New patent applications are published on Thursdays by the USPTO. This week's Spotlight Applications address aspects of digital fingerprinting and digital watermarking. Assigned to Sorenson Media, the first application discloses techniques for detecting channel change in automatic content recognition fingerprint matching. Assigned to VMware, the second application discloses watermarking and scalability techniques for a virtual desktop planning tool.
20170013324, "Detecting channel change in automatic content recognition fingerprint matching," assigned to Sorenson Media, Inc.
The disclosed system and method receives, from a media device, a query fingerprint that includes an ordered sequence of frames of content being consumed, and matches the query fingerprint with frame fingerprints by determining a distance between respective frame fingerprints and corresponding ones of the ordered sequence of frames. The matching may be performed by populating a binary tree structure with respective distances, chronologically across leaf nodes of the binary tree structure, and then noting a value of the root node. If below a maximum threshold, then there is a match. The system and method may detect a channel change at the media device by traversing a right or a left sub-tree of the binary tree structure, and locating a low value (or a zero) at the first or second nodes of the sub-tree. This channel change may be useful in targeting media to content being streamed to the media device.
20170011486, "Watermarking and scalability techniques for a virtual desktop planning tool," assigned to VMware.
A method for measuring performance of virtual desktop services offered by a server including a processor is described. A first encoded watermark is embedded into user interface display generated by a virtual desktop when initiating an operation. The first encoded watermark includes pixels identifying the operation and indicating its initiation. A second encoded watermark is embedded into the user interface upon completion of the operation indicating completion of the operation. An action performance time is then computed and stored in a memory. Multiple performance times may be compiled from multiple operations of multiple virtual desktops to assess the performance of the system as a whole.