Non-contact temperature measurement in thermal imaging systems and methods

Abstract

A system and method including an image capture component configured to capture infrared images of a scene, and a logic device configured to identify a target in the images, acquire temperature data associated with the target based on the images, evaluate the temperature data and determine a corresponding temperature classification, and process the identified target according to the temperature classification. The logic device identifies a person and tracks the person over a subset of the images, identifies a measured location of the target in the subset of images based on a neural network identified feature point of the target, and measures a temperature at the location using corresponding values from one or more captured thermal images. The logic device is further configured to calculate a core body temperature of the target using the temperature data to determine whether the target is feverish, and calibrate using one or more blackbodies.

Description

[0001] Cross-references to related applications

[0002] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 006,063, filed April 6, 2020, entitled “Non-Contact Temperature Measurement in Thermal Imaging Systems and Methods,” the entire contents of which are incorporated herein by reference.

[0003] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 090,691, filed October 12, 2020, entitled “Non-contact temperature measurement in thermal imaging systems and methods,” the entire contents of which are incorporated herein by reference. Technical Field

[0004] One or more embodiments of this disclosure generally relate to non-contact temperature measurement, and more specifically, for example, to systems and methods for receiving and processing thermal data for health screening applications. Background Technology

[0005] Thermal cameras can be used to measure the surface temperature of people to help identify individuals with fever. While thermal cameras work quite well in certain controlled environments, deploying non-contact thermal measurement systems in public places, crowded environments, and other environments where people move freely presents challenges. Other issues arise in critical applications, such as identifying individuals with fever during pandemics, where real-time processing, adaptation to changing environments, and accurate identification are crucial to avoiding false positives or missed infections.

[0006] For example, using thermal imaging for individual screening can create bottlenecks, causing delays in movement and bringing people in queues closer together. This bottleneck can increase the risk of transmission in high-traffic areas such as airports and train stations. In some cases, monitoring people in groups or clusters is preferable, such as in workplaces (offices, production lines), classrooms, public transportation, courtrooms, or other locations where populations may be at risk of infection or a resurgence of an epidemic / pandemic (such as seasonal influenza).

[0007] In view of the foregoing, there is a continued need in the art for improved systems and methods for non-contact temperature measurement. Summary of the Invention

[0008] This document describes improved systems and methods for non-contact temperature measurement. The scope of this disclosure is defined by the claims, which are incorporated herein by reference. A more complete understanding of embodiments of this disclosure, and their additional advantages, will be gained by considering the following detailed description of one or more embodiments. Reference will be made to the accompanying drawings, which will be briefly described first. Attached Figure Description

[0009] Figure 1A An example thermal imaging system is shown, deployed at a checkpoint and configured to identify people with elevated body temperature according to one or more embodiments of the present disclosure.

[0010] Figure 1B An example thermal imaging system according to one or more embodiments of the present disclosure is shown.

[0011] Figure 1C Example operation of a thermal imaging system according to one or more embodiments of the present disclosure is shown.

[0012] Figure 1D Incremental considerations for avoiding false positives and false negatives are illustrated according to one or more embodiments.

[0013] Figure 2A An example thermal imaging monitoring system according to one or more embodiments of the present disclosure is shown.

[0014] Figure 2B Example thermal camera shape factors and example implementations are shown according to one or more embodiments of the present disclosure.

[0015] Figure 2C An example implementation of a thermal camera according to one or more embodiments of the present disclosure is shown.

[0016] Figure 3A An example virus detection system and network according to one or more embodiments of this disclosure are shown.

[0017] Figure 3B An example of a remote virus detection system according to one or more embodiments of the present disclosure is shown.

[0018] Figure 3C An example virus detection training process according to one or more embodiments of this disclosure is shown.

[0019] Figure 3D An example virus detection verification process according to one or more embodiments of this disclosure is shown.

[0020] Figure 3E Training images with annotations according to one or more embodiments of this disclosure are shown.

[0021] Figure 3F The properties of example facial poses and annotations for training datasets according to one or more embodiments of this disclosure are shown.

[0022] Figure 3G Example distance variations in a training dataset according to one or more embodiments of this disclosure are shown.

[0023] Figure 3H An example procedure for detecting elevated temperature in a subject, according to one or more embodiments, is shown.

[0024] Figure 4 This is a flowchart illustrating an example process in bold, showing use case information according to one or more embodiments of this disclosure.

[0025] Figure 5A The present disclosure illustrates one or more embodiments that can be implemented according to this disclosure. Figure 4 Example boldface labels used in the process.

[0026] Figure 5B An example passive blackbody and telemetry system according to one or more embodiments of the present disclosure is shown.

[0027] Figure 6 This is a flowchart illustrating an example process for identifying pixels on a human face according to one or more embodiments of the present disclosure.

[0028] Figure 7 It is a graph showing temperature measurements over time according to one or more embodiments.

[0029] Figure 8 An example three-dimensional thermal system including drift monitoring is shown according to one or more embodiments of the present disclosure.

[0030] Figure 9 An example airport millimeter-wave scanner according to one or more embodiments of the present disclosure is shown.

[0031] Figure 10A An example sampling interface according to one or more embodiments of the present disclosure is shown.

[0032] Figure 10B An example method for scanning elevated body temperature according to one or more embodiments of the present disclosure is shown.

[0033] Figure 10C The use of one or more embodiments according to this disclosure is illustrated. Figure 10A Example image of the method.

[0034] Figure 11An example method is shown for using an isotherm in conjunction with an alarm threshold in an elevated body temperature system, according to one or more embodiments.

[0035] Figure 12A and 12B An example method for improving measurement accuracy using temperature offset according to one or more embodiments is shown.

[0036] The embodiments and advantages of this disclosure can be best understood by referring to the following detailed description. It should be understood that the same reference numerals are used to identify one or more of the same elements shown in the figures. Detailed Implementation

[0037] According to various embodiments, this disclosure describes improved systems and methods for non-contact temperature measurement. The various systems and methods disclosed herein can be used with thermal cameras for individual screening (e.g., handheld thermal cameras, stationary thermal cameras, etc.), other temperature measurement systems deployed to provide real-time screening of people in various environments (including crowds and high-traffic areas), and for detecting elevated skin or body temperatures.

[0038] This disclosure improves processing and accuracy and can be used in critical scenarios such as monitoring and controlling people with infectious diseases during epidemics or pandemics (e.g., monitoring the spread of viruses during a global outbreak of coronaviruses such as COVID-19). Detecting, tracking, and containing the spread of infectious diseases is often a top priority for governments and healthcare professionals to help slow and contain the spread of viruses. In some embodiments disclosed herein, thermal imaging systems can be positioned in high-traffic areas to provide rapid screening to detect individuals with fever. Machine learning is used to analyze thermal images and / or other data to detect, tag, and track people in crowds, identify head positions, recognize facial features (e.g., foreheads), and obtain temperature measurements of the forehead, corners of the eyes, or other desired locations. In some embodiments, zoom optics are used to target specific temperature measurement locations on the person being tracked.

[0039] Thermal imaging for virus / infection surveillance

[0040] In various embodiments, thermal imaging technology is described that can be used to rapidly identify individuals with fever in large crowds to support increased surveillance and security measures, as well as prioritizing monitoring and response assets. Governments, healthcare providers, public / private enterprises, individuals, and / or other entities can use this technology to provide enhanced situational awareness across multiple mission areas. In some embodiments, the thermal imaging device is configured to capture panoramic, high-resolution thermal images, which are then analyzed using a machine learning system capable of autonomously measuring the temperature of a crowd. The thermal imaging device can be configured to rapidly identify individuals who may have elevated temperatures, with the goal of, for example, isolating the individual from the crowd and / or preventing the individual from entering the crowd and prioritizing the individual for subsequent diagnostic screening.

[0041] The embodiments disclosed herein contribute to improved situational awareness, enabling real-time, proactive tracking of potential public transmission of diseases such as COVID-19. Disease transmission can be seasonal (e.g., flu season typically includes winter) and can be mitigated by a variety of factors, including effective mitigation methods, demographics, immunity, and others. The possibility of resurgence before an effective vaccine is approved leaves populations with significant susceptibility, even when the disease appears to be under control. The systems and methods disclosed herein provide a novel thermal imaging platform that enables continuous, presumptive surveillance of susceptible populations to rapidly identify and prevent disease resurgence, including in the period before a vaccine is completed and widely distributed.

[0042] In one implementation, the system is designed to protect designated groups of interest. The system can be deployed in workplaces (e.g., factories, hospitals, etc.), schools, shops, homes, and other locations. Thermal cameras are placed to continuously monitor the premises and conduct early detection of individuals (or other desired targets) meeting specific criteria. The system can be configured to provide situational awareness by prioritizing assets to support and assist disease treatment and minimize the impact of transmission.

[0043] In another embodiment, the system is configured to protect a location by screening people at point of entry (POE), targeting potential patients and taking steps to keep them out. For example, the system can be deployed at public transportation locations (e.g., airports, train stations, and subway stations), cruise ships, entertainment venues (e.g., arenas, theaters), and other locations. The system can be configured to minimize potential contamination through a high-speed processing system that automatically screens large numbers of people. In some embodiments, the system captures thermal images, tags people and other objects of interest, and tracks the locations of the tagged people and objects as they move through one or more scenes.

[0044] Now refer to Figure 1A , 1B Embodiments of this application are described in more detail in 1C. Figure 1A An example thermal imaging system deployed at a checkpoint is shown, configured to provide surveillance of large crowds to identify individuals with elevated temperatures that may indicate fever. Fever is often an early symptom of infection, and temperature screening has been deployed in many airports and other public transportation environments to attempt to screen and / or isolate potentially infectious individuals. Temperature measurement can be used as a coarse screening and triage technique in global healthcare crises to support the prioritization of critical medical assets.

[0045] like Figure 1A As shown, a system 50 incorporating a high-resolution thermal camera 52 can be used at screening checkpoints in transportation hubs for passenger screening. Thermal imaging at the screening checkpoint can be used as an aid to help identify individuals who may have a fever and require further diagnostic screening by a healthcare professional. In step 1 of the illustrated example, the system 50 including the high-resolution thermal camera 52 is deployed to provide remote monitoring of people at one or more checkpoints. In step 2, an operator 54 views data on a display 56 and identifies elevated temperatures determined by the thermal imaging system 50. If it is determined that the person being screened has a fever, in step 3, the person is directed to a second location for further screening (e.g., by a healthcare professional). If no elevated temperature is detected at the first checkpoint or the second screening location, the person can pass through the checkpoint. Otherwise, the person may be identified as unhealthy and at risk of infecting others, and is prevented from passing through the checkpoint. In some embodiments, the thermal camera 52 is integrated into a security checkpoint in a handheld device with high temperature resolution for use as a mobile aid to diagnostics, such as by screening differences in skin surface temperature.

[0046] refer to Figure 1B An example thermal imaging system 100 that can be used for infectious disease surveillance will now be described according to one or more embodiments. The thermal imaging system 100 can be used to capture and process thermal images to identify, track, and monitor the temperature of a person in a scene 170. As shown, the thermal imaging system 100 can be used to image a scene 170 located within the field of view of the thermal imaging system 100. The thermal imaging system 100 includes a processing unit 110, a memory unit 120, an image capture unit 130, an optical unit 132 (e.g., one or more lenses configured to receive electromagnetic radiation through an aperture 134 in a camera unit 101 and transmit the electromagnetic radiation to the image capture unit 130), an image capture interface unit 136, a display unit 140, a control unit 150, a communication unit 152, and other sensing units 142.

[0047] In various embodiments, the thermal imaging system 100 can be implemented as a handheld thermal imaging device and / or a fixed thermal imaging device, including a camera component 101 to capture image frames of scene 170, for example, within the field of view of the camera component 101. In some embodiments, the camera component 101 may include an image capture component 130, an optical component 132, and an image capture interface component 136 housed in a protective housing. The thermal imaging system 100 can represent any type of camera system suitable for imaging scene 170 and providing associated thermal image data. The thermal imaging system 100 can be implemented with the camera component 101 at various types of fixed locations and environments (e.g., airports, event venues, office buildings, etc.). In some embodiments, the camera component 101 may be mounted in a fixed device to capture continuous images of scene 170. In some embodiments, the thermal imaging system 100 may include a portable device and may be implemented, for example, as a handheld device and / or, in other examples, coupled to various types of portable devices and / or vehicles (e.g., land-based vehicles, ships, aircraft, spacecraft, or other vehicles).

[0048] Processing unit 110 may include, for example, a microprocessor, a single-core processor, a multi-core processor, a microcontroller, a logic device (e.g., a programmable logic device configured to perform processing operations), a digital signal processing (DSP) device, one or more memories for storing executable instructions (e.g., software, firmware, or other instructions) and / or any other suitable combination of processing devices and / or memories for executing instructions to perform any of the various operations described herein. Processing unit 110 is adapted to interface with and communicate with units 101, 102, 120, 140, 142, 150, and 152 to perform the methods and processing steps described herein. Processing unit 110 is also adapted to detect and classify objects in images captured by image capture unit 130 via image processing module 180, person tracking module 182, and temperature analysis module 184.

[0049] It should be understood that processing operations and / or instructions may be integrated into software and / or hardware as part of processing unit 110, or integrated into code (e.g., software or configuration data) that may be stored in memory unit 120. Embodiments of the processing operations and / or instructions disclosed herein may be stored in a non-transitory manner by a machine-readable medium (e.g., memory, hard disk drive, compact disk, digital video disk, or flash memory) for execution by a computer (e.g., a logic- or processor-based system) to perform the various methods disclosed herein.

[0050] In one embodiment, memory component 120 includes one or more memory devices (e.g., one or more memories) for storing data and information. The one or more memory devices may include various types of memory, including volatile and non-volatile memory devices such as RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Read-Only Memory), flash memory, or other types of memory. In one embodiment, processing component 110 is adapted to execute software stored in memory component 120 and / or a machine-readable medium to perform various methods, processes, and operations as described herein.

[0051] In one embodiment, the image capture component 130 includes one or more sensors for capturing image signals representing an image of scene 170. In one embodiment, the sensors of the image capture component 130 provide representation (e.g., conversion) of the captured thermal image signals of scene 170 into digital data (e.g., via an analog-to-digital converter included as part of the sensor or separate from the sensor as part of the thermal imaging system 100). The thermal sensors may include multiple infrared sensors (e.g., infrared detectors) implemented on a substrate in an array or otherwise. For example, in one embodiment, the infrared sensors may be implemented as a focal plane array (FPA). The infrared sensors may be configured to detect infrared radiation (e.g., infrared energy) from the target scene, including, for example, the mid-wave infrared (MWIR) band, the long-wave infrared (LWIR) band, and / or other thermal imaging bands that may be required in a particular embodiment. For example, the infrared sensors may be implemented as microbolometers or other types of thermal imaging infrared sensors arranged in any desired array pattern to provide multiple pixels.

[0052] Processing unit 110 may be a logic device adapted to receive image signals from image capture unit 130, process image signals (e.g., to provide image data to be processed), store image signals or image data in memory unit 120, and / or retrieve stored image signals from memory unit 120. In various aspects, as described herein, processing unit 110 may be remotely located, and processing unit 110 may be adapted to remotely receive image signals from image capture unit 130 via wired or wireless communication with image capture interface unit 136.

[0053] Display component 140 may include an image display device (e.g., a liquid crystal display (LCD)) or various other types of generally known video displays or monitors. In various embodiments, control component 150 may include user input and / or interface devices, such as a keyboard, control panel unit, graphical user interface, or other user input / output. Control component 150 may be adapted to be integrated as part of display component 140 to operate as a user input device and display device, for example, a touchscreen device adapted to receive user input signals from different portions of a touchscreen display.

[0054] Processing unit 110 may be adapted to communicate with image capture interface unit 136 (e.g., by receiving data and information from image capture unit 530). Image capture interface unit 136 may be configured to receive image signals (e.g., image frames) from image capture unit 130 and transmit the image signals to processing unit 110 directly or via one or more wired or wireless communication units (e.g., represented by connection 137) in a manner similar to communication unit 152 further described herein. In various embodiments, camera unit 101 and processing unit 110 may be positioned close to or far from each other.

[0055] The communication component 152 may be implemented as a network interface component adapted to communicate with the cloud / network 154, which includes other devices in the network, and may include one or more wired or wireless communication components. In various embodiments, the cloud / network 154 may be implemented as a single network or a combination of multiple networks, and may include wired or wireless networks, including wireless LANs, wide area networks, the Internet, cloud network services, and / or other suitable types of communication networks.

[0056] In various embodiments, thermal imaging system 100 provides the ability to detect, track, and determine the temperature of a person in scene 170 in real time. For example, thermal imaging system 100 may be configured to capture images of scene 170 using camera component 101 (e.g., an infrared camera). The captured images may be received by processing component 110 and stored in memory component 120. Image processing module 180 and person tracking module 182 may extract a subset of pixel values ​​of scene 170 corresponding to the detected person from each captured image. Temperature analysis module 184 analyzes available information to estimate the temperature of the tracked person and stores the results in memory component 120, a database, or other memory storage device according to system design preferences. In some embodiments, thermal imaging system 100 may transmit thermal image data or other sensed, calculated, and / or determined data to host system 156 and database 157 via network 154 (e.g., the Internet or the cloud) for remote processing and / or storage.

[0057] The person tracking module 182 and temperature analysis module 184 are configured to provide analysis of captured thermal images and other data to detect, track, and determine the temperature of people. The person tracking module 182 may also include other person counting and tracking functions, for example, to measure traffic flow through the area corresponding to scene 170. In various embodiments, the person tracking module 182 is connected to one or more databases and / or other sensor interfaces that provide additional data to detect / track people and determine their temperatures. For example, the database may store criteria for identifying people, reference images of known conditions, field-of-view parameters for each image capture device (e.g., for estimating the size and location of detected people and objects), common learning and configuration activities for the image capture devices, and other person tracking information.

[0058] Temperature analysis module 184 analyzes one or more thermal images of the tracked person to determine the person's temperature. In some embodiments, temperature analysis module 184 is configured to detect a measurement location on the tracked person, instruct camera component 101 to magnify the display at the desired location (e.g., via an optical or digital zoom component, such as an infrared camera with zoom optics 102) and capture a thermal image of the measurement location. Temperature analysis module 184 may also receive additional sensing data from other sensing components 142, as well as system-related data and environmental parameters. In some embodiments, facial recognition is used to identify faces at different angles and / or distances, and then find the optimal location to measure the person's temperature.

[0059] Depending on the application or implementation of the sensing, other sensing components 142 may include environmental and / or operational sensors that provide information to processing component 110 (e.g., by receiving sensor information from each sensing component 142). In various embodiments, other sensing components 142 may be adapted to provide data and information related to environmental conditions, such as internal and / or external temperature conditions, lighting conditions (e.g., daytime, nighttime, dusk, and / or dawn), humidity levels, specific weather conditions (e.g., sun, rain, and / or snow), distance (e.g., laser rangefinder), ambient sound, visible image sensors, and / or other sensor types. Therefore, other sensing components 142 may include one or more conventional sensors known to those skilled in the art for monitoring various conditions (e.g., environmental conditions) that may affect the data provided by image capture component 130 (e.g., affect the appearance of the image).

[0060] In some embodiments, an infrared camera with zoom optics 102 is configured to provide accurate thermal temperature measurements of target 103 in scene 170. Camera component 101 may include a wide field-of-view (FOV) infrared (IR) microbolometer camera that provides 360-degree imaging for identifying people in the wide scene 170 by running a neural network algorithm to recognize and track people in the scene. After a person is identified by person-tracking component 182, the paired IR camera with zoom optics 102 is instructed to focus on target 103 and perform a series of temperature measurements, which are analyzed by an algorithm to determine the number of data points required over time and to set a deviation threshold to trigger an alarm. The system then reports (e.g., to another system, to a monitoring person, etc.) that the subject should be isolated for further diagnostic evaluation by a medical professional.

[0061] In some embodiments, camera component 101 is a visible light camera, and person tracking module 182 is configured to perform face recognition on the captured images and determine the location of faces in scene 170. A thermal camera (e.g., a paired IR camera with zoom optics 102) is aligned and calibrated with the visible light camera to identify the region where faces are located in the thermal image. In various embodiments, the field of view of the visible light camera is at least as large as that of the paired thermal camera. Alignment of the captured images is facilitated by accurately calibrating differences in pointing error, rotation, field of view, and distortion between the two cameras. Temperature measurements can be made more reliable by limiting measurements in the thermal image to locations known to be faces. In some embodiments, cases where faces are obscured or where heated areas that are not faces can be rejected.

[0062] In some embodiments, the thermal camera is calibrated to a known temperature and can determine the absolute temperature of scene 170. Both cameras can also be (factory) calibrated to allow pixels in one camera to be mapped to pixels in the other. In some embodiments, parameters to be determined relative to the cameras include pointing difference (translation / tilt), rotation difference (roll), FOV difference (zoom), and / or distortion difference. These differences can be determined during factory calibration. Furthermore, the distance between the two cameras (parallax) can be a parameter used by the system for calibration and, in some embodiments, is configured to be as small as possible.

[0063] The person tracking module 182 is configured to detect the size and orientation of the face, as well as other varying factors such as glasses, masks, beards, etc. From the identified and identified face, the corresponding location in the thermal image can be inferred using the calibration terms mentioned above. In the thermal image, temperature can be measured at the target location where the face is known to be in position and where measurements of the face would produce the most accurate temperature. Visual face recognition can be used to avoid mismeasurements of objects that are close to human skin temperature but are not actually human. It can be used to track people passing through the field of view and measure skin temperature at the moment when the expected orientation of the face relative to the camera produces accurate results.

[0064] Other sensing components 142 may include devices that relay information to processing component 110 via wireless communication. For example, each sensing component 142 may be adapted to receive information from satellites via local broadcast (e.g., radio frequency), via mobile or cellular networks and / or via information beacons in infrastructure or various other wired or wireless technologies.

[0065] In various embodiments, the thermal imaging system is configured to achieve high accuracy and reduce measurement errors using one or more systems and methods described herein, taking into account factors such as: distance from the target to the thermal camera, capturing and analyzing high-resolution images of the target (e.g., systems and methods for increasing the number of pixels on the target), identifying the area on the target to be measured, the time for measuring and tracking the target, on-site system calibration, the presence of a blackbody in the field of view or radiation measurement calibration, approaching the area of ​​interest on the target (e.g., the tear duct), the time for temperature stabilization after entering the field of view (e.g., if the target enters from the outside, the target temperature may need time to adjust to a new setting), and / or other factors.

[0066] refer to Figure 1C Example operation of a thermal imaging system (e.g., thermal imaging system 100) will now be described according to one or more embodiments. The systems and methods described herein offer improved reliability and accuracy compared to conventional techniques for detecting elevated body and / or skin temperatures. In some embodiments, the technique is applied person-to-person (e.g., relative to other people) to improve performance compared to absolute temperature comparisons, which fail to account for variable environmental conditions and the average temperature of people in the scene. In some embodiments, the elevated body temperature detection described herein does not require calibration with a blackbody in the scene, but a blackbody can be optionally used when needed, which improves deployment flexibility compared to conventional systems. The systems and methods described herein can compensate for several factors, including distance from the camera and person-to-person differences.

[0067] In the illustrated embodiment, process 190 begins by capturing and processing thermal images of the scene in step 191. Thermal images can be captured in real time to monitor crowds and / or captured and stored for later processing. In some embodiments, other data is also captured, such as visible light images, audio, time, date, location, temperature, and other data that can be aligned with the thermal images to aid in identification of people. The thermal images are captured by a thermal camera, which can be deployed at points of crowd flow obstruction and / or other locations suitable for capturing thermal images of people for the temperature measurements described herein.

[0068] In step 192, the captured thermal images are processed to identify people of interest. People within the images are detected and tracked, and the system is configured to identify one or more tracked people for further temperature detection processing. In some embodiments, motion statistics between frames of the captured thermal images are used to identify people moving in the scene. In other embodiments, a trained convolutional neural network (or other machine learning module) is configured to receive one or more thermal images and identify people in the images. The system can also be configured to identify the location of a person in the image (e.g., identify a bounding box around the person), the size of the person, and then individually select one or more identified people for further processing to determine the person's temperature. The system can use automatic person recognition, where people can identify themselves by interacting with a public phone booth, user-guided recognition can be used to guide the system to focus on a specific individual, and / or other human identification techniques can be used.

[0069] In step 194, the system processes one or more thermal images to determine temperature data associated with a person of interest. In some embodiments, the system uses the image captured from step 191 to measure the user's temperature. In other embodiments, the system tracks the person of interest to identify opportunities for thermal image capture, such as a visible forehead in an image and / or eyes in an image. The system may be instructed to capture an image, zoom in on the person and capture the image, and / or perform other image capture steps. After capturing the thermal image, the thermal image is processed to identify the location for measurement and determine the temperature at the measurement location. In some embodiments, step 194 includes operating a neural network to identify people in a crowd and determine the measurement location (e.g., forehead) associated with each individual. The system may then perform alignment zoom (e.g., optical zoom, digital zoom, etc.) on the measurement location and capture a new thermal image for measuring the person's temperature.

[0070] In various embodiments, the system is configured to process one or more thermal images and / or other data to determine and / or estimate the core temperature of a person of interest. This processing may include detecting skin temperature and using it to determine the person's core body temperature. For example, skin temperature may stabilize within 4-5 minutes, thus providing a correlation with core body temperature. In some embodiments, thermal images are captured using a high-resolution thermal camera configured with an accuracy range within 0.1 degrees 80% of the time and within 0.2 degrees 100% of the time, allowing the system to accurately detect minute changes in person-to-person temperature. In various embodiments, the scene is processed as a blackbody for measurement using a low-cost process.

[0071] In some embodiments, the system uses multiple measurement points to estimate core body temperature. Automated algorithms can be used to update the temperature distribution. Other measurement techniques may include user temperature offset functionality to improve measurement accuracy, and / or system drift for stereo thermal imaging and monitoring of depth maps (e.g., distance compensation). Various camera and processing parameters can be adjusted or compensated to improve temperature accuracy, including target distance, image resolution, area on the target to be measured, time on the target, on-site calibration, calibration using blackbody or radiation measurements, specific regions of interest (e.g., forehead, tear ducts), time to target temperature stabilization, etc. In various embodiments, parameters are processed to extract features for input into a neural network trained to recognize core body temperature and / or other temperature-related data.

[0072] Next, in step 195, the system evaluates the temperature data associated with the person of interest and classifies the individual. In various embodiments, the temperature data is evaluated to determine whether the person of interest has an elevated body temperature. In some embodiments, direct readings from a thermal camera at the measurement location are used. In other embodiments, the system evaluates other available data to obtain a more accurate classification result. For example, the system may process captured thermal images to determine the average temperature measurement of people passing through the scene (e.g., adjusting subsequent thermal image processing of the person of interest in response to temporal changes in temperature associated with environmental conditions). The system may be configured to group people according to criteria such as age (e.g., young children, older adults, etc.) to provide an accurate assessment of the average temperature for each group. For example, grouping may be based on a facial recognition process to determine approximate age. Temperature data may be compared to a baseline, adjusted based on other data and criteria, processed via a neural network, and / or other processing may be performed.

[0073] In various embodiments, the system compares the estimated core body temperature and / or measured skin temperature (or other point temperature, such as eye temperature) of the person of interest with the average temperature of statistically similar other people. In some embodiments, a neural network can be trained to classify detected people into groups (e.g., by age range, sex, size). Each group of people can have its own temperature model, which allows the system to adjust the temperature data in real time to limit the number of false positives and / or false negatives, thereby obtaining a more accurate model. The measurements of this group adjustment can be combined with other adjustments described herein, such as the movement path of the people being tracked through an area monitored by the system (e.g., temperature changes as people enter a building) or changes in temperature over time. In some embodiments, the temperature data model can be provided as input to the trained neural network and / or statistical analysis process to determine the adjusted temperature of the person. In various embodiments, the system is trained / configured to have an appropriate increment between a baseline temperature determined based on EBT (e.g., based on the average value of people in the scene, distance, grouping of people such as age range, movement path, etc.) and the EBT determination to avoid false positives or missing people with EBT (see, e.g., Figure 1D In some embodiments, the process includes isotherm data of the scene in conjunction with an alarm threshold.

[0074] In step 196, the person of interest is then processed based on the assessment. For example, if the person has a normal temperature, information about the individual can be stored for further statistical analysis, system training, or other processing. If the person is determined to have an elevated body temperature (EBD), further processing can be performed based on classification. In some systems, this information is stored for further analysis. In other systems, the system's users are notified of the EBD, and the person of interest is directed to another location for subsequent health screening and actions, such as disease testing, quarantine, preventing the person of interest from entering crowded areas, and / or other actions.

[0075] Continuous thermal imaging for virus / infection detection

[0076] Pandemic response is a tremendous challenge for governments and societies. Infectious diseases can disrupt businesses, schools, travel, public events, and other activities. If any, before a vaccine is developed, infection could spread through community growth for months or years. High-density deployment of surveillance equipment can be used to quickly identify people with fever and rapidly isolate them from the population before they can infect others. (Reference) Figure 2A Security personnel 204 can use monitoring devices such as thermal cameras 202 and interactive user displays 203 to quickly screen passengers 206 for elevated temperatures indicating possible fever. Handheld cameras 210 can also be used to provide high-resolution thermal imaging 212 of the skin.

[0077] In conventional systems, thermal cameras are typically integrated at security checkpoints using a person-in-the-loop approach, which limits screening to one person at a time by a security personnel. These methods create significant congestion within the crowd, potentially leading to more direct contact between people that could transmit the disease. Many diseases can spread across communities along with routine contact via airborne contamination, and using simple checkpoints and individual screening is not an effective way to manage and mitigate virus transmission.

[0078] In various embodiments of this disclosure, the improved thermal camera platform incorporates novel machine learning methods for screening individuals in large populations. The platform can wirelessly connect to a monitoring network to rapidly detect new epidemic activity, thereby providing early intervention to prevent the growth and / or resurgence of infectious diseases. Machine learning is used to more efficiently identify people in real time within a scene and track their surface temperatures in large groups to identify fever as an early indicator of illness among individuals in the population.

[0079] The thermal camera platform can have a shape factor similar to that of a ceiling or wall-mounted smoke detector or thermal security camera. Figure 2B Examples of thermal imaging camera shape factors are illustrated, including thermal security camera 220 and thermal security camera 230, which can be used to capture infrared images of crowds for fever screening, as shown in image 222. In some embodiments, the thermal camera platform includes a wide field-of-view (FOV) infrared (IR) microbolometer camera that provides 360-degree imaging for identifying people in a wide scene by running a neural network algorithm to identify and track people in the scene. These shape factors support large-interval distances and real-time large-area monitoring. When a person is identified by the thermal camera, a paired IR camera with zoom optics focuses on the target and performs a series of temperature measurements, which are analyzed by an algorithm to determine the number of data points required over time and set a deviation threshold to trigger an alarm. The system then reports to the monitoring personnel that the subject should be isolated for further diagnostic evaluation by a medical professional. Reference Figure 2C The thermal camera can be placed in a hospital waiting room 240 and will issue an alarm signal (image 244) on a display 250 when it detects a person with a fever. For example, the thermal camera can also be deployed in office environments 242 and manufacturing facilities 246.

[0080] Machine Learning for Virus / Infection Detection

[0081] Various aspects of this disclosure can be implemented using trained neural networks and / or other machine learning processes, including analyzing captured images to detect and locate people, identifying measurement locations on people, determining a person's core body temperature and / or determining if an individual has a fever. Reference will now be made to... Figure 3A-D is used to describe embodiments of neural network systems and methods that can be used in this disclosure.

[0082] refer to Figure 3A This document describes embodiments of a virus detection system. The virus detection system 300 may be implemented on one or more servers, such as application servers that perform data processing and / or other software operations for training, storing, and using neural networks employed by the virus detection system 300. In some embodiments, components of the virus detection system 300 may be distributed across a communication network (e.g., cloud / network 322). The communication network 322 may include one or more local area networks, such as wireless local area networks (WLANs), wide area networks (e.g., the Internet or cloud networks), and other wired or wireless communication paths suitable for facilitating communication between the components described herein. The virus detection system 300 includes a communication component 314 operable to facilitate communication with one or more local virus monitoring systems 320 via the communication network 322.

[0083] In various embodiments, the virus detection system 300 may operate as a networked virus detection system, such as a cloud-based virus detection system, or it may be configured to operate in a dedicated system, such as a virus surveillance system that processes thermal images and other data captured in real time from one or more virus surveillance devices (e.g., thermal imaging cameras described herein). The virus detection system 300 may be configured to analyze the captured data and return information about virus identification (e.g., an alert with an identifier of an individual detected as having a fever). The virus detection system 300 may also include a database 302 for storing the captured data, training datasets, and other information.

[0084] As shown in the figure, the virus detection system 300 includes one or more processors 304 that perform data processing and / or other software execution operations. The processor 304 may include logic devices, microcontrollers, processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other devices that the virus detection system 300 can use to execute appropriate instructions, such as software instructions stored in memory 306 for a training system 308 for virus detection, other data processing algorithms 310, and a trained virus detection neural network 312 (e.g., a convolutional neural network trained using a training dataset stored in database 302) and / or other applications.

[0085] Memory 306 may be implemented in one or more memory devices (e.g., memory components) that store executable instructions, data, and information (including image data, video data, audio data, and network information). Memory devices may include various types of memory for information storage, including volatile and non-volatile memory devices such as RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Read-Only Memory), flash memory, disk drives, and other types of memory described herein.

[0086] Each local virus surveillance system 320 can be implemented as a computing device, such as a thermal imaging camera, a handheld non-contact temperature sensing device, a desktop computer or web server, a mobile computing device (e.g., a mobile phone), a tablet computer, a laptop computer, or other computing devices having communication circuitry (e.g., wireless communication circuitry or wired communication circuitry) for connection to other devices in the virus detection system 300. In some embodiments, the local virus surveillance system 320 may include one or more unmanned vehicles (e.g., drones), such as unmanned aerial vehicles, unmanned ground vehicles, or other unmanned vehicles. For example, unmanned vehicles can be deployed to monitor a location while limiting operator transmission, and the unmanned vehicle may be configured with temperature sensors and processing systems to detect people and identify individuals with elevated body temperatures. In some embodiments, the drone may include one or more speakers for providing instructions or information to nearby people, distance sensors (e.g., lasers), GPS components for determining location, navigation components, communication components for communicating with the host system and / or the operator, and other components as needed. Drones can be used to identify infected individuals in public locations, for surveillance and quarantine, and in other scenarios where fixed and / or handheld temperature monitoring is not feasible.

[0087] Communication component 314 may include circuitry for communicating with other devices using various communication protocols. In various embodiments, communication component 314 may be configured to communicate via a wired communication link (e.g., via a network router, switch, hub, or other network device) for wired communication purposes. For example, the wired link may be implemented using power line cable, coaxial cable, fiber optic cable, or other suitable cable or wire supporting the corresponding wired network technology. Communication component 314 may also be configured to connect to a wired network and / or device via a wired communication component such as an Ethernet interface, power line modem, digital subscriber line (DSL) modem, public switched telephone network (PSTN) modem, cable modem, and / or other suitable component for wired communication. Communication component 314 may also support proprietary wired communication protocols and interfaces.

[0088] One or more trained virus detection neural networks 370 can be implemented in a remote real-time environment, such as... Figure 3B As shown. Virus detection system 350 may include a thermal imaging camera or other device or system operable to receive and / or generate thermal images and process the received thermal images using one or more detection processes described herein. In the illustrated embodiment, virus detection system 350 includes a processor and memory 360 operable to store one or more trained neural networks and implement a neural network runtime interface 370 thereon.

[0089] In various embodiments, a training dataset containing virus detection information can be used to train one or more neural networks and / or other machine learning algorithms for use in a virus detection system. (Reference) Figure 3C An embodiment of the neural network training process will now be described. In one embodiment, neural network 380 is a convolutional neural network (CNN) that receives a training dataset 382 including up-to-date virus detection information and outputs a classification of the data. This disclosure describes various neural networks that can be trained for one or more determinations in a virus detection system, including but not limited to detecting and tracking people in a crowd from visible light and / or thermal images, detecting the location of temperature measurements on an individual, detecting the individual's temperature (e.g., core body temperature, skin temperature, etc.) and / or determining whether the individual has a fever or not.

[0090] In some embodiments, the training dataset 382 includes data simulating real-time input to the virus surveillance system described herein, and may include real data captured from infrared, visible light, or other types of cameras, as well as other sensor data captured from the environment. The data may also include synthetic data generated to simulate real-time sensor data. In one embodiment, training begins with a forward pass through a neural network 380, which includes feature extraction, multiple convolutional and pooling layers, multiple fully connected layers, and an output layer including the desired classification. The CNN parameters can then be updated using a backward pass through the neural network 380 to account for errors generated during the forward pass (e.g., misclassified data compared to the label). Other training procedures may be used according to this disclosure in various embodiments.

[0091] Figure 3DAn embodiment for validating a trained neural network is illustrated. A set of fully annotated validation test data 392 representing real-world data is fed into the trained neural network 390. The validation test data 393 includes various thermal image data and sensor data for classification. Detected errors (e.g., image misclassification) can be analyzed and fed back to the training system to update the training model, which in turn updates the training dataset 382. In various embodiments, detected errors can be corrected by adding more data examples (e.g., more types of environments), thereby increasing the resolution of the data and / or increasing the accuracy of thermal modeling to help distinguish data types. By adjusting the training dataset to improve accuracy in real time, operators can avoid costly delays in implementing accurate virus detection and surveillance systems.

[0092] In various embodiments, the elevated body temperature detection system is designed to operate using one or more trained CNNs. In some embodiments, measurement errors are reduced and the detected temperature is stabilized by using a CNN-based face tracking, translation, tilt, and zoom (PTZ) camera, a high-resolution infrared camera with optical zoom, and / or other systems and methods for keeping thermal image pixels on the target face (e.g., keeping the number of pixels in the measurement area above a minimum threshold).

[0093] In some embodiments, the CNN is trained to identify the optimal measurement location within captured pixels of the tracked person, which may depend on the area of ​​skin and / or body available for measurement. The CNN localization is determined as the area best available to the system for measurement accuracy, such as the target's forehead, tear ducts, and / or other locations.

[0094] In more general applications, such as hotel lobbies, restaurants, shopping malls, and other high-traffic areas, a CNN is trained to determine the distance from the thermal camera to the person being tracked, and the temperature measurement is automatically adjusted based on the target distance from the camera. In cases where person tracking spans one or more camera fields of view, the system can also be configured to monitor a person's body temperature, as it changes over long periods. In these embodiments, the system can track each identified person and maintain the quality of the measurement metrics. In some embodiments, the system is configured to identify the measurement location (e.g., eyes, skin), the number of pixels (or a very sufficient number of pixels) at the target measurement location, the distance from the camera, the time on the target, and / or other parameters. In some cases, the average skin temperature of the crowd can be determined from an image, which is determined to have high-quality parameters that make the temperature measurement sufficiently reliable.

[0095] In many applications, the monitored person is moving through a scene, and the CNN is trained to recognize times when conditions are favorable for measurement (e.g., a face looking directly at the camera, a face without glasses or a mask, etc.). CNNs can also be used to identify people who do not follow accurate measurement procedures, such as those who do not remove their glasses or hats as instructed. In some cases, the system can be configured to notify the target when a person is not provided with a view containing sufficient measurement data. For example, the system may include lasers, directional sounds, or other markers to prompt the target to look towards the thermal imaging system, allowing the thermal imaging system to capture thermal images for more accurate temperature measurements.

[0096] In various embodiments, the system is configured to store field-generated data for further analysis and / or training of one or more neural networks. For example, data from the deployed system can be fed back into the CNN training process to improve the CNN for specific environments (e.g., detecting the temperature of people in airports, people attending public sporting events, etc.), desired classification objectives (e.g., training the CNN to detect temperature / symptom profiles associated with certain infectious diseases), and / or for more accurate performance. In some jurisdictions, privacy regulations may prevent / restrict data sharing for this purpose.

[0097] refer to Figure 3E -G, Other embodiments of CNN training systems and methods will now be described. As disclosed herein, temperature screening systems use one or more thermal cameras to measure a person to check if they may be in a fever state. In some embodiments, facial recognition is used to identify faces at different angles and / or distances and then find the desired location to measure the person's temperature. For example, in many systems, accurate corner-of-the-eye temperature measurements are used for reliable results.

[0098] In various embodiments, a CNN trained on thermal data is combined with a tracker to provide a rich set of information, including face detection, facial landmarks (e.g., corners of the eyes), head pose, and facial attributes (e.g., glasses, masks, hats, etc.). This additional information is used to generate more accurate measurements and reduce system complexity. For example, systems and methods in various embodiments use face detection to identify the person to be measured. Facial landmarks are used to locate the corners of the eyes, which are used to accurately measure elevated skin temperature and for fever screening. By tracking head pose, the system can identify when the person of interest is looking at the camera to validate the measurement. Head pose is used to validate the measurement by ensuring that the person of interest is correctly “looking at the camera.” Attributes such as masks and glasses are further used to validate the measurement and check for occlusion of the corners of the eyes. By using pose and face detection, the system can additionally estimate the distance to the target. In turn, the elevated temperature system can correct for the influence of distance for higher temperature measurement accuracy.

[0099] refer to Figure 3E The thermal training image 400 is annotated for face detection, including the identification of facial feature points and attributes. For example, the thermal image 400 includes various annotations identifying head pose attributes, such as bounding boxes identifying the location of the person or object in the image 400, such as bounding boxes identifying the location of "person," the location of "face," and the location of "mask." It may also include annotations identifying various facial attributes that provide head pose information, such as annotations identifying the location of a person's chin, the side of the mouth, the nose, and the inner and outer positions of the left and right eyes in the image. Images can be collected in the field, developed through a testing system, generated comprehensively, and / or collected through other methods. Annotations can be manually verified via a user interface 410, which includes a list of available annotations and associated location identifiers, such as boxes, points, graphs, icons, or other identifiers that can be placed on the image 400. The trained CNN can be integrated into a tracker system for detecting elevated skin temperature and / or body temperature.

[0100] In some embodiments, one or more CNNs are trained to provide face detection. By recognizing faces within an elevated skin temperature system, the system is configured to ensure that measurements are taken only when a person is present in the thermal image. This helps reduce false positives caused by other thermal objects (hot cups) in the scene that have a similar temperature to a person with a fever. Furthermore, this facilitates measurement consistency, person counting, multi-person measurements, and temperature measurement in scenes where stationary objects are not required.

[0101] One or more CNNs provide localization of various feature points on the face (e.g., ...). Figure 3E (As shown). In some embodiments, these feature points are used to identify measurement locations, such as the corners of the eyes on the face. The corners of the eyes are areas on the face that are associated with a person's core body temperature and, consequently, whether a person has a fever. Using the corners of the eyes feature points, the region of highest temperature of interest is identified and used for temperature measurement. The elevated temperature system can then use this measurement to determine whether a person has a fever according to the system's decision rules. Other benefits of the feature points described herein include, but are not limited to, not requiring a stationary subject, and reducing measurement variance / false positives due to potential hotspots on the face (e.g., a hot summer day).

[0102] CNNs, combined with feature trackers, can be used to provide head pose information. (Reference) Figure 3F Image 420 shows the use of feature points (e.g., corresponding to...) Figure 3EThe system identifies faces by recognizing feature points. It captures images of the scene and processes them using a trained CNN to identify the location of faces and each tracked feature (if visible). The system checks whether the tracked person has looked at the camera, at least in part, based on the relative position of feature points with respect to the camera. For example, image 422 shows a captured image of a face looking down, and image 424 shows a face looking at the camera. In various embodiments, the head position is measured and compared to a certain pitch, yaw, and roll threshold to determine when the head was positioned for accurate temperature measurement during image capture. If the person is looking away (e.g., image 422), the system can mark the measurement as invalid. If the person is looking directly at the camera within the threshold (e.g., image 424), the system can mark the corresponding temperature measurement as valid. In some embodiments, the method can be used to automatically prompt the person to look correctly at the camera, thus avoiding situations where the corners of the eyes are obscured due to head posture. For example, based on calculated pitch, yaw, and / or roll, the system can prompt the person to adjust their / her head orientation towards the camera. Furthermore, angle corrections can be applied to the measurements to obtain better absolute accuracy.

[0103] In various embodiments, the CNN provides attributes such as mask / no mask, glasses, eye occlusion, etc. Eye occlusion / glasses is used to invalidate the measurement and prompt the person to remove the occlusion (e.g., glasses). Mask / no mask can potentially be used to calibrate temperature measurements, as a potential source of variance in the measured temperature may be due to the mask. Furthermore, it can be used to verify whether people entering a building are wearing the required PPE.

[0104] Based on the above information, the tracking system estimates the distance from the camera to the subject. In some embodiments, the training data includes training images of the subject at different distances, such as... Figure 3G As shown, Figure 3G Images 450 depict multiple subjects 460 and multiple distances 470. Each subject 460 may have a temperature measured at different distances 470 from the camera, resulting in different measured temperatures, which are then adjusted based on the distance. In various embodiments, a CNN can be trained to determine the distance from the camera to the subject, and / or known properties of the camera and system configuration, as well as knowledge of the scene (e.g., distance to the monitored entrance), can be used to determine the distance.

[0105] refer to Figure 3H An example process 480 for detecting an elevated temperature in a subject will now be described according to one or more embodiments. In step 482, a neural network is trained to track facial feature points, such as those described above. Figure 3A to 3GThe facial feature points described. In various embodiments, facial features may include bounding boxes identifying a subject (e.g., one or more people), bounding boxes identifying facial features or objects (e.g., face, mask, etc.), and the locations of multiple facial features (e.g., chin, eyes, nose, mouth, etc.). In step 484, a trained neural network is used to process the captured scene image to identify the subject and facial features. In step 486, the system measures the temperature at a region of interest associated with one or more facial feature points (e.g., corners of the eyes), which may include an estimated distance to the face, enabling potential correction for the temperature measurement at the corners of the eyes. In step 488, the system measures head pose orientation based on the detected facial feature points. In some embodiments, the system employs a nominal 3D facial model and matches it to feature points and pose, measuring yaw, lateral, and / or pitch to determine if the subject is facing the camera. In step 490, the system compares the measured head pose orientation to one or more thresholds to validate the temperature measurement. If the head pose orientation is within one or more thresholds, the temperature is validated. Otherwise, the temperature measurement may be discarded. Furthermore, the system can validate temperature measurements by checking if a person is too far or too close to the camera, and can adjust the temperature measurement based on estimated distance and / or head posture orientation. By using a CNN on the thermal image instead of the visible image, the temperature enhancement system does not need to correct for parallax when attempting to locate the corner of the eye in the visible image and transfer that location to the thermal image, thus reducing measurement variance.

[0106] Systems and methods for raising body temperature

[0107] Various embodiments of one or more aspects of improving systems and methods for raising body temperature will now be described in more detail with reference to the remaining figures.

[0108] 1. Systems and methods for improving measurement stability

[0109] In one or more embodiments, the usage scenario (e.g., Figure 1B Scenario 170) is used as a "blackbody" to achieve improved measurement stability. (Reference) Figure 4 The process may include selecting multiple random or predetermined measurement locations within a captured thermal image of the scene in step 502. Next, in step 504, local contrast is measured while excluding points with high contrast (e.g., pixels that may indicate strong edges). For example, high-contrast points can be determined by comparing the contrast of adjacent pixels and excluding measurement locations where the contrast exceeds a threshold contrast value. In some embodiments, by selecting a new location for each excluded location, the number N of selected pixels can be kept constant, resulting in a minimal set of points with low spatial contrast in the identified scene.

[0110] Then, in step 506, the signal at each of the N points is measured over time. For example, measurements can be taken at regular time intervals T, and the collected values ​​can be stored in memory. Next, in step 508, points that exhibit significant signal changes over time (e.g., signal standard deviations exceeding a threshold) are excluded from the set of points used to improve stability. In some embodiments, the number of locations N can be kept constant by selecting new locations for each excluded location, thereby identifying a minimal set of points with low temporal variation in the scene.

[0111] Then, in step 510, the system monitors and compensates for drift in the measurement. In some embodiments, the average or median of N stable points is calculated to monitor drift in the measurement. For example, high temporal frequency drift can be eliminated by subtracting a high-attenuation aggregate value of N samples from the sampled signal values ​​in the image. The aggregate value can be the average / median of the samples used or other weighted combinations. Spatial and temporal stability can be continuously monitored to identify outliers or pixel samples that are frequently occluded by people passing through the scene. Slow temporal drift can be allowed to compensate for temperature variations throughout the scene.

[0112] In some embodiments, a black label can be applied to a surface in the scene that is at ambient temperature. The location of the black label can be used as one of the predetermined measurement locations described above. The surface to which the label is attached should not be a surface that is significantly heated by any temperature other than ambient air temperature. Figure 5A As shown, the label can have a high in-band emissivity for the camera used; for example, it might be a label like the Acktar Black. The label serves as a reference point for the ambient temperature in the scene. One drawback of applying the label to a surface with high thermal conductivity is that it may not be able to track temperature changes as well as a thermal camera / lens combination. One solution to this is to use a spacer ring under the label to create an air gap beneath it. This is similar to a microbolometer element, or a sensor on a rattlesnake or viper, where a thin film is stretched across an opening in the snake's face to reduce its thermal contact with the snake. The label will then have a lower effective thermal mass and track the air temperature with a thermal time constant closer to that of a thermal camera / lens combination. The time constant of the label can also be customized by applying it to an aluminum disk. The thicker the disk, the higher the time constant, which can be determined experimentally. Labels applied directly to surfaces in the scene can have different time constants depending on the thermal conductivity of what the label is attached to.

[0113] In some embodiments, additional stabilization can be achieved using image processing algorithms that separate the background from the person. This can be performed using various techniques, including image analysis or spatial analysis to identify objects in three-dimensional space, learning by identifying and removing the background, etc. The background can be used to stabilize the signal in a shorter time, especially if periodic manual calibration is required.

[0114] refer to Figure 5B An example blackbody is described according to one or more embodiments, which can be used to improve the accuracy of radiation measurements in thermal imagers with low system cost and complexity. System 550 includes a thermal camera 560 configured to capture thermal images of scene 562 (e.g., thermal image 564 including images of object 566 and blackbody 568). Scene 562 includes a passive blackbody 570 incorporating the thermal camera 560. In various embodiments, the passive blackbody includes a low-cost passive blackbody used as a reference for radiation measurements within the scene. Temperature sensor 572 senses the ambient temperature at the blackbody and includes a wireless interface 574 for transmitting sensor information to the thermal camera 560 via a telemetry interface.

[0115] In system 550, the accuracy of remote temperature measurement using thermal imaging camera 560 is improved because system 550 provides an accurate reference temperature for blackbody 570. Thermal camera 560 captures a thermal image 564 of the scene, which includes an image 568 of blackbody 570, and uses this portion of the thermal image to provide an accurate reference relating pixels to temperature. This system is efficient and does not require expensive active blackbody and / or knowledge of the actual blackbody temperature (e.g., the actual blackbody temperature can vary due to preheating time and environmental influences). The illustrated embodiment replaces the active blackbody of a conventional system with a passive, non-powered blackbody that provides a highly accurate measurement of the blackbody temperature using electronic or other methods, and then transmits this temperature using a telemetry interface (e.g., Bluetooth) to transmit the blackbody temperature to the radiometric thermal imaging sensor in real time.

[0116] In one embodiment, the passive blackbody 570 comprises an aluminum plate with a high emissivity coating. The temperature sensor is one or more low-power, high-accuracy solid-state temperature sensors whose digital output is in thermal contact with the observed surface. The connection to the thermal camera 560 is a low-power wireless link (e.g., Bluetooth), through which the blackbody temperature is transmitted. The temperature sensor 572 and the wireless interface can operate for extended periods, for example, using a lithium coin cell battery and / or using solar energy or other low-power methods.

[0117] In some embodiments, system 550 may also include a collapsible bellows that protects the blackbody from airflow that could cause temperature fluctuations or changes on the surface. For example, the same bellows could be used to make the blackbody stand upright in a scene.

[0118] 2. Classify people using motion statistics from measurement points.

[0119] One challenge in fever monitoring systems is correctly identifying points in captured images as part of a human face. This disclosure describes an improved thermal imaging system with built-in analysis capable of correctly classifying sampled data points as part of a face. The method described herein provides more efficient and accurate face detection than conventional face detection algorithms.

[0120] Knowing that the collected temperature values ​​come from the face is important for many systems so they can automatically update population mean / median statistics to determine if a person has a higher than normal skin temperature. Currently, collecting mean / median values ​​is a manual process requiring user interaction and is prone to errors. User fatigue can also hinder demographic updates, thus requiring automated collection.

[0121] In one or more embodiments, the trajectory of the maximum value sampled is analyzed. A person moving through a scene typically follows several possible motion paths. They walk toward the camera or along a path determined by the physical layout of the scene and the camera's mounting position through the FoV. A thermal camera can detect the highest temperature in or within a region of interest (ROI) in the scene. If a person is moving and the surface temperature of the person's face is higher than the background, the location of the highest "temperature" pixel will follow a recognizable path in the pixel's XY plane. For example, if no one is present, the maximum value may be very stationary and may have a value lower than typical values ​​measured on the skin of a human face. If the maximum value location "jumps" from one location to another in an unstable pattern, this could also be an indication that the maximum measurement does not belong to the face of a single person in motion. These values ​​may also be excluded from demographic data collection.

[0122] refer to Figure 6 According to one or more embodiments, an example process 600 for identifying samples to be tracked and measured will not be described. First, in step 602, the system determines the trajectory of the largest sample value. Next, in step 604, the system compares the trajectory with one or more trajectory models. In step 606, the system collects sample values ​​that follow a predefined motion trajectory model. For example, the system may specify to collect only sample values ​​that follow a predefined motion trajectory model. In this way, people from different locations in the scene who have been exposed to different environmental thresholds and therefore may be expected to have different skin temperatures than the group the system is designed to monitor at that location are excluded.

[0123] Alternatively, two or more motion trajectories can be defined, and separate statistics can be collected for these groups. For example, this could be people entering or leaving an area. This allows the system to have two (or more) separate alarm levels based on the sample trajectories. In step 608, the system detects alarm conditions for sample values ​​that follow the corresponding trajectory model.

[0124] 3. Estimate actual core body temperature using multiple measurements over time.

[0125] Now refer to Figure 7 The diagram illustrates an implementation for estimating actual core body temperature. It is well known that people exposed to cold outside air and those exposed to warm car air will have very different skin temperatures, even if their core body temperatures are similar. This difference is expected to decrease over time as people are exposed to the same environment (e.g., waiting inside an airport or train station). By tracking the same individual (e.g., through tracking algorithms or through image-based recognition and re-recognition), a surveillance system can monitor the skin temperature in an individual's face over time.

[0126] By measuring the rate of change of skin temperature over time, the system can estimate the asymptotic value of a person's eventual "steady-state." This allows the system to compare people from different environments and will reduce the number of false negatives from people in colder environments and the number of false positives from people in warmer environments. Samples collected over time may come from the same thermal imager (e.g., if people are in a queue and therefore exist in FoV for a long time), or samples may come from different thermal imagers at different locations along the movement path of the person being monitored. For example, a person could be tracked at an airport entrance and then followed as that person attempts to check in or go through security.

[0127] Figure 700 illustrates that if the system extrapolates to future times, the large temperature difference at the time of the first sampling can be reduced to a very small difference. For example, as facial temperature adapts to a new environment, the system can model the rate of change of facial temperature, which can then be used with a minimal number of samples (e.g., only two samples) with a known time interval between them to estimate the “true” core body temperature or the “true” facial temperature.

[0128] 4. Use a three-dimensional thermal system to monitor system drift.

[0129] refer to Figure 8According to one or more embodiments, an example method for monitoring system drift for recalibration will be described. System 800 includes at least two thermal cameras, thermal camera #1 802 and thermal camera #2 804, that provide stereo thermal (e.g., for creating a depth map). Individual temperature measurements Temp1 and Temp2 are taken at point 812 in scene 810. Because the temperature measurements are distance-dependent, the temperature measurements are distance-adjusted (e.g., Temp1 is adjusted using distance D1 and Temp2 is adjusted using distance D2) and the temperature measurements are compared. If both values ​​are within an acceptable error threshold, the measurement is accepted. If the adjusted measurement is outside the error threshold, the system can be “marked” as needing recalibration or repair, and the measurement can be marked as erroneous.

[0130] 5. Focus deployment at existing crowd congestion points.

[0131] refer to Figure 9 According to one or more embodiments of this disclosure, an example airport millimeter-wave scanner with virus detection will now be described. Adding a thermal imager to a currently existing millimeter-wave scanning inlet, such as millimeter-wave scanner 900, will allow for focused measurements of a target face at a known or measurable distance to the target. The typical “hands raised overhead, palms out” posture required for millimeter-wave scanning provides the ability to capture the target in a known orientation, making it easy to focus / capture the facial / eye area (or other areas of interest: inside the wrist, armpit, throat – which can be used as a reference or indicating measurement). A camera (e.g., thermal camera 904) can be added as an accessory or integrated into the scanner.

[0132] In some embodiments, a thermal camera 906 may be placed at the entrance / exit of scanner 900 (e.g., on or near the outside of the scanner) to capture a thermal image of a person 902 from a known distance as they enter / exit the scanner. A blackbody or “black label” in the camera’s field of view (FOV) (as shown above) can also be used for continuous calibration. Most thermal transients associated with high / low external temperatures will be normalized before the target reaches the scan point, allowing for more accurate temperature measurements.

[0133] By deploying thermal cameras at existing congestion points, temperature scanning does not disrupt crowd flow because thermal images can be captured and evaluated in parallel with millimeter-wave scanning. Additional thermal cameras can also be deployed at other known locations, such as baggage check / registration kiosks or TSA ID / passport verification points.

[0134] 6. Use a low-cost powered blackbody to correct camera drift.

[0135] If camera digital data can be drift-corrected using a powered blackbody source in the scene, the relative surface temperature in the scene can be monitored using a thermal camera that is not calibrated for radiometric measurements. The blackbody operates at a constant temperature (which would be close to 38 degrees Celsius, or, if someone is hot, the temperature of their forehead). The blackbody source has high emissivity, so it can act as a stable radiation source with very low reflectivity. This source can be relatively small, in this case, about 1 inch × 0.5 inches. The requirement that the target's apparent size in the image should be at least 10 × 10 pixels for accurate temperature measurement is relaxed here, as we don't need to measure absolute temperature; we only need a stable radiation source. Since the blackbody operates at temperatures above ambient, it only needs to be heated—no TE cooler is required. A low-cost powered blackbody can consist of wire-wound power resistors thermally bonded to a heat sink made of aluminum with a high-emissivity black coating on the visible side. Wire-wound resistors in metal or ceramic housings are relatively inexpensive, and the housings have flat surfaces, making it easy to attach them to another flat surface, in this case, the heat sink. The heat sink can have a thermistor attached to it, close to the resistor housing. A very simple electronic circuit monitors the resistance of the thermistor and uses a power transistor circuit to adjust the drive current of the resistor. The blackbody will operate under controlled conditions at a very stable temperature under ambient temperature conditions. It is not necessary to operate the blackbody at a precise temperature value, only for it to be stable. By using a shroud around the front of the blackbody to improve temperature stability, airflow can be directed away from the visible surface. By eliminating the need for absolute temperature accuracy, the requirement for calibration is removed, resulting in cost savings due to reduced circuit complexity and less manual handling. Thermistors used for temperature control are inexpensive and interchangeable by fractions of a degree Celsius, so these sources should anyway be within ±0.5°C of each other.

[0136] In some embodiments, the blackbody is designed as a 5mm coated aluminum plate heated by a film heater. This offers the advantage of being larger and more uniform than a wire-wound resistor concept. The heater will be manufactured with a thermistor bonded to the back with epoxy resin, and this thermistor can be read by a control circuit that can also transmit its temperature to the camera system via Bluetooth or another interface. The aluminum plate may have a black label applied to it, which has a very high emissivity. This reduces the touch labor required for printing. The heater will be controlled by a PID controller with an adjustable setpoint, which can be set via Bluetooth from an app or via a micro-USB cable from a host PC. In one embodiment, the blackbody housing is designed as a plastic sphere with holes on the side. The camera views through the hole and observes a 1-inch recess in the plate within the sphere. The sphere is positioned on a pivot base so that the holes can be aligned with the camera. The recessed design makes the surface of the blackbody emitter less susceptible to airflow.

[0137] 7. Automatic algorithm for updating temperature distribution

[0138] An embodiment for updating the temperature distribution will now be described. An example algorithm is provided below:

[0139] 1. Initialize temperature distribution (initialize mean and variance. Fever distribution is higher than normal temperature!)

[0140] 2. Set up isotherm alarms covering all possible human body temperatures (including fever).

[0141] 3. Loop

[0142] 4. If an isotherm alarm is triggered, capture the image and send it to SW.

[0143] 5. SW performs facial detection

[0144] 6. If a face is detected

[0145] 7. Then measure the temperature

[0146] 8. If the temperature is above the threshold (or may be close to the fever distribution).

[0147] 9. Then an alarm will be triggered.

[0148] 10. Update the corresponding distribution (using a controlled learning factor).

[0149] 11. Sleep (over time)

[0150] 12. Go to 3.

[0151] In various embodiments, the system utilizes existing contrast metrics (for focused or unfocused faces), motion segmentation, visual skin color distribution (if applicable), and / or a combination of several algorithms / attributes to reduce false alarms.

[0152] To reduce / remove static objects within a scene, a Gaussian distribution-based mixture motion segmentation algorithm can be used. In such an algorithm, each pixel is modeled as a Gaussian distribution, for example, with a mean and variance. If a pixel is close to the distribution (e.g., located somewhere between the mean and variance), it is called background; otherwise, it is called foreground. This process adapts to the scene and marks static objects as background and moving objects as foreground.

[0153] Furthermore, by using isotherms, only pixels within the desired temperature range are considered foreground candidates. This process further reduces the number of errors in the scene. After the scene is segmented into foreground and background (e.g., a binary map with face pixel candidates), clustering size filtering can be applied to the binary map. This can be achieved by applying morphological filters (e.g., on and off (erosion dilation)). This process filters out small, isolated foreground pixels that are considered noise or false positive face pixels. After performing the aforementioned steps, the remaining foreground pixels are considered "face" pixels that can be sent to the face detection algorithm.

[0154] 8. Improved Elevated Body Temperature Scanning

[0155] In some systems, a measurement frame with the highest temperature is used, and the alarm threshold is based on the increment of the average temperature of the people measured by the system. The average typically needs to be updated manually by the operator via pressing a button for a certain number of reference people. To obtain a good average, this process may need to be performed hourly for multiple reference people in crowded areas. One problem with this approach is that users do not always reliably perform the process, resulting in a fixed average and a high probability of false positives and false negatives.

[0156] In various embodiments, more frequent updates to the average value allow the system to use stricter constraints, thereby reducing false positives and false negatives. In one approach, a selfie mode with flipped images is introduced to make it easier for test subjects to correct their position in the image (e.g., aligning with an outline or silhouette on the screen). In another embodiment, GUI support is provided in the screening mode, such as... Figure 10A As shown.

[0157] As shown in the figure, when the screening mode is enabled (as shown in image (A)), all irrelevant coverage is removed, and the outline of the head is displayed to help align the subject / subject / patient at the correct distance. Different outlines can be provided as needed to accommodate different types of people, different angles, different distances, etc. Additionally, a box covering the central portion of the face is enabled. As shown in image (B), the graphic display shows the current average sample size, incremental temperature, the current value of the measurement box, and the "Scan" button to perform the scan.

[0158] When screening mode is enabled, the user is instructed to obtain multiple samples (e.g., 10 samples). Samples can be obtained by clicking the "Scan" button or through other on-screen options (e.g., displaying the text "10 more scans needed"). As shown in image (C), the screen updates to show the remaining scans. After X minutes, the user is reminded to obtain a new sample, as shown in image (D). When an alarm is triggered, as shown in image (E), the overlay color changes (e.g., turns red, flashes / blinks, etc.) to alert the user. A beep or other signal may also be provided.

[0159] In other sampling methods, test subjects press an easily accessible button when they are ready to be measured, which updates the average value. Face detection algorithms for visual image streams, thermal image streams, or other streams can be used to detect faces and set measurement boxes in aligned IR images.

[0160] refer to Figure 10B and 10C The example frame and sample images used for training face detection will now be described. In this method, matched IR and visible light images can be used to automatically annotate IR faces detected from visual images. Process 1000 begins by capturing IR and visible light images in step 1002. In step 1004, a face detection algorithm is run on the captured visible light image, and if a possible face is detected (step 1006), a second face detection algorithm is run in step 1008. If a face with a discernible feature is detected (step 1010), the system sets the frame and measures the maximum temperature in step 1012. If the temperature measurement is within the confidence limits of the reference sample (step 1014), the measurement is used to update the average value in step 1016. Example images using this method are shown below. Figure 10C As shown.

[0161] 9. Isotherms combined with alarm thresholds in EBT

[0162] refer to Figure 11The image in question will now be described with reference to one or more embodiments to illustrate an embodiment using isotherms in conjunction with alarm thresholds in an EBT (Emergency Temperature Detection and Control). When users utilize the EBT function in a camera, they typically establish a set of average temperatures and set the camera to alarm when a person's temperature exceeds the average temperature by x degrees Celsius. This alarm is then displayed as an image overlaid with a red flash value on the left, as shown in image (A). However, if there are many people in the crowd, the operator may not be able to determine who triggered the alarm, especially if people are moving.

[0163] In some embodiments, isotherms are applied to the area where an alarm is triggered. Currently, isotherms require absolute temperature thresholds that are not suitable for EBT. In one embodiment, isotherms for EBT functionality are enabled and continuous. This can be implemented in the GUI, so when a user selects EBT mode, the user can enable the isotherms (along with configuring sounds, etc.). Image (B) shows the interface of EBT with isotherms when an alarm is triggered.

[0164] 10. Temperature offset to improve measurement accuracy

[0165] Some thermal cameras include the ability to perform temperature offset measurements to measure surfaces with known temperatures and use those temperatures to adjust the radiance in the image. This feature offers several advantages, including higher accuracy in temperature measurements and less non-uniformity compensation (NUC) processing. While NUC can be disabled by using the described temperature offset, resulting in a slow degrade in image quality, the radiometric performance remains acceptable for many applications.

[0166] In fever screening scenarios, using temperature offset can minimize camera measurement errors. In one approach, a fever screening solution averages a group of people through a manual process to obtain a baseline. This average is used to eliminate camera measurement errors and offsets caused by people coming from cold (or warm) areas (e.g., the outside). In certain specific cases, using temperature offset can reduce the need for averaging. For example, if cameras are placed inside buildings where people spend a significant amount of time. This could be used, for instance, to monitor lunch queues to see if anyone has a fever that day, or to monitor walkways in large buildings, including hospitals. Figure 12A The settings for this method are shown in the figure.

[0167] The temperature offset can be adjusted for both the presence and absence of a blackbody. In many cases, having a blackbody may not be ideal. An alternative solution is to use a device that measures the ambient temperature and sends that temperature to the camera. This device has the same temperature as the ambient temperature. This method is set up... Figure 12B As shown in the image.

[0168] Where applicable, the various embodiments provided in this disclosure may be implemented using hardware, software, or a combination of hardware and software. Furthermore, where applicable, the various hardware and / or software components described herein may be combined into composite components comprising software, hardware, and / or both, without departing from the spirit of this disclosure. Where applicable, the various hardware and / or software components described herein may be separated into sub-components comprising software, hardware, or both, without departing from the spirit of this disclosure.

[0169] The software according to this disclosure, such as non-transitory instructions, program code, and / or data, may be stored on one or more non-transitory machine-readable media. It is also contemplated that the software identified herein may be implemented using one or more networked and / or unnetworked general-purpose or special-purpose computers and / or computer systems. Where applicable, the order of the various steps described herein may be changed, combined into compound steps, and / or separated into sub-steps to provide the functionality described herein.

[0170] The above embodiments illustrate, but are not limited to, the present invention. It should also be understood that many modifications and variations are possible based on the principles of the present invention. Therefore, the scope of the invention is defined only by the appended claims.

Claims

1. A thermal imaging method, the method comprising: Identify targets in multiple infrared images; Temperature data associated with the target is obtained, at least in part, based on the infrared image; The temperature data is evaluated to determine the corresponding temperature category; The target is identified based on the temperature classification process; Select the first plurality of measurement locations from one or more captured thermal images; Measure the local contrast at the first measurement location and exclude measurement locations with contrast higher than the high contrast threshold to generate a second set of multiple measurement locations; The signal at each of the second measurement locations is measured over time; Exclude measurement locations whose changes exceed a change threshold over time, and generate a third, multiple measurement location; Monitor the signal at each of the third measurement locations; and Use monitored signals to reduce signal measurement errors caused by system drift.

2. The method according to claim 1, wherein: The target is people; The identification includes identifying the person and tracking the person on a subset of the infrared images; The acquisition includes identifying the measurement location of the person within a subset of the infrared images, and determining temperature data associated with that location using corresponding values ​​from one or more of the infrared images; and The assessment includes using the temperature data to calculate the person's core body temperature.

3. The method according to claim 1, wherein: The evaluation includes using a convolutional neural network (CNN) trained to classify the target as either a feverish target or a non-feverish target; and The process includes generating an alarm condition if the temperature classification indicates that the target has a fever.

4. The method according to claim 1, wherein, The acquisition includes analyzing thermal data acquired from the measurement location, environmental data associated with the camera location, crowd temperature data associated with multiple individuals associated with the camera location, and / or target temperature data over time using a convolutional neural network (CNN) trained to generate core body temperature.

5. The method according to claim 1, wherein, The identification includes analyzing the infrared image using a convolutional neural network (CNN), and the method further includes: The CNN is trained to recognize facial feature points, including the chin, mouth, nose, left eye, and / or right eye. The CNN is trained to identify the distance between the target and the image capture device; and The CNN is trained to recognize one or more objects in an image, including masks, glasses, and / or hats.

6. The method according to claim 1, wherein: The identification includes identifying one or more facial feature points of the target; The acquisition includes: identifying a region of interest based on the one or more facial feature points, and measuring the temperature of the region of interest using corresponding values ​​from one or more of the infrared images; and The assessment includes measuring head pose orientation based on the one or more facial feature points to determine whether the target is in the measurement position.

7. The method according to claim 1, wherein: The temperature data is correlated with measurement locations on the target, including the cheeks, tear ducts, and / or forehead. The assessment includes correlating the measurement location with the target's core body temperature; and The method also includes analyzing a visual image of the target to identify the measurement location.

8. The method according to claim 1, further comprising: A passive blackbody is configured for the thermal camera that is set to capture the infrared image; A temperature sensor is placed near the passive blackbody. Sensing the temperature associated with the blackbody; The sensed temperature is wirelessly transmitted to the thermal camera; and The thermal camera is calibrated using at least one of the sensed temperature and an infrared image including the passive blackbody.

9. The method according to claim 1, further comprising: Measuring the temperature of a surface with a known temperature, wherein the surface is a blackbody, an object with a known temperature, and / or a device for measuring ambient temperature; and The temperature data is adjusted using the measured temperature to account for measurement error.

10. The method according to claim 1, further comprising: A hollow body is provided for the thermal camera that captures the infrared image, the hollow body including a high emissivity surface at the end facing the thermal camera; and The image of the high emissivity surface is processed into a blackbody.

11. The method according to claim 1, wherein: The acquisition includes: Capture multiple visible light images using a visible light camera. Track the target in the visible light image. Identify the measurement location on the target, and The infrared image was captured using a thermal camera; and The visible light camera has a wider field of view than the thermal camera.

12. The method of claim 11, further comprising: The thermal camera is directed toward the measurement location for magnified display; and The measurement location includes more pixels in the infrared image than in the visible light image.

13. The method according to claim 1, wherein, The identification includes: Capture multiple visible light images of the scene; The desired measurement location is detected in the visible light image; Acquire an infrared image of the desired measurement location; Analyze the infrared image to identify multiple markers within the measurement location; Choose one of the flags based on the likelihood of accurate temperature measurement; Establish the region of interest associated with the selected marker; Measure the maximum temperature within the region of interest; and The reference temperature data for the scene is updated using the measured temperature.

14. The method according to claim 1, wherein: The acquisition includes sampling temperature data from the measurement location over time and comparing the target temperature trajectory with one or more trajectory models; The evaluation includes applying the corresponding trajectory model to the sampled temperature data and applying the model to predict core body temperature; and The method also includes detecting trajectory alarm conditions for sampled temperature data that follow a corresponding trajectory model.

15. The method according to claim 1, further comprising: Determine multiple temperature profiles associated with multiple paths to a target within a scene, wherein one or more extensions of the paths span multiple scenes captured by multiple imaging devices; Determine one of the paths associated with the target; and The temperature data is adjusted based on the temperature curve of one of the paths determined in the path.

16. The method according to claim 1, further comprising: A reference temperature sample is obtained by presenting a user interface with user prompts configured to guide the user to obtain an accurate reference temperature sample. Present visual markers associated with a desired target, including the outline and / or alignment box of a head, wherein the visual markers are adapted to the target distance and / or angle; and If the target is determined to have a fever, an alarm indication will be displayed.

17. The method according to claim 1, further comprising: Initialize the temperature distribution associated with fever; Set an isotherm alarm covering the temperature distribution; Capture thermal images and perform face detection in the thermal images in response to the isotherm alarm; Measure the temperature of the face; and If the measured temperature is higher than the threshold, the temperature distribution is updated.

18. A system configured to perform the method of claim 1, the system comprising: A thermal camera, configured to capture the infrared image; and A logic device configured to perform the identification, the acquisition, the evaluation, and the processing.

19. The system of claim 18, further comprising: A passive blackbody, wherein the passive blackbody is configured to be positioned in the field of view of the thermal camera; A temperature sensor, configured to be located near the passive blackbody and to sense the temperature associated with the blackbody; A wireless interface configured to transmit the sensed temperature to the thermal camera; and The logic device is configured to calibrate the thermal camera using at least one of the sensed temperature and an infrared image including the passive blackbody.

Images