Method and apparatus for analyzing image data
View Patent ↗A method for the automatic detection of fields or frames in film or video content having similar or substantially the same luminance or color component values, for example to detect black frames. Image sample values are assigned to a sample value range having a signed weight associated therewith, and a contribution to a discriminator function is determined for each sample range depending on the signed weight and the number of input samples in the sample value range. The discriminator function output is then compared with a threshold.
1. A method of analyzing sampled or sub-sampled image data to determine whether image data sample values are close to a black level, the method comprising the steps:
defining a plurality of sample value ranges, wherein each of the sample value ranges has a signed weight assigned thereto and a sample value range including the black level has a weight of opposite sign from the weight assigned to at least one other sample value range;
counting, by a processor, the number of samples having values within each said range and forming a weighted sum of the count values; and
comparing, by the processor, the weighted sum with a threshold value to determine whether image data sample values are close to the black level.
2. The method according to claim 1 , wherein the weight assigned to the sample value range nearest to, but not including, the black level has a weight of the same sign as the sample value range including the black level.
3. The method according to claim 1 , wherein the weights having opposing sign to the weight associated with the sample value range including the black increase in magnitude with closer proximity of the associated sample value range to the black level.
4. The method according to claim 1 , wherein the threshold value is zero.
5. The method as claimed in claim 1 , wherein the weight assigned to the sample value range containing the black level is negative.
6. The method as claimed in claim 1 , wherein the weighted sum is calculated from a representative set of samples from an image.
7. The method as claimed in claim 6 , where the representative set of samples is pixels of a television picture field.
8. The method as claimed in claim 7 , wherein successive odd- or even-television picture fields are analyzed.
9. The method according to claim 1 , wherein image data from a sequence of images is analyzed and a positive decision is output only when a positive result is determined for each of a defined number of consecutive images in the said sequence.
10. The method as claimed in claim 1 , where one or more of: the number of sample value ranges; the extent of the sample value ranges; the magnitude of the weight to be assigned to the sample value ranges; and the sign of the weight to be assigned to the sample value ranges are obtained from optimisation of a training sequence.
11. A computer-readable medium storing code adapted to implement a method in accordance with claim 1 .
12. A method of analyzing sampled or sub-sampled image data to determine whether image data sample values are close to a defined value, the method comprising the steps of:
forming, by a processor, a weighted sum from relative contributions from a plurality of image data sample value ranges, the relative contribution depending on the number of samples in the range and a signed weight associated with each sample value range, wherein a sample value range including the defined value therein has a negative weight and at least one other sample value range has a positive weight; and
determining, by the processor, a positive result when the weighted sum is less than the threshold value.
13. The method according to claim 12 , wherein the said image data is luminance data.
14. The method according to claim 12 , wherein the said image data is primary color component data.
15. The method according to claim 12 , wherein the said defined value is black level.
16. The method according to claim 12 , wherein the said defined value is white level.
17. The method according to claim 12 , wherein the said image data is color difference data.
18. The method according to claim 17 , wherein the defined value represents zero color difference.
19. The method according to claim 12 , wherein the positive weights increase in magnitude with closer proximity of the associated sample value range to the said defined value.
20. The method according to claim 12 , wherein the threshold value is zero.
21. The method according to claim 12 , wherein image data from a sequence of images is analyzed and a positive decision is output only when a positive result is determined for each of a defined number of consecutive images in the said sequence.
22. A computer-readable medium storing code adapted to implement a method in accordance with claim 12 .
23. A method of analyzing sampled or sub-sampled image data to determine whether image data sample values are close to a defined value comprising the steps, for each image data sample, of:
allocating the image data sample to one of at least three sample value ranges;
accumulating, by a processor, a signed weight associated with the allocated sample value range to form a weighted sum, the weight associated with a sample value range including the defined value therein being of opposite sign from the weight assigned to at least one other sample value range; and
determining, by the processor, a positive result when the weighted sum for all sample values is less or more than the threshold value, if the weight assigned to the sample value range including the defined value therein is negative or positive, respectively.
24. The method according to claim 23 , wherein said defined value is black level.
25. The method according to claim 23 , wherein the weight assigned to the sample value range nearest to, but not including, the defined value has a weight of the same sign as the sample value range including the defined value.
26. The method according to claim 23 , wherein the weights having opposing sign to the weight associated with the sample value range including the defined value increase in magnitude with closer proximity of the associated sample value range to the said defined value.
27. The method as claimed in claim 23 , wherein the weight assigned to the sample value range containing the defined value is negative.
28. The method according to claim 23 , wherein image data from a sequence of images is analyzed and a positive decision is output only when a positive result is determined for each of a defined number of consecutive images in the said sequence.
29. A computer-readable medium storing code adapted to implement a method in accordance with claim 23 .