plurality of f max frequency detection values;
b) comparing the plurality of f max frequency detection values to a threshold; and
c) adjusting the plurality of f max frequency detection values based upon the comparison wherein the plurality of f max frequency detection values are adjusted, at least in part, by subtracting a variable multiple from the plurality of f max frequency detection values based on a value of f max .
7. A method for the prevention of false detection of interlace motion artifacts comprising:
a) obtaining a composite fmax frequency detection value, a level-boosted fmax/2 frequency detection value and a level-boosted fmax/4 frequency detection values;
b) comparing the composite f max frequency detection value, the level-boosted f max /2 frequency detection value and the level-boosted f max /4 frequency detection value to a threshold; and
c) adjusting the composite f max frequency detection value, the level-boosted f max /2 frequency detection value and the level-boosted f max /4 frequency detection value based upon comparison.
8. The method of claim 7 wherein the composite f max frequency detection value is adjusted by:
a) comparing the composite f max frequency detection value to a first low frequency threshold;
b) multiplying a first low frequency scale factor by the level-boosted f max /2 frequency detection value and subtracting from the composite f max frequency detection value if the composite f max frequency detection value is less than the first low frequency, threshold; and
c) multiplying a second low frequency scale factor by the level-boosted f max /4 frequency detection value and subtracting from the composite f max frequency detection value if the composite f max frequency detection value is greater than the first low frequency threshold.
9. The method of claim 8 wherein the composite f max frequency detection value is adjusted by:
a) comparing the level-boosted f max /4 frequency detection value to a second low frequency threshold;
b) multiplying a third low frequency scale factor by the level-boosted f max /4 frequency detection value and subtracting from the composite f max frequency detection value if the level-boosted f max /4 frequency detection value is less than the second low frequency threshold; and
c) multiplying a fourth low frequency scale factor by the level-boosted f max /4 frequency detection value and subtracting from the composite f max frequency detection value if the level-boosted f max /4 frequency detection value is greater than the second low frequency threshold.
10. The method of claim 9 further comprising setting the composite f max frequency detection value to zero if the composite f max frequency detection value is less than zero.
11. The method of claim 9 wherein the composite f max frequency detection value is lowpass filtered.
12. The method of claim 11 wherein the lowpass filtering is comprises:
a) obtaining a first and second previous f max values, the composite f max frequency detection value and a next and second next f max values;
b) doubling the first previous, and next f max values;
c) octupling the composite f max frequency detection value;
d) summing the doubled first previous f max value, the doubled next f max value, the octupled f max frequency detection value with the second previous and second next f max value; and
e) dividing the sum by 8.