US20190332848A1 - Facial enrollment and recognition system - Google Patents

Abstract

A facial recognition system includes a memory for storing a facial image database, wherein the facial image database includes a plurality of entries each corresponding to a different person, and wherein each entry includes a person identifier along with one or more facial images of the person. The facial recognition system further includes a facial recognition module that is operatively coupled to the memory. The facial recognition module is configured to receive a new facial image, and to select one or more facial recognition engines based on one or more facial image parameters of the new facial image, and to use the selected facial recognition engines to compare the new facial image with facial models that are based upon facial images stored in the facial image database in order to identify the person in the new facial image.

Description

BACKGROUND
• The present disclosure relates generally to facial recognition systems, and more particularly, to facial recognition systems that employ a facial recognition engine to compare a facial image with a representation that is based upon facial images stored in a facial image database created and maintained by the facial recognition system.
SUMMARY
• The disclosure relates to facial recognition systems that employ one or more facial recognition engines to compare a facial image with a representation that is based upon facial images stored in a facial image database created and maintained by the facial recognition system. In some instances, the facial recognition system may be configured to monitor still and/or video sources capturing facial images of individuals within a space, and then utilize one or more facial recognition engines to identify the individuals within the space. In some cases, the facial recognition system may report back to a building automation system with the identity of the individuals seen within the space so that the building automation system may take appropriate action. For example, if the building automation system includes an HVAC system, the HVAC system may change a temperature set point in response to being informed that a particular person has arrived home. If the building automation system includes a security system, the security system may unlock a door of a building in response to being informed that a particular authorized person is present at the door. These are just examples.
• A particular example of the disclosure is a facial recognition system that includes an input, an output and a memory for storing a facial image database that includes a plurality of entries each corresponding to a different person, and wherein each entry includes a person identifier along with one or more facial images of the person. The system includes a facial recognition module that is operably coupled to the memory, the input and the output. The facial recognition module is configured to receive a new facial image via the input and to ascertain one or more facial image parameters from the new facial image, and then select a subset of facial recognition engines from a larger set of available facial recognition engines based at least in part on one or more of the ascertained facial image parameters. The ascertained facial image parameters may include, for example, the size of the facial image in pixels, a relative brightness of the facial image, a relative contrast of the facial image, a relative back lighting of the facial image, a relative blurriness of the facial image and/or any other suitable image parameter(s). In some cases, the ascertained facial image parameters may include whether the captured image shows the individual looking directly at the camera, or up or down and/or to the left or to the right. In some cases, the ascertained facial image parameters may include whether and/or how much of the face is obstructed by a hat, hair, glasses or other object. Some facial recognition engines may perform better on facial images under certain facial image parameters than other facial recognition engines. In any event, each of the selected facial recognition engines may compare the new facial image to facial representations that are based upon the facial images in the facial image database to try to identify the person identifier that likely corresponds to the new facial image. The facial recognition module may then send a person ID to a control module via the output, wherein the control module may control one or more building control devices based at least in part on the person ID (e.g. change a setpoint, unlock a door, etc.).
• Another example of the disclosure includes a method of recognizing individuals within a building space. Access is gained to a facial image database that includes a plurality of enrolled persons, where the facial image database includes a facial image for each of the plurality of enrolled persons under each of a plurality of different facial conditions. One or more video feeds are monitored that provide images of spaces within the building space, at least some of the images including images of persons within the building space. The one or more video feeds are processed to detect one or more facial images of a person within the building space and one or more facial recognition engines may be selected to compare the detected facial image with facial models that are based on the facial images in the facial image database. Selecting the one or more facial recognition engines is based at least in part on one or more image criteria of the detected facial image. An identified one of the plurality of enrolled persons included in the facial image database that is identified in the detected facial image may be received from the selected one or more facial recognition engines and the identified one of the plurality of enrolled persons may be reported to a building automation system. One or more building control devices of the building automation system may be controlled based at least in part on the identified one of the plurality of enrolled persons.
• Another example of the disclosure includes a method of identifying an individual. The method includes monitoring a video feed that provides a series of images of activity in or around a building and extracting one or more images from the series of images of the video feed. The extracted one or more images are analyzed to find facial images, and the facial images are quantified to find a query-able facial image. One or more facial recognition engines are selected based at least in part upon one or more image properties of the query-able facial image. The query-able facial image is sent to the selected one or more facial recognition engines. The selected one or more facial recognition engines are configured to compare the query-able facial image with facial models that are based upon facial images within the facial image database. Facial recognition engine results are provided that include an identity of a person shown within the query-able facial image (if the person is present in the facial image database) and in some cases an associated confidence value. One or more building control devices may then be controlled based at least in part on the identity of the person shown within the query-able facial image.
• The above summary of some illustrative embodiments is not intended to describe each disclosed embodiment or every implementation of the present disclosure. The Figures, and Description, which follow, more particularly exemplify some of these embodiments.
BRIEF DESCRIPTION OF THE FIGURES
• The disclosure may be more completely understood in consideration of the following description in connection with the accompanying drawings, in which:
• FIG. 1 is a schematic block diagram of an illustrative facial recognition system;
• FIG. 2 is a schematic block diagram of an illustrative enrollment module forming a portion of the facial recognition system ofFIG. 1;
• FIG. 3 is a schematic block diagram of an illustrative capture module forming a portion of the facial recognition system ofFIG. 1;
• FIG. 4 is a schematic block diagram of an illustrative facial recognition system;
• FIG. 5 is a schematic block diagram of an illustrative facial image database usable in the illustrative facial recognition systems ofFIG. 1 andFIG. 4;
• FIG. 6 is a flow diagram showing an illustrative method of recognizing individuals within a space; and
• FIG. 7 is a flow diagram showing an illustrative method of identifying an individual.
• While the disclosure is amenable to various modifications and alternative forms, specifics thereof have been shown by way of example in the drawings and will be described in detail. It should be understood, however, that the intention is not to limit the disclosure to the particular embodiments described. On the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the disclosure.
DESCRIPTION
• For the following defined terms, these definitions shall be applied, unless a different definition is given in the claims or elsewhere in this specification.
• All numeric values are herein assumed to be modified by the term “about,” whether or not explicitly indicated. The term “about” generally refers to a range of numbers that one of skill in the art would consider equivalent to the recited value (i.e., having the same function or result). In many instances, the terms “about” may include numbers that are rounded to the nearest significant figure.
• The recitation of numerical ranges by endpoints includes all numbers within that range (e.g. 1 to 5 includes 1, 1.5, 2, 2.75, 3, 3.80, 4, and 5).
• As used in this specification and the appended claims, the singular forms “a”, “an”, and “the” include plural referents unless the content clearly dictates otherwise. As used in this specification and the appended claims, the term “or” is generally employed in its sense including “and/or” unless the content clearly dictates otherwise.
• It is noted that references in the specification to “an embodiment”, “some embodiments”, “other embodiments”, etc., indicate that the embodiment described may include one or more particular features, structures, and/or characteristics. However, such recitations do not necessarily mean that all embodiments include the particular features, structures, and/or characteristics. Additionally, when particular features, structures, and/or characteristics are described in connection with one embodiment, it should be understood that such features, structures, and/or characteristics may also be used connection with other embodiments whether or not explicitly described unless clearly stated to the contrary.
• The following description should be read with reference to the drawings in which similar structures in different drawings are numbered the same. The drawings, which are not necessarily to scale, depict illustrative embodiments and are not intended to limit the scope of the disclosure.
• FIG. 1 is a schematic block diagram of an illustrativefacial recognition system10 that may, for example, be configured to create and update a database of facial images and to use that database of facial images to identify an individual. In cases, thefacial recognition system10 includes anenrollment module12 and acapture module14. While illustrated as separate components, theenrollment module12 and thecapture module14 may individually or in combination be manifested in a controller that may be part of a building automation system (e.g. an HVAC panel such as a thermostat, a security panel, etc.). In some cases, theenrollment module12 and/or thecapture module14, or at least some functionally of one or both modules, may be manifested in a server or a cloud-based application. In some cases, theenrollment module12 and/or thecapture module14, or at least some functionally of one or both modules, may be manifested in mobile device such as a tablet computer, laptop computer or smartphone. In some cases, theenrollment module12 and/or thecapture module14, or at least some functionally of one or both modules, may be manifested in a desktop computer.
• Theenrollment module12, may, for example, be responsible for creating and maintaining afacial images database16. As will be discussed, theenrollment module12 may obtain facial images from a variety of different sources and may in some cases analyze the facial images for quality before storing the facial images in thefacial images database16. In some cases, theenrollment module12 may be responsible for periodically updating the facial images stored in thefacial images database16 to account for changing styles (hair styles, facial hair, glasses, etc.), aging and the like of the individuals in thefacial images database16.
• Thecapture module14 may be responsible for obtaining facial images of individuals to be identified. In some cases, thecapture module14 may receive live video of a space, and may analyze the live video to find facial images of persons to be identified. Once thecapture module14 finds one or more facial images of person(s) that are to be identified, thecapture module14 may provide the one or more facial images to afacial recognition module18. One of the tasks of thefacial recognition module18 is to determine which of a variety of differentfacial recognition engines20 are to be used to query thefacial images database16 in order to identify the persons in the one or more facial images. In some cases, whichfacial recognition engine20 to use may be determined at least in part by one or more image parameters of the one or more facial images. In some cases, thefacial recognition module18 includes a facial recognitionengine evaluation module22 that determines which of thefacial recognition engines20 are to be used. In some cases, a singlefacial recognition engine20 may be used. In other situations, two, three or more distinctfacial recognition engines20 may be used. InFIG. 1, thefacial recognition engines20 include anENGINE #1 labeled as24, anENGINE #2 labeled as26 through an ENGINE #N labeled as28.
• The facial recognitionengine evaluation module22 may analyze a facial image to determine one or more image parameters of the facial image, and may utilize the one or more image parameters to determine which of thefacial recognition engines20 should be used in order to identify the person shown in the facial image. In some instances, the facial recognition engine evaluation module22 (or thefacial recognition module18 itself) may evaluate multiple images of a particular individual (e.g. multiple images of a video sequence) to determine which of the multiple images are most likely to provide good results. In some cases, the image parameters may include, for example, the size of the facial image in pixels, a relative brightness of the facial image, a relative contrast of the facial image, a relative back lighting of the facial image, a relative blurriness of the facial image, and/or any other suitable image parameter(s). In some cases, the image parameters may include whether the captured image shows the individual looking directly at the camera, or up or down and/or to the left or to the right. In some cases, the image parameters may include whether and/or how much of the face is obstructed by a hat, hair, glasses or other object. The engine evaluation model may store a table that maps certain image parameters to certainfacial recognition engines20.
• In some cases, a specificfacial recognition engine20 may be selected if the image parameters fall within a range suitable for thatfacial recognition engine20. The suitable range thresholds may be selected manually based on engine specifications, or may be based upon prior testing performed on a sample of facial images. In some cases, the suitable range thresholds may also be selected automatically, and may be adjusted over time, based on comparison of different facial recognition engine results on facial images processed by the system. Examples are shown in Table One, below:

• TABLE ONE
  	Engine A	Engine B	Engine C
  Image Parameter	Range	Range	Range
  Face size (pixels)	40 to 90	120 to unlimited	60 to unlimited
  Face Contrast	0.9 to 1.0	0.5 to 1.0	0.8 to 1.0
  Face Orientation	−20° to +20°	−30° to +30°	−40° to +40°
  tilt
  Facial Occlusion	no occlusion	eye occlusion	mouth and forehead
  	of facial	allowed	occlusion allowed
  	landmarks	(glasses)

• It will be appreciated that these are just examples, and a variety of other image parameters may be considered. For each facial recognition engine, the input parameters may be described as a range, as specific values, or a specific set of conditions. In some cases, a weighed sum combining various parameters may also be considered. For example, if the face size for a particular facial image is between 40 and 90 pixels, facial sharpness needs to be greater than or equal to 0.8; and if the face size is greater than 90 pixels, facial sharpness needs to be greater than or equal to 0.7.
• A variety of commercially available facial recognition engines may be utilized. Examples of suitable cloud-based APIs (Application Programming Interface) for facial recognition include but are not limited to Microsoft Face API and Amazon Rekognition. Examples of suitable facial recognition engines that may be employed as on-site software or be integrated into other products include but are not limited to NEC Face Recognition, MorphoArgusand Cognitec.
• Once the facial recognitionengine evaluation module22 determines which of thefacial recognition engines20 to select (often a subset of a larger set of available facial engines20), thefacial recognition module18 may provide the selected facial recognition engine(s)20 with the facial image. In some cases, one or more of thefacial recognition engines20 may be hosted on a remote server, but this is not required. Upon querying, the selected facial recognition engine(s)20 may return an identity of the person shown in the facial image(s) that was sent to the selected facial recognition engine(s)20. The selected facial recognition engine(s)20 may also return a confidence value that provides an indication of how confident (i.e. likely) that the identity of the person is correct. For example, if the best available facial image sent to the selected facial recognition engine(s) is poorly lit and blurry, the confidence value may be relatively low. Conversely, if the best available facial image sent to the selected facial recognition engine(s)20 is well-light and clear, and is an image of the individual looking directly or nearly directly at the camera, the confidence value may be relatively high.
• FIG. 2 is a schematic block diagram of anillustrative enrollment module12. As noted, a function of theenrollment module12 is to obtain facial images that may be placed in the facial images database16 (FIG. 1) and subsequently used to build or update a facial model based upon facial images in thefacial images database16 in order to identify persons in captured facial images. In some instances, theenrollment module12 includes animage input module30. Theimage input module30 may obtain facial images from a variety of different sources. Examples of suitable image sources include aselfies module32, aphotos module34, a capturedimages module36 and asocial media module38. In some cases, theselfies module32 may instruct individuals who are expected to be in the building space and are enrolling in the facial recognition system to take a series of selfies. For example, the individuals may be instructed to take and upload selfies showing themselves, or at least their faces, looking directly at the camera, looking left, looking right, looking up and looking down. In some cases, theselfies module32 may instruct the individual to take multiple selfies with their hair up and their hair down, with and without facial jewelry like earrings, nose piercings, lip piercings, and/or various glasses, for example. In some cases, theselfies module32 may be implemented, at least in part, on a mobile device such as a smartphone or tablet computer, but this is not required.
• In some cases, thephotos module34 may be configured to go through online and/or otherwise electronic photo libraries looking for suitable facial images. These can include photo libraries stored on a personal computer, on the cloud, on a mobile device, and/or any other device. Thephotos module34 may assemble multiple facial images for a particular individual, may display the multiple facial images, and ask the individual to confirm that each of the images is in fact of that individual. In some instances, the capturedimages module36 may include or otherwise be operably coupled with a still camera, a video camera and the like, and may capture facial images as individuals move about the space. The capturedimages module36 may compile these images, and in some cases may ask the individuals to confirm their identity.
• Thesocial media module38 may scan social media accounts, such as but not limited to Facebook, Snapchat and the like, looking for suitable facial images of an individual. In some cases, thesocial media module38 may display the found images and ask for identity confirmation.
• In some cases, theenrollment module12 may include an imagequality assessment module40 that receives facial images from theimage input module30 and analyzes the received facial images to confirm that the images are of sufficient quality to be of use. The facial images that are believed to be of sufficient quality, and represent a suitable variety of poses and images (looking left, looking right, etc.) may be passed on to animage organization module42. In some cases, a particular facial image pose may be determined to be of less-than sufficient quality, and the imagequality assessment module40 may ask theimage input module30 to obtain a higher quality facial image of that particular facial image pose if possible.
• Facial images that are deemed to be of sufficient quality, and of appropriate facial poses, are forwarded to theimage organization module42. In some cases, theimage organization module42 may at least partially contribute to the organization of facial images within the facial images database16 (FIG. 1). Facial images may be organized in any suitable manner. In some cases, for example, the facial images for a particular individual may include images of the individual looking directly at the camera, images of the individual looking above the camera, images of the individual looking below the camera, images of the individual looking to the left of the camera, and images of the individual looking to the right of the camera. In some cases, facial images for a particular individual may be organized by whether they are wearing their hair up or down, have facial hair or are clean-shaven, whether or not they are wear jewelry, glasses and the like. In some cases, facial images for a particular individual may also be organized by the size of the facial image in pixels, the relative brightness of the facial image, the relative contrast of the facial image, the relative back lighting of the facial image, the relative blurriness of the facial image and/or by any other suitable image parameter. These are just examples.
• FIG. 3 is a schematic block diagram of thecapture module14. While thecapture module14 is configured to capture facial images of individuals within a space so that they can be identified, in some cases, thecapture module14 may also assist theenrollment module12 in initially capturing facial images for populating the facial images database16 (FIG. 1). In some cases, thecapture module14 includes avideo capture module50. In some instances, thevideo capture module50 may be operably coupled to one or more still cameras and/or video cameras that are distributed within a space. In some cases, still images may be captured by the still cameras and/or from image frames captured by the video cameras. In some cases, still images may be captured 30 times (frames) per second using a video camera, although this is just an example.
• The still images may be forwarded to aface detection module52, which analyzes the still images looking for facial images. Once a possible facial image is detected, in some cases, subsequent still images are analyzed by aface tracking module54 looking for confirmation the individual is still there and/or looking for better quality facial images of that individual for subsequent identification. A faceimage evaluation module56 may review the facial images to ascertain whether and which of the captured image(s) are of sufficient quality and/or pose to be of use in identifying the individual shown in the captured image(s).
• FIG. 4 is a schematic block diagram of an illustrativefacial recognition system60. Thefacial recognition system60 includes aninput62 and anoutput64. Amemory67 may be configured to store thefacial images database16. In some cases, thefacial images database16 includes a plurality of entries each corresponding to a different person, and each entry includes a person identifier along with one or more facial images of the person. In some instances, thefacial images database16 may include multiple facial images for each person identifier, with some of the multiple facial images representing the person at one or more of different facial angles, different facial lighting, different facial size in terms of pixels, and different facial obstructions. Examples of different facial obstructions include but are not limited to differing hair style, wearing glasses, not wearing glasses, wearing a hat, not wearing a hat, and differing ages.
• Thefacial recognition module18 is operably coupled to theinput62, theoutput64 and to thememory67 and is configured to receive a new facial image via theinput62 and to ascertain one or more facial image parameters from the new facial image. Thefacial recognition module18 is configured to select a subset of facial recognition engines20 (FIG. 1) from a larger set of available facial recognition engines20 (FIG. 1) based at least in part on one or more of the ascertained facial image parameters. The ascertained facial image parameters may include, for example, the size of the facial image in pixels, a relative brightness of the facial image, a relative contrast of the facial image, a relative back lighting of the facial image, a relative blurriness of the facial image and/or any other suitable image parameter(s). In some cases, the ascertained facial image parameters may include whether the captured image shows the individual looking directly at the camera, or up or down and/or to the left or to the right. In some cases, the ascertained facial image parameters may include whether and/or how much of the face is obstructed by a hat, hair, glasses or other object.
• At least some of thefacial recognition engines20 may include cloud-based facial recognition engines, but this is not required. In some cases, the selected subset offacial recognition engines20 may include two or more distinctfacial recognition engines20. In some cases, the selected subset offacial recognition engines20 may include only a singlefacial recognition engine20.
• Each of thefacial recognition engines20 is configured to compare the new facial image to facial models or other facial representations that are based upon the facial images in thefacial image database16 and to identify a person identifier that likely corresponds to the new facial image. Thefacial recognition module18 may be configured to evaluate the person identifiers and confidence levels returned by the selectedfacial recognition engines20. If the returned person identifiers are the same, thefacial recognition module18 will as sign a high confidence to the output person ID. However, if the returned person identifiers differ, thefacial recognition module18 may output the person ID with the highest combined confidence, or may select additional facial recognition engines to evaluate the new facial image. In case of disagreement, the output person ID will be assigned lower confidence. Thefacial recognition module18 is configured to send a person ID to acontrol module66 via theoutput64, wherein thecontrol module66 is configured to control one or morebuilding control devices68,70 based at least in part on the person ID. In some cases, thefacial recognition module18 may be further configured to process the person identifiers identified by each of the subset offacial recognition engines20 to determine the person ID that is sent to thecontrol module66.
• In some cases, at least one of thefacial recognition engines20 may provide a confidence level of the person identifier that likely corresponds to the new facial image. In some cases, particularly if the new facial image was sent to multiplefacial recognition engines20 for identification, thefacial recognition module18 may be configured to determine a confidence level in the person ID that is based at least in part on the confidence level of the person identifier provided by each of one or more of the subset offacial recognition engines20. In some cases, if the confidence level in the person ID is below a threshold confidence level, thefacial recognition module18 may select a differentfacial recognition engine20, or a different subset offacial recognition engines20, and may try again. This may be repeated until an acceptable confidence level is achieved. If an acceptable confidence level cannot be achieved, thefacial recognition module18 may report to theoutput module64 that an unknown person was seen in the new facial image.
• FIG. 5 is a schematic illustration of an illustrativefacial images database116. In some cases, thefacial images database116 may be considered as being an illustrative but non-limiting example of thefacial images database16. It will be appreciated that the information within thefacial images database16 may be organized in any suitable fashion. Thefacial images database116 includes a plurality of facial images that are organized by individual. To illustrate, thefacial images database116 may include anINDIVIDUAL #1 labeled as118, anINDIVIDUAL #2 labeled as120 through an INDIVIDUAL #P labeled as122. A number of facial images are organized underneath each individual118,120,122. As illustrated, the individual118 includes anIMAGE #1 labeled as124, anIMAGE #2 labeled as126 through an IMAGE #M labeled as128. Similarly, the individual120 includes anIMAGE #1 labeled as134, anIMAGE #2 labeled as136 through an IMAGE #M labeled as138 and the individual122 includes anIMAGE #1 labeled as144, anIMAGE #2 labeled as146 through an IMAGE #M labeled as148.
• In some cases, the images for each individual118,120,122 may be organized in a similar fashion. For example, theimage124 may represent a straight on view of the individual118, theimage134 may represent a straight on view of the individual120 and theimage144 may represent a straight on view of the individual122. Theimages126,136,146 may represent left profiles of theindividuals118,120,122, respectively. These are just examples. It will be appreciated that the various views of each individual118,120,122, and perhaps views with and without facial obstructions and/or other characteristics, may be organized in a similar manner. In some cases, facial images for a particular individual may be organized by whether they are wearing their hair up or down, have facial hair or are clean-shaven, whether or not they are wear jewelry, glasses and the like. In some cases, facial images for a particular individual may be organized by the size of the facial image in pixels, the relative brightness of the facial image, the relative contrast of the facial image, the relative back lighting of the facial image, the relative blurriness of the facial image and/or by any other suitable image parameter. These are just examples.
• FIG. 6 is a flow diagram showing anillustrative method150 of recognizing individuals within or around a building space. As generally shown atblock152, access is gained to a facial image database (such as thefacial images databases16,116) that includes a plurality of enrolled persons, where the facial image database includes a facial image for each of the plurality of enrolled persons under each of a plurality of different facial conditions. In some cases, the plurality of different facial conditions include two or more of the person looking up, the person looking down, the person looking to the left, the person looking to the right and the person looking straight ahead.
• In some cases, the facial image database organizes the facial images for each of the plurality of enrolled persons at each of a plurality of different facial conditions into predetermined separate categories. Examples of the separate categories include one or more of the person with glasses, the person without glasses, the person with their hair worn up, the person with their hair worn down, the person clean shaven, the person not clean shaven, the person wearing jewelry, the person not wearing jewelry, and the person wearing a hat, the person not wearing a hat and the person wearing a scarf. In some instances, the separate categories include one or more of the person looking to the left, the person looking to the right, the person looking up, the person looking down, and the person looking straight ahead. In some cases, the separate categories may include the size of the facial image in pixels, the relative brightness of the facial image, the relative contrast of the facial image, the relative back lighting of the facial image, the relative blurriness of the facial image and/or by any other suitable image parameter. These are just examples.
• One or more video feeds providing images of spaces within or around the building space may be monitored, at least some of the images include images of persons within the building space, as seen atblock154. As seen atblock156, the one or more video feeds may be processed to detect one or more facial images of a person within or around the building space. One or more facial recognition engines may be selected, as generally indicated atblock158. In some cases, selecting the one or more facial recognition engines is based at least in part on one or more image criteria of the one or more detected facial images. The one or more image criteria may include, for example, the size of the facial image in pixels, a relative brightness of the facial image, a relative contrast of the facial image, a relative back lighting of the facial image, a relative blurriness of the facial image and/or any other suitable image parameter(s). In some cases, the one or more image criteria may include whether the captured image shows the individual looking directly at the camera, or up or down and/or to the left or to the right. In some cases, the one or more image criteria may include whether and/or how much of the face is obstructed by a hat, hair, glasses or other object. Some facial recognition engines may perform better on facial images under certain image criteria than other facial recognition engines.
• As seen atblock160, an identified one of the plurality of enrolled persons included in the facial image database that is identified in the one or more detected facial images may be received from the selected one or more facial recognition engines. The identified one of the plurality of enrolled persons may be reported to a building automation system, as generally seen atblock162, and one or more building control devices of the building automation system may be controlled based at least in part on the identified one of the plurality of enrolled persons, as indicated atblock164.
• In some cases, the building automation system includes an HVAC system, and the building control device may include a building control user interface device that allows the identified one of the plurality of enrolled persons to change one or more building control parameters only when the identified one of the plurality of enrolled persons has been granted permission to change one or more building control parameters. In some instances, the building automation system includes an access control system, and a building access device is controlled to allow entry of the identified one of the plurality of enrolled persons only when the identified one of the plurality of enrolled persons has been granted permission to enter.
• FIG. 7 is a flow diagram illustrating amethod170 of identifying an individual. In some cases, themethod170 includes creating a facial images database by soliciting facial images of each of a plurality of enrolled persons under each of a plurality of facial conditions that include one or more of the person looking to the left, the person looking to the right, the person looking up, the person looking down, and the person looking straight ahead. A video feed that provides a series of images of activity in or around a building is monitored, as indicated atblock174. One or more images may be extracted from the series of images of the video feed, as seen atblock176. The extracted one or more images may be analyzed to find facial images, as indicated atblock178, and are quantified to find a query-able facial image, as noted atblock180.
• As indicated atblock182, one or more facial recognition engines may be selected based at least in part upon one or more image properties of the query-able facial image and the query-able facial image may be sent to the selected one or more facial recognition engines as indicated atblock184, where the selected one or more facial recognition engines are configured to compare the query-able facial image with facial models that are based upon facial images stored within the facial image database. As noted atblock186, facial recognition engine results that include an identity of a person shown within the query-able facial image as well as an associated confidence value may be provided. In some cases, and as indicated atblock188, one or more building control devices may be controlled based at least in part on the identity of the person shown within the query-able facial image.
• It should be understood that this disclosure is, in many respects, only illustrative. Changes may be made in details, particularly in matters of shape, size, and arrangement of steps without exceeding the scope of the disclosure. This may include, to the extent that it is appropriate, the use of any of the features of one example embodiment being used in other embodiments.

Claims (20)

What is claimed is:

1. A facial recognition system, comprising:

an input;

an output;

a memory for storing a facial image database, wherein the facial image database includes a plurality of entries each corresponding to a different person, and wherein each entry includes a person identifier along with one or more facial images of the person;

a facial recognition module operatively coupled to the memory, the input and the output, the facial recognition module configured to:

receive a new facial image via the input and ascertain one or more facial image parameters from the new facial image;

select a subset of facial recognition engines from a larger set of available facial recognition engines based at least in part on one or more of the ascertained facial image parameters, wherein each of the facial recognition engines is configured to compare the new facial image to facial representations that are based upon the facial images stored in the facial image database and identify a person identifier that likely corresponds to the new facial image; and

send a person ID to a control module via the output, wherein the control module is configured to control one or more building control devices based at least in part on the person ID.

2. The facial recognition system ofclaim 1, wherein the selected subset of facial recognition engines consists of one facial recognition engine selected from the larger set of available facial recognition engines.

3. The facial recognition system ofclaim 1, wherein the selected subset of facial recognition engines comprises two or more facial recognition engines selected from the larger set of available facial recognition engines.

4. The facial recognition system ofclaim 3, wherein the facial recognition module is further configured to process the person identifiers identified by each of the subset of facial recognition engines to determine the person ID that is sent to the control module.

5. The facial recognition system ofclaim 1, wherein at least one of the facial recognition engines provide a confidence level of the person identifier that likely corresponds to the new facial image.

6. The facial recognition system ofclaim 5, wherein the facial recognition module is configured to determine a confidence level in the person ID that is based at least in part on the confidence level of the person identifier provided by each of one or more of the subset of facial recognition engines.

7. The facial recognition system ofclaim 6, wherein the facial recognition module is configured to select a different subset of facial recognition engines from the larger set of available facial recognition engines when the confidence level in the person ID is below a threshold confidence level.

8. The facial recognition system ofclaim 1, wherein the facial image database comprises multiple facial images for each person identifier, with some of the multiple facial images representing the person at one or more of different facial angles, different facial lighting, different facial size in terms of pixels, and different facial obstructions.

9. The facial recognition system ofclaim 8, wherein the different facial obstructions comprise one or more of differing hair style, wearing glasses, not wearing glasses, wearing a hat, not wearing a hat, and differing ages.

10. The facial recognition system ofclaim 1, wherein at least one of the larger set of available facial recognition engines comprise a cloud-based facial recognition engine.

11. A method of recognizing individuals within a building space, the method comprising:

gaining access to a facial image database that includes a plurality of enrolled persons, where the facial image database includes a facial image for each of the plurality of enrolled persons under each of a plurality of different facial conditions;

monitoring one or more video feeds providing images of spaces within the building space, at least some of the images include images of persons within the building space;

processing the one or more video feeds to detect one or more facial images of a person within the building space;

selecting one or more facial recognition engines to compare the detected one or more facial images with facial models that are based upon the facial images in the facial image database, where selecting the one or more facial recognition engines is based at least in part on one or more image criteria of the one or more detected facial images;

receiving from the selected one or more facial recognition engines an identified one of the plurality of enrolled persons included in the facial image database that is identified in the one or more detected facial images;

reporting the identified one of the plurality of enrolled persons to a building automation system; and

controlling one or more building control devices of the building automation system based at least in part on the identified one of the plurality of enrolled persons.

12. The method ofclaim 11, wherein the plurality of different facial conditions include two or more of the person looking up, the person looking down, the person looking to the left, the person looking to the right and the person looking straight ahead.

13. The method ofclaim 11, wherein the facial image database organizes the facial images for each of the plurality of enrolled persons at each of a plurality of different facial conditions into predetermined separate categories that can be separately queried by the one or more facial recognition engines.

14. The method ofclaim 13, wherein the separate categories include one or more of the person with glasses, the person without glasses, the person with their hair worn up, the person with their hair worn down, the person clean shaven, the person not clean shaven, the person wearing jewelry, the person not wearing jewelry, and the person wearing a hat, the person not wearing a hat, and a person wearing a scarf.

15. The method ofclaim 13, wherein the separate categories include one or more of the person looking to the left, the person looking to the right, the person looking up, the person looking down, and the person looking straight ahead.

16. The method ofclaim 11, wherein the building automation system comprises an HVAC system, and the method further comprises controlling a building control user interface device to allow the identified one of the plurality of enrolled persons to change one or more building control parameters only when the identified one of the plurality of enrolled persons has been granted permission to change one or more building control parameters.

17. The method ofclaim 11, wherein the building automation system comprises an access control system, and the method further comprises controlling a building access device to allow entry of the identified one of the plurality of enrolled persons only when the identified one of the plurality of enrolled persons has been granted permission to enter.

18. A method of identifying an individual, the method comprising:

monitoring a video feed that provides a series of images of activity in or around a building;

extracting one or more images from the series of images of the video feed;

analyzing the extracted one or more images to find facial images;

quantifying the facial images to find a query-able facial image;

selecting one or more facial recognition engines based at least in part upon one or more image properties of the query-able facial image;

sending the query-able facial image to the selected one or more facial recognition engines, the selected one or more facial recognition engines configured to compare the query-able facial image to facial models that are based upon facial images stored in a facial image database;

providing facial recognition engine results that include an identity of a person shown within the query-able facial image as well as an associated confidence value; and

controlling one or more building control devices based at least in part on the identity of the person shown within the query-able facial image.

19. The method ofclaim 18, further comprising creating the facial image database by soliciting facial images of each of a plurality of enrolled persons under each of a plurality of facial conditions.

20. The method ofclaim 19, wherein the plurality of facial conditions includes one or more of the person looking to the left, the person looking to the right, the person looking up, the person looking down, and the person looking straight ahead.

Images