Apparatus and process for detecting scene breaks in a sequence of video frames

Claim 2

A process for detecting scene breaks in a sequence of images, comprising: a) providing a plurality of successive images from an image source; b) converting every image I from the plurality of successive images from an analog to a digital form; c) detecting the location of intensity edges in every image I and storing the information about the location of the intensity edges in a plurality of successive binary images; d) determining a proportion P out of exiting edge pixels in a binary image E from the plurality of successive binary images; e) determining the proportion P in of entering edge pixels in a binary image E 1 from the plurality of successive binary images; f) obtaining a plurality of successive edge change fractions P by obtaining an edge change fraction P=max (P in , P out ) for each binary image E from the plurality of successive binary images; and g) detecting at least one peak in the plurality of successive edge change fractions, the peak being indicative of a scene break.

Claim 3

The process according to claim 1 wherein the peak occurs between two consecutive images, the peak being indicative of a cut.

Claim 4

The process according to claim 1 wherein the peak occurs during more than two consecutive images, the peak being indicative of a fade or a dissolve.

Claim 5

The process according to claim 1 where the images from an image source are video images.

Claim 6

The process according to claim 4 wherein the peak occurs between two consecutive images, the peak being indicative of a cut.

Claim 7

The process according to claim 4 wherein the peak occurs during more than two consecutive images, the peak being indicative of a fade or a dissolve.

Claim 8

A process for detecting scene breaks in a sequence of images, comprising: a) providing a plurality of successive images from an image source; b) converting every image I from the plurality of successive images from an analog to a digital form; c) detecting the location of intensity edges in every image I and storing the information about the location of the intensity edges in a plurality of successive binary images; d) determining a proportion P out of exiting edge pixels in a binary image E from the plurality of successive binary images by determining the fraction of edge pixels in E that are farther than a predetermined distance from the respectively closest pixels in a binary image E 1 ; e) determining the proportion P in of entering edge pixels in the binary image E 1 from the plurality of successive binary images by determining the fraction of edge pixels in E 1 that are farther than a predetermined distance from the respectively closest pixels in E; f) obtaining a plurality of successive edge change fractions by obtaining an edge change fraction P=max (P in , P out ) for each binary image E from the plurality of successive binary images; and g) detecting at least one peak in the plurality of successive edge change fractions, the peak being indicative of a scene break.

Claim 9

A process for detecting scene breaks in a sequence of images, comprising: a) providing a plurality of successive images from an image source; b) converting every image I from the plurality of successive images from an analog to a digital form; c) detecting the location of intensity edges in every image I and storing the information about the location of the intensity edges in a plurality of successive binary images; d) determining a proportion P out of exiting edge pixels in a binary image E from the plurality of successive binary images by determining the fraction of edge pixels in E that are farther than a predetermined distance from the respectively closest pixels in a binary image E 1 ; e) determining the proportion P in of entering edge pixels in the binary image E 1 from the plurality of successive binary images by determining the fraction of edge pixels in E 1 that are farther than a predetermined distance from the respectively closest pixels in E; f) obtaining a plurality of successive edge change fractions by obtaining an edge change fraction P=max (P in , P out ) for each binary image E from the plurality of successive binary images; and g) detecting at least one peak in the plurality of successive edge change fractions, the peak being indicative of a scene break.

Claim 10

The process according to claim 7 wherein the peak occurs between two consecutive images, the peak being indicative of a cut.

Claim 11

The process according to claim 7 wherein the peak occurs during more than two consecutive images, the peak being indicative of a fade or a dissolve.

Claim 12

The process according to claim 7 where the images from an image source are video images.

Claim 13

The process according to claim 10 wherein the peak occurs between two consecutive images, the peak being indicative of a cut.

Claim 14

The process according to claim 10 wherein the peak occurs during more than two consecutive images, the peak being indicative of a fade or a dissolve.

Claim 15

A process for detecting scene breaks in a sequence of images, comprising: a) providing a plurality of successive images from an image source; b) converting every image I from the plurality of successive images from an analog to a digital form; c) compensating for global motion to align consecutive images before obtaining edge change fractions, the compensating for global motion comprising: i) computing a primary motion and at least one secondary motion; and ii) segmenting an image into pixels undergoing primary motion and into pixels undergoing at least one secondary motion; d) detecting the location of intensity edges in every image I and storing the information about the location of the intensity edges in a plurality of successive binary images; e) determining a proportion P out of exiting edge pixels in a binary image E from the plurality of successive binary images by determining the fraction of edge pixels in E that are farther than a predetermined distance from the respectively closest pixels in a binary image E 1 ; f) determining the proportion P in of entering edge pixels in the binary image E 1 from the plurality of successive binary images by determining the fraction of edge pixels in E 1 that are farther than a predetermined distance from the respectively closest pixels in E; g) obtaining a plurality of successive edge change fractions by obtaining an edge change fraction P=max (P in , P out ) for each binary image E from the plurality of successive binary images; and h) detecting at least one peak in the plurality of successive edge change fractions, the peak being indicative of a scene break.

Claim 16

A process for detecting scene breaks in a sequence of images, comprising: a) providing a plurality of successive images from an image source; b) converting every image I from the plurality of successive images from an analog to a digital form; c) compensating for global motion to align consecutive images before obtaining edge change fractions, the compensating for global motion comprising: i) computing a primary motion and at least one secondary motion; and ii) segmenting an image into pixels undergoing primary motion and into pixels undergoing at least one secondary motion; d) detecting the location of intensity edges in every image I and storing the information about the location of the intensity edges in a plurality of successive binary images; e) determining a proportion P out of exiting edge pixels in a binary image E from the plurality of successive binary images by determining the fraction of edge pixels in E that are farther than a predetermined distance from the respectively closest pixels in a binary image E 1 ; f) determining the proportion P in of entering edge pixels in the binary image E 1 from the plurality of successive binary images by determining the fraction of edge pixels in E 1 that are farther than a predetermined distance from the respectively closest pixels in E; g) obtaining a plurality of successive edge change fractions by obtaining an edge change fraction P=max (P in , P out ) for each binary image E from the plurality of successive binary images; and h) detecting at least one peak in the plurality of successive edge change fractions, the peak being indicative of a scene break.

Claim 17

The process according to claim 13 wherein the peak occurs between two consecutive images, the peak being indicative of a cut.

Claim 18

The process according to claim 13 wherein the peak occurs during more than two consecutive images, the peak being indicative of a fade or a dissolve.

Claim 19

The process according to claim 13 where the images from an image source are video images.

Claim 20

The process according to claim 16 wherein the peak occurs between two consecutive images, the peak being indicative of a cut.

Claim 21

The process according to claim 16 wherein the peak occurs during more than two consecutive images, the peak being indicative of a fade or a dissolve.

Claim 24

The system as in claim 19 where the images from an image source are video images.

Claim 25

The system as in claim 19 further comprising means for compensating for global motion to align consecutive images I before obtaining edge change fractions.

Claim 26

The system as in claim 21 where the images from an image source are video images.

Claim 27

The system as in claim 21 further comprising: i) means for computing a primary motion and at least one secondary motion; and ii) means for segmenting an image I into pixels undergoing primary motion and into pixels undergoing at least one secondary motion.

Claim 28

The system as in claim 23 where the images from an image source are video images.