System and method for electronic surveillance

A system for surveillance of a scene comprising:

• at least one camera configured to capture the entire scene under surveillance via a stream of high resolution sequential image frames;
• a processor configured to:
• produce a high resolution image stream from the sequential high resolution image frames;
• produce a low resolution image stream by capturing a reduced number of pixels of each image frame of the high resolution sequential high resolution image frames;
• store the high resolution image stream in short term storage as an independent data stream to the low resolution image stream;
• detect one or more acts, events or objects within the captured scene, by processing a low resolution image stream;
• wherein the processor is further configured to process sequential frames of the low resolution image stream using a detection routine which comprises detecting changes in image data of the sequential frames which are associated with the one or more acts, events or objects within the captured scene, by applying one or more of a number of algorithms to the low resolution image stream, wherein the detected changes are not restricted to changes associated with a motion of an item or an entity and;
• the processor being further configured to, when one or more acts, events or objects are detected, derive location data referencing a sub-area within the sequential frames of the low resolution image stream where the act, event or object is detected, and, generate, using the location data referencing the sub-area within the sequential frames of the low resolution image stream for each of the one or more detected acts, events or objects, frame coordinate data defining a sub-frame portion within a frame of the higher resolution image data where an act, event or object is detected;
• and
• a data management module configured to:
• store the low resolution image stream, where no act, event, or object is detected, in long term memory;
• using the frame coordinate data, extract one or more sub-frame portions from within a frame of the high resolution image stream that correspond to the detected acts, events or objects to produce one or more isolated sub-frame portions, the one or more isolated sub-frame portions extracted from the high resolution image stream being stored independently of the low resolution image stream;
• and
• produce a video stream by retrieving said one or more isolated sub-frame portions extracted from within the frames of the high resolution stream, and said low resolution image stream, from long term memory, and combining said retrieved high resolution sub-frames and said low resolution image stream;
• wherein both the high resolution image stream and the low resolution image stream are produced before said processor detecting of one or more acts, events or objects, so that said detection routine is performed on said low resolution image stream.

An electronic visual surveillance method utilizing sequential image frames formatted in relatively lower and higher resolution image data, wherein the lower resolution image data is produced by capturing a reduced number of pixels from each image frame of the higher resolution image data, the image frames of the higher resolution image data capturing a scene under surveillance, the method comprising:

• storing the high resolution image stream in short term storage as an independent data stream to the low resolution image stream;
• storing the lower resolution image data in long term memory and continuously processing image frames from the lower resolution image data using a detection routine to detect changes in image data of the sequential frames within the low resolution image stream, said detected changes corresponding to one or more detected acts, events or objects within the captured scene, the detected change not being restricted to changes corresponding to a motion of an item or an entity;
• the detection routine comprising detecting the one or more acts, events or objects by applying one or more of a number of motion detection and non-motion detection algorithms to detect one or more of a plurality of different types of acts, events or objects;
• deriving, for each detected act, event or object, location data referencing a sub-area within the sequential frames of the low resolution image stream where the act, event or object is detected;
• generating, using the location data referencing the sub-area within the sequential frames of the low resolution image stream for each of the one or more detected acts, events or objects, frame coordinate data defining a sub-frame area within a frame of the higher resolution image data where an act, event or object is detected;
• extracting the sub-frame area of the higher resolution image data using the frame coordinate data to provide an isolated sub-frame area;
• storing the isolated sub-frame area extracted from the higher resolution image data independently of the lower resolution image data in long term memory;
• and
• producing a video stream by retrieving said isolated sub-frame area extracted from within the high resolution image data, and said lower resolution image data, from long term memory, and combining said retrieved high resolution sub-frame area and said lower resolution image data;
• wherein both the high resolution image stream and the low resolution image stream are produced before said detecting of one or more acts, events or objects, so that said detection routine is performed on said low resolution image stream.

An electronic surveillance system comprising:

• a camera configured to output a stream of sequential image frames capturing a scene under surveillance;
• a processor configured to generate a higher resolution image data from the stream of sequential image frames, and continually process lower resolution image data generated by capturing a reduced number of pixels from each image frame of the sequential image frames, to detect changes in said image data that correspond to one or more acts, events, or objects within the captured scene, the detected changes not being restricted to change corresponding to a motion of an item or an entity, using a detection routine to detect one or more acts, events or objects in the scene captured in the sequential image frames,
• wherein the higher resolution image data is stored in short term storage as an independent data stream to the lower resolution image data;
• wherein the detection routine comprises:
• detecting the acts, events or objects by applying one or more of a number of algorithms to detect one or more of a plurality of different types of behaviour;
• for each of one or more detected act, event or object, the processor being arranged to derive location data referencing a sub-area within the sequential frames of the low resolution image stream where the act, event or object is detected, and generate, using the location data referencing the sub-area within the sequential frames of the low resolution image stream for each of the one or more detected acts, events or objects, frame coordinate data defining a sub-frame area within higher resolution image data produced from the sequential image frames where an act, event or object has been detected;
• a data management module configured to:
• store the longer resolution image data, where no act, event, or object is detected, stored in long term memory;
• extract the sub-frame area of the higher resolution image data using the frame coordinate data to produce an isolated sub-frame area, and store the isolated sub-frame area extracted from the high resolution image data independently of the lower resolution image data;
• and
• an identification engine configured to identify attributes of the act, event or object from the isolated sub-frame area extracted from the relatively higher resolution image data;
• wherein both the high resolution image data and the lower resolution image data are produced before said detecting of one or more acts, events or objects, so that said detection routine is performed on said low resolution image stream;
• wherein a video stream is produced, by retrieving said isolated sub-frame area extracted from the higher resolution image data, and said lower resolution image data, from long term memory, and combining said retrieved high resolution sub-frame area and said lower resolution image stream.

A method for surveillance of a scene comprising the steps of:

• capturing a scene via a stream of sequential image frames;
• producing a high resolution image stream from the sequential image frames, and producing a low resolution image stream of the entire captured scene, each image frame of the low resolution image stream having a reduced number of pixels of a corresponding image frame of the sequential image frames, the high resolution image stream being stored in short term storage as an independent data stream to the low resolution image stream, the high and low resolution image streams comprising image data for a concurrent time period;
• storing the low resolution image stream in a long term memory;
• detecting changes in image data of sequential frames of the low resolution image stream, the detected changes corresponding to one or more detected act, event, or object, the detected changes not being restricted to changes associated with a motion of an item or an entity, by processing the sequential frames of the low resolution image stream using a detection routine which comprises applying one or more of a number of algorithms to detect one or more of a plurality of different types of acts, events, or objects;
• for each detected act, event or object, deriving location data referencing a sub-area within the sequential frames of the low resolution image stream where the act, event or object in the captured scene is detected;
• generating, using the location data referencing the sub-area within the sequential frames of the low resolution image stream for each of the one or more detected acts, events or objects, frame coordinate data defining a sub-frame portion within a frame of the higher resolution image data where an act, event or object is detected, and
• using the frame coordinate data, extracting the one or more sub-frame portions from within the frame of the high resolution image stream corresponding to the one or more detected acts, events or objects to produce one or more isolated sub-frame portions;
• storing the one or more isolated sub-frame portions extracted from within the frame of the high resolution image stream independently of the low resolution image stream in long term memory;
• and
• producing a video stream by retrieving said one or more isolated sub-frame portions extracted from within the frames of the high resolution, and said low resolution image stream, from long term memory, and combining said retrieved high resolution sub-frames and said low resolution image stream;
• wherein the one or more sub-frame portions are extracted based upon a signal effected by the detection of the one or more acts, events or objects within the low resolution image stream;
• wherein both the high resolution image stream and the low resolution image stream are produced before said detecting of one or more acts, events or objects, so that said detection routine is performed on said low resolution image stream.

A system for surveillance of a scene comprising:

• at least one camera configured to capture a scene via a stream of sequential image frames;
• a processor configured to generate a high resolution image stream from the captured sequential image frames, and to generate a low resolution image stream by capturing a reduced number of pixels of each image frame of the captured sequential image frames, the high resolution and low resolution image streams comprising image data of the entire captured stream, the image data covering a concurrent time period;
• the processor being further configured to detect one or more changes in image data of the sequential frames of the low resolution image stream, the detected change or changes corresponding to one or more detected acts, events, or objects, the detected changes not being restricted to changes associated with a motion of an item or an entity, by processing the sequential frames of the low resolution image stream using a detection routine which comprises detecting the one or more acts, events or objects by applying one or more algorithms to detect one or more of a plurality of different types of acts, events or objects;
• and
• when one or more acts, events or objects are detected within the captured scene, derive location data referencing a sub-area within the sequential frames of the low resolution image stream where the act, event or object is detected, and, generate, using the location data referencing the sub-area within the sequential frames of the low resolution image stream for each of the one or more detected acts, events or objects, frame coordinate data defining a sub-frame portion within a frame of the higher resolution image data where an act, event or object is detected, and
• a data management module configured to
• store the low resolution image stream in long term memory and extract one or more sub-frame portions using the frame coordinate data from within a frame of the high resolution image stream that correspond to detected acts, events or objects to produce one or more isolated sub-frame portions, the one or more isolated sub-frame portions extracted from the high resolution image stream being stored in long term memory independently of the low resolution image stream;
• and
• produce a video stream by retrieving said one or more isolated sub-frame portions extracted from within the frames of the high resolution, and said low resolution image stream, from long term memory, and combining said retrieved high resolution sub-frames and said low resolution image stream;
• wherein the one or more sub-frame portions are extracted based upon a signal effected by the detection of the one or more acts, events or objects within the low resolution image stream;
• wherein both the high resolution image stream and the low resolution image stream are produced before said detecting of one or more acts, events or objects, so that said detection routine is performed on said low resolution image stream.