Detection of long shots in sports video
A method of identifying long shots of sports video comprising receiving a video frame comprising a plurality of pixels, classifying each of the plurality of pixels as a candidate field pixel or a candidate non-field pixel, determining whether at least a predefined percentage of the plurality of pixels are candidate field pixels, calculating a first standard deviation, the first standard deviation being the standard deviation of the hues of all candidate field pixels, and classifying the video frame as a long shot of sports video when at least the predefined percentage of the plurality of pixels are candidate field pixels and the first standard deviation is equal to or lower than a predefined maximum standard deviation value.
1. A method of identifying long shots of sports video comprising:
receiving a video frame comprising a plurality of pixels;
classifying each of said plurality of pixels as a candidate field pixel or a candidate non-field pixel;
determining whether at least a predefined percentage of said plurality of pixels are candidate field pixels;
calculating a first standard deviation, the first standard deviation being the standard deviation of the hues of all candidate field pixels; and
classifying said video frame as a long shot of sports video when at least said predefined percentage of said plurality of pixels are candidate field pixels and said first standard deviation is equal to or lower than a predefined maximum standard deviation value.
2. The method of claim 1 , further comprising:
calculating a field mean hue, said field mean hue being the average hue of all said candidate field pixels;
determining whether said field mean hue is within a predefined range;
calculating a second standard deviation, the second standard deviation being the standard deviation of the hues of the subset of candidate field pixels having hue components within said predefined range; and
classifying said video frame as a long shot of sports video when at least said predefined percentage of said plurality of pixels are candidate field pixels and at least one of said first standard deviation and said second standard deviation are equal to or lower than said predefined maximum standard deviation value.
3. The method of claim 2 , wherein the steps of calculating said first standard deviation, said field mean hue and said second standard deviation comprises generating a hue histogram of said video frame and calculating said first standard deviation, said field mean hue and said second standard deviation from said hue histogram.
4. The method of claim 1 , wherein the classifying of each of said plurality of pixels as a candidate field pixel or a candidate non-field pixel comprises determining a hue component threshold, classifying each of said plurality of pixels that have hue components less than said hue component threshold as candidate field pixels; and classifying each of said plurality of pixels that have hue components greater than said hue component threshold as candidate non-field pixels.
5. The method of claim 2 , wherein said hue component threshold is determined using Otsu's method.
6. The method of claim 4 , wherein the classifying of each of said plurality of pixels as a candidate field pixel or a candidate non-field pixel further comprises determining a value component threshold, classifying each of said plurality of pixels that have value components greater than said value component threshold as candidate field pixels; and classifying each of said plurality of pixels that have value components greater than said value component threshold as candidate non-field pixels.
7. The method of claim 6 , wherein said hue component threshold and said value component threshold is determined using Otsu's method.
8. The method of claim 1 , wherein said predefined percentage is 50%.
9. The method of claim 8 , wherein determining whether at least 50% of said plurality of pixels are candidate field pixels comprises determining whether a total mean hue is less than a hue component threshold, wherein said total mean hue is the mean hue of all of said plurality of pixels.
10. The method of claim 1 , wherein said predefined range is a range of hue values typical for the playing surface of a sporting event.
11. The method of claim 1 , further comprising informing an encoder that said video frame is not a long shot of sports video when more than said predefined percentage of said plurality of pixels are determined to be candidate non-field pixels.
12. The method of claim 1 , further comprising informing an encoder that said video frame is not a long shot of sports video when the mean hue of all candidate field pixels is determined to be outside of said predefined range.
13. The method of claim 1 , further comprising informing an encoder that said video frame is not a long shot of sports video when both of said first standard deviation and said second standard deviation are higher than said predefined maximum standard deviation value.
14. The method of claim 1 , further comprising informing an encoder that said video frame is a long shot of sports video when at least one of said first standard deviation and said second standard deviation are equal to or lower than a predefined maximum standard deviation value.
15. The method of claim 1 , further comprising calculating a third standard deviation, said third standard deviation being the standard deviation of the value components of candidate field pixels; comparing said third standard deviation to a predefined maximum; and determining that said frame is not a long shot when said third standard deviation is less than said predefined maximum.
16. A system, comprising:
a memory configured to receive and store a video frame comprising a plurality of pixels; and
a processor configured to:
classify each of said plurality of pixels as a candidate field pixel or a candidate non-field pixel;
determine whether at least a predefined percentage of said plurality of pixels are candidate field pixels;
calculate a first standard deviation, the first standard deviation being the standard deviation of the hues of all candidate field pixels;
classify said video frame as a long shot of sports video when at least said predefined percentage of said plurality of pixels are candidate field pixels and said first standard deviation is equal to or lower than a predefined maximum standard deviation value.
17. The system of claim 16 , wherein said processor is further configured to:
calculate a field mean hue, said field mean hue being the average hue of all said candidate field pixels;
determine whether said field mean hue is within a predefined range;
calculate a second standard deviation, the second standard deviation being the standard deviation of the hues of the subset of candidate field pixels having hue components within said predefined range; and
classify said video frame as a long shot of sports video when at least said predefined percentage of said plurality of pixels are candidate field pixels and at least one of said first standard deviation and said second standard deviation are equal to or lower than said predefined maximum standard deviation value.