Object archival systems and methods

A computer system for processing video comprising:

• an encoder including a processor configured to preprocess initial video data to detect one or more significant objects represented in the initial video data by:
• identifying reoccurring salient components in the initial video data;
• qualifying the significance of the reoccurring salient components;
• and
• selecting one or more of the most significant of the qualified salient components, the selected one or more most significant components resulting in the detected one or more significant objects;
• segmenting the detected one or more significant objects form the non-object components in the initial video data, the segmentation resulting in object and non-object portions of the initial video data that are processed differently, such that the non-objects and the detected one or more significant objects are encoded separately using distinct encoding processes;
• an object modeler configured to create one or more object models for the detected one or more significant objects to provide a compact representation of the detected one or more significant objects, compare one or more of the object models to determine whether there are substantially equivalent object models, and eliminate the redundant substantially equivalent object models;
• an object library archiving the object models;
• and
• a decoder accessing a codec and the object library to use one or more of the archived object models when reconstructing salient objects of an encoded video file, the decoder reconstructing non-salient objects without accessing the object library, the encoded video file is an encoding of other video data, the other video data being distinct from the initial video data used to created the archived object models.

A computer system for processing a video file comprising:

• one or more processors configured to preprocess initial video data to detect one or more significant objects represented in the initial video data by:
• identifying reoccurring salient components in the initial video data;
• qualifying the significance of the reoccurring salient components;
• and
• selecting one or more of the most significant of the qualified salient components, the selected one or more most significant components resulting in the detected one or more significant objects;
• segmenting the detected one or more significant objects form the non-object components in the initial video data, the segmentation resulting in object and non-object portions of the initial video data that are processed differently, such that the non-objects and the detected one or more significant objects are encoded separately using distinct encoding processes;
• an object modeler configured to create one or more object models for the detected one or more significant objects to provide a compact representation of the detected one or more significant objects, compare one or more of the object models to determine whether there are substantially equivalent object models, and eliminate the redundant substantially equivalent object models;
• an object model archive configuring to archive the object models, the object models being configured for use when processing at least one other video file including other video data, the other video data being distinct from the initial video data used to create the archived object models;
• coding salient objects of the at least one other video file using a transform encoding process with reference to the archived object models, and coding non-object portions of the at least one other video file using a transform encoding process without reference to the object model archive.

A method of processing a video file, the method comprising the computer implemented steps of:

• processing a plurality of initial video files to detect one or more significant objects represented in the video files data by:
• identifying reoccurring salient components in the video files;
• qualifying the significance of the reoccurring salient components;
• and
• selecting one or more of the most significant of the qualified salient components, the selected one or more most significant components resulting in the detected one or more significant objects;
• segmenting the detected one or more significant objects form the non-object components in the initial video data, the segmentation resulting in object and non-object portions of the initial video data that are processed differently, such that the non-objects and the detected one or more significant objects are encoded separately using distinct encoding processes;
• creating one or more object models for the detected one or more significant objects to provide a compact representation of the detected one or more significant objects;
• comparing one or more of the archived object models to determine whether there are substantially equivalent object models stored in the archive;
• reducing the size of the archive by eliminating the redundant substantially equivalent object models;
• and
• using one or more of the object models when coding the salient objects of a subject video file while not using the object models when coding the non-objects of the subject video, the subject video is not one of the plurality of initial video files from which the object models are derived.

A codec for processing a video file, the codec comprising computer readable instructions stored on a non-transitory computer readable medium so as when executed by one or more computer processors, the computer readable instructions causing processing of the video file by:

• reconstructing salient objects of the video file using archived object models while reconstructing non-object portions of the video file without using the archived object models, in which at least a portion of the object models are different than the salient objects depicted in the video files,
• where the archived object models are created by:
• identifying reoccurring salient components in the initial video data;
• qualifying the significance of the reoccurring salient components;
• and
• selecting one or more of the most significant of the qualified salient components, the selected one or more most significant components resulting in the detected one or more significant objects;
• segmenting the detected one or more significant objects form the non-object components in the initial video data, the segmentation resulting in object and non-object portions of the initial video data that are processed differently, such that the non-objects and the detected one or more significant objects are encoded separately using distinct encoding processes;
• creating one or more object models for the detected one or more significant objects to provide a compact representation of the detected one or more significant objects;
• archiving the object models;
• comparing one or more of the archived object models to determine whether there are substantially equivalent object models stored in the archive;
• and
• reducing the size of the archive by eliminating the redundant substantially equivalent object models.