Method and apparatus for detecting scene-cuts in a block-based video coding system

Claim 2

In a block-based video encoder that partitions each input image of a sequence of input images into a plurality of macroblocks, computes at least one motion vector for each of said macroblocks, and generates a sequence of predicted images, where each predicted image in said sequence comprises a plurality of predicted macroblocks derived from said input images and said motion vectors, a method of detecting that a first image is substantially different from a second image within said sequence of input images comprising the steps of: (a) estimating a coding mode that is to be used by the block-based video encoder to code a macroblock within said second image; (b) incrementing a count value whenever the coding mode is estimated to be a particular type of coding mode; (c) repeating steps a and b until the coding mode is estimated for each macroblock in the second image; and (d) when the count value exceeds a threshold level, setting an indicator flag.

Claim 3

The method of claim 1 wherein said estimating step further comprises the steps of: comparing a predicted macroblock to an input macroblock from said first image to generate a residuals macroblock; computing a variance of the input macroblock and the residuals macroblock; comparing the variance of the input macroblock and the variance of the residuals macroblock to a decision function; and in response to a particular outcome of said comparing step, incrementing said count value.

Claim 4

The method of claim 1 wherein said particular type of coding mode is an intra-block coding mode.

Claim 5

The method of claim 1 wherein the count value exceeds the threshold level whenever a scene cut has occurred in said sequence of images.

Claim 6

The method of claim 1 wherein said threshold level is set to thirty percent of the total number of macroblocks in the input image.

Claim 7

In a block-based video encoder that partitions each input image of a sequence of input images into a plurality of macroblocks, computes at least one motion vector for each of said macroblocks, and generates a sequence of predicted images, where each predicted image in said sequence comprises a plurality of predicted macroblocks derived from said input images and said motion vectors, a method of detecting that a first image is substantially different from a second image within said sequence of input images comprising the steps of: (a) comparing a predicted macroblock to an input macroblock from said first image to generate a residuals macroblock; (b) computing a variance of the input macroblock and the residuals macroblock; (c) comparing the variance of the input macroblock to the variance of the residuals macroblock; (d) if the variance of the residuals macroblock exceeds the variance of the input macroblock, incrementing a count value; (e) repeating steps a, b, c, d, and e, until each macroblock in the predicted image is compared to each input macroblock; and (f) when the count value exceeds a threshold level, setting an indicator flag.

Claim 8

In a block-based video encoder that partitions each input image of a sequence of input images into a plurality of macroblocks, computes at least one motion vector for each of said macroblocks, and generates a sequence of predicted images, where each predicted image in said sequence comprises a plurality of predicted macroblocks derived from said input images and said motion vectors, a method of detecting that a first image is substantially different from a second image within said sequence of input images comprising the steps of: (a) comparing a predicted macroblock to an input macroblock from said first image to generate a residuals macroblock; (b) computing a variance of the input macroblock and the residuals macroblock; (c) comparing the variance of the input macroblock to the variance of the residuals macroblock; (d) if the variance of the residuals macroblock exceeds the variance of the input macroblock, incrementing a count value; (e) repeating steps a, b, c, d, and e, until each macroblock in the predicted image is compared to each input macroblock; and (f) when the count value exceeds a threshold level, setting an indicator flag.

Claim 9

The method of claim 6 wherein said variance of the residuals macroblock exceeds the variance of the input macroblock to produce a count value that exceeds the threshold level whenever a scene cut has occurred in said sequence of images.

Claim 10

The method of claim 6 wherein said threshold level is set to approximately thirty-three percent of the total number of macroblocks in the input image.

Claim 11

In a block-based video encoder that partitions each input image of a sequence of input images into a plurality of macroblocks, computes at least one motion vector for each of said macroblocks, and generates a sequence of predicted images, where each predicted image in said sequence comprises a plurality of predicted macroblocks derived from said input images and said motion vectors, apparatus of detecting that a first image is substantially different from a second image within said sequence of input images comprising: a coding mode estimator, connected to a motion estimator within said block-based video encoder and a source of input images, for estimating the coding mode that is to be used by said block-based video encoder to code each macro-block of said second image; a counter, connected to said coding mode estimator, for incrementing a count value whenever a number of macroblocks coded using a first coding mode and a number of macroblocks coded using a second coding mode compare favorably with a decision function; and thresholding means, connected to said counter, for setting an indicator flag when the count value exceeds a threshold level.

Claim 12

In a block-based video encoder that partitions each input image of a sequence of input images into a plurality of macroblocks, computes at least one motion vector for each of said macroblocks, and generates a sequence of predicted images, where each predicted image in said sequence comprises a plurality of predicted macroblocks derived from said input images and said motion vectors, apparatus of detecting that a first image is substantially different from a second image within said sequence of input images comprising: a coding mode estimator, connected to a motion estimator within said block-based video encoder and a source of input images, for estimating the coding mode that is to be used by said block-based video encoder to code each macro-block of said second image; a counter, connected to said coding mode estimator, for incrementing a count value whenever a number of macroblocks coded using a first coding mode and a number of macroblocks coded using a second coding mode compare favorably with a decision function; and thresholding means, connected to said counter, for setting an indicator flag when the count value exceeds a threshold level.

Claim 13

The apparatus of claim 9 wherein said coding mode estimator comprises: subtractor, connected to said motion estimator and a source of said input image, for comparing each of the predicted macroblocks to an input macroblock to generate a residuals macroblock; a residuals variance generator, connected to said subtractor, for computing a variance of the residuals macroblocks; a input variance generator, connected to said source of the input image, for computing a variance of the input macroblocks; a decision circuit, connected to the residuals variance generator and the input variance generator, for comparing the variance of the input macroblocks and the variance of the residuals macroblocks to a decision function, wherein when the variance of said residuals macroblock and the variance of the input macroblock compares with the decision function to produce a first output, the block-based video encoder is estimated to be using the second coding mode and when the variance of said residuals macroblock and the variance of the input macroblock compares with the decision function to produce a second output, the block-based video encoder is estimated to be using the first coding mode.

Claim 14

The apparatus of claim 10 wherein said first coding mode is an inter-block coding mode and the second coding mode is an intra-block coding mode.

Claim 15

The apparatus of claim 9 wherein the count value exceeds the threshold level whenever a scene cut has occurred in said sequence of images.

Claim 16

The apparatus of claim 9 wherein said threshold level is set to approximately thirty-three percent of the total number of macroblocks in the input image.

Claim 17

The apparatus of claim 10 wherein said decision circuit is a comparator, connected to the residuals variance generator and the input variance generator, for comparing the variance of the input macroblocks to the variance of the residuals macroblocks, wherein when the variance of said residuals macroblock exceeds the variance of the input macroblock, the block-based video encoder is estimated to be using the second coding mode and when the variance of said input macroblock exceeds the variance of the residuals macroblock, the block-based video encoder is estimated to be using the first coding mode.