Image metadata creation for improved image processing and content delivery
First Claim
1. A media processor for the creation of metadata from a set of images, said media processor comprising:
- an input for a first set of data, said first set of data comprising image data, metadata or image data/metadata associated with said set of images, said set of images comprising a sequence of images; and
a metadata processor being capable of receiving said first set of data and computing a set of metadata from said first set of data, said set of metadata comprising statistical data derived from said sequence of images over a plurality of different temporal intervals and said set of metadata representing a media fingerprint which identifies said set of images without altering said set of images; and
whereinsaid metadata processor is capable of generating and storing a multi-scale version of said set of metadata, said multi-scale version including first metadata generated based on statistical attributes associated with said sequence of images over a first temporal interval and second metadata generated based on statistical attributes associated with said sequence of images over a second temporal interval;
said first temporal interval and said second temporal interval are associated with increasing levels of said multi-scale version of metadata; and
each increasing level of said multi-scale version of metadata is associated with an increasing confidence level in said media fingerprint;
said set of metadata further comprises at least one of statistical attributes derived directly from said first set of data and statistical attributes derived from image appearance correlates of said first set of data;
said media processor is further capable of performing display management (DM) processing of a set of image data to be rendered on a target display having a set of display specifications; and
to perform said DM processing, said media processor is operative to compute a first set of statistics from said set of image data over time,compare a portion of said first set of statistics against said set of display specifications,test for a match condition for said portion of said first set of statistics against said set of display specifications, andrender a portion of said set of image data upon said target display, said target display comprising said set of display specifications.
1 Assignment
0 Petitions
Accused Products
Abstract
Several embodiments of a media processor for the creation of metadata from a set of images are given. The media processor receives a set of image data and computes metadata from the image data. The metadata may comprise statistical data based on the image data; from which the media processor may generate and store a multi-scale version of the metadata. The metadata may comprise statistical attributes derived directly from the image data or statistical attributes derived from image appearance correlates of the image data. The media processor may generate a subsampled set of the metadata and the subsampled set may be stored in an efficient data structure, such as a pyramidal structure. The pyramidal structure may be generated from spatial image information, from temporal image information or from spatial-temporal image information.
68 Citations
20 Claims
-
1. A media processor for the creation of metadata from a set of images, said media processor comprising:
-
an input for a first set of data, said first set of data comprising image data, metadata or image data/metadata associated with said set of images, said set of images comprising a sequence of images; and a metadata processor being capable of receiving said first set of data and computing a set of metadata from said first set of data, said set of metadata comprising statistical data derived from said sequence of images over a plurality of different temporal intervals and said set of metadata representing a media fingerprint which identifies said set of images without altering said set of images; and
whereinsaid metadata processor is capable of generating and storing a multi-scale version of said set of metadata, said multi-scale version including first metadata generated based on statistical attributes associated with said sequence of images over a first temporal interval and second metadata generated based on statistical attributes associated with said sequence of images over a second temporal interval; said first temporal interval and said second temporal interval are associated with increasing levels of said multi-scale version of metadata; and each increasing level of said multi-scale version of metadata is associated with an increasing confidence level in said media fingerprint; said set of metadata further comprises at least one of statistical attributes derived directly from said first set of data and statistical attributes derived from image appearance correlates of said first set of data; said media processor is further capable of performing display management (DM) processing of a set of image data to be rendered on a target display having a set of display specifications; and to perform said DM processing, said media processor is operative to compute a first set of statistics from said set of image data over time, compare a portion of said first set of statistics against said set of display specifications, test for a match condition for said portion of said first set of statistics against said set of display specifications, and render a portion of said set of image data upon said target display, said target display comprising said set of display specifications. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
-
-
19. A media processor for the creation of metadata from a set of images, said media processor comprising:
-
an input for source data associated with said set of images, said set of images comprising a sequence of images; and a metadata processor operative to receive said source data, generate first metadata based on statistics of said sequence of images associated with a first temporal interval, generate second metadata based on statistics of said sequence of images associated with a second temporal interval, said second temporal interval being different than said first temporal interval such that said first metadata and said second metadata are derived based on different amounts of said source data, and generate a multi-scale version of metadata based on said first metadata and said second metadata, said multi-scale version of said metadata representing a media fingerprint which identifies said set of images without altering said set of images; and
whereinsaid first temporal interval and said second temporal interval are associated with increasing levels of said multi-scale version of metadata; and each increasing level of said multi-scale version of metadata is associated with an increasing confidence level in said media fingerprint. - View Dependent Claims (20)
-
Specification