System and method for monitoring a premises based on parsed codec data

An intrusion detection system for detecting conditions at a physical premises, the system comprising:

• one or more camera sensors deployed at the physical premises;
• and
• a central monitoring station, the central monitoring station comprising a data processing system configured to perform operations comprising:
• receiving, from a video sensor system deployed at the physical premises, compressed video data that is encoded by a block-based encoder in the video sensor system;
• parsing the compressed video data to extract from the compressed video data, one or more macroblock arrays that correspond to areas of a frame of the compressed video data, the one or more macroblock arrays comprising motion vector data and color vector data;
• applying a cluster analysis to the one or more macroblock arrays to reduce the one or more macroblock arrays to cluster data;
• aggregating the cluster data over a series of frames;
• identifying one or more first segments of the cluster data associated with first changes in the motion vector data and the color vector data less than a predefined amount and one or more second segments of the cluster data associated with second changes in the motion vector data and the color vector data greater than the predefined amount;
• truncating and concatenating the cluster data to remove the one or more first segments and retain the one or more second segments;
• applying a pattern recognition algorithm to the cluster data to detect one or more irregularities in the cluster data;
• determining whether the one or more irregularities correspond to an occurrence of one or more events at the physical premises by:
• identifying a time that the motion vector data indicates the one or more events have occurred;
• and
• analyzing the color vector data at the time to determine that the color vector data confirms that the one or more events have occurred at the time;
• and
• applying the one or more events to a rules engine to determine a response to the one or more events.

One or more machine readable hardware storage devices storing instructions that are executed by one or more processing devices of a data processing system to perform operations comprising:

• receiving, from a video sensor system deployed at a physical premises, block-based encoded video data;
• parsing the block-based encoded video data to extract one or more macroblock arrays that correspond to areas of a frame of video data, the one or more macroblock arrays comprising motion vector data and color vector data;
• applying a cluster analysis to reduce the one or more macroblock arrays to cluster data;
• aggregating the cluster data over a series of frames;
• identifying one or more first segments of the cluster data associated with first changes in the motion vector data and the color vector data less than a predefined amount and one or more second segments of the cluster data associated with second changes in the motion vector data and the color vector data greater than the predefined amount;
• truncating and concatenating the cluster data to remove the one or more first segments and retain the one or more second segments;
• applying a pattern recognition algorithm to the cluster data to detect one or more irregularities in the cluster data;
• determining whether the one or more irregularities correspond to an occurrence of one or more events at the physical premises by:
• identifying a time that the motion vector data indicates the one or more events have occurred;
• and
• analyzing the color vector data at the time to determine that the color vector data confirms that the one or more events have occurred at the time;
• and when the one or more events are determined,
• applying the one or more events to a rules engine to determine a response to the one or more events.