Browse Prior Art Database

ADVANCED METADATA BASED SEARCHABLE DIGITAL VIDEO RECORDER (DVR) / NETWORK VIDEO RECORDER (NVR) SYSTEM

IP.com Disclosure Number: IPCOM000199111D
Publication Date: 2010-Aug-26
Document File: 12 page(s) / 80K

Publishing Venue

The IP.com Prior Art Database

Abstract

A technique for a searchable Digital Video Recorder (DVR) or Network Video Recorder (NVR) system integrated with an advanced metadata based search engine is disclosed. The proposed technique searches metadata ranging from low-level features of scene objects to high-level semantics of scene events that are stored and indexed with video data in DVR / NVR. Content-based video search is achieved in an efficient manner by utilizing the present technique.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 20% of the total text.

Page 1 of 12

RP13492

  ADVANCED METADATA BASED SEARCHABLE DIGITAL VIDEO RECORDER (DVR) / NETWORK VIDEO RECORDER (NVR) SYSTEM

BRIEF ABSTRACT

    A technique for a searchable Digital Video Recorder (DVR) or Network Video Recorder (NVR) system integrated with an advanced metadata based search engine is disclosed. The proposed technique searches metadata ranging from low-level features of scene objects to high-level semantics of scene events that are stored and indexed with video data in DVR / NVR. Content-based video search is achieved in an efficient manner by utilizing the present technique.

KEYWORDS

    Digital Video Recorder (DVR), Network Video Recorder (NVR), Closed Circuit Television (CCTV), Internet Protocol (IP) camera, video analytics (VA), metadata, event, scene, video, encoder, hard disk.

DETAILED DESCRIPTION

    Generally, a digital video recorder (DVR) refers to a device that records video in a digital format to a mass storage device. Network Video Recorder (NVR) is a network camera that enables simultaneous recording and remote access to both live views and recorded images. NVRs are to network video transmitters and cameras what DVRs are to conventional Closed Circuit Television (CCTV) systems.

1

Page 2 of 12

RP13492

    Conventionally, the most common search functions of DVR/NVR products are based on camera identification (ID), time stamps among others. Typically, some DVR/NVR products are provided with search features based on either event/alarm tags generated by video analytics (VA) software or event / text insertions from external devices, such as an access control system or transaction machine. It is efficient to do a video search based on such event / alarm tags. However, such features are not flexible enough for search and offline video analytics as the events or alarms are usually predefined. Both search accuracy and flexibility are confined by the generated events or alarms. For example, consider a line cross detection which is a common function for a number of VA software to detect objects crossing a line. In such a situation it is impossible to search for objects crossing a completely different line with event based search unless the video is reprocessed. However, if we record metadata of an object's trajectory instead of the line cross event, much more search flexibility may be gained. Besides such event based search, there are other advanced video search solutions based on advanced metadata like face descriptor, but its functionality is limited in face similarity searches.

    Further, conventional DVR/NVR usually provides limited search functionalities and search query options. Typically, users have to perform random or sequential browsing and exploration of large amount of video data to find the item of interest (IOI), which are both labor-intensive and time consuming. Additionally, searching in large-scale video surveillance system with hundreds to thousands of video sources is very difficult.

    Therefore as surveillance cameras are incre...