Processing system

Abstract

The present invention provides a processing system. The processing system includes: a first camera system configured to capture a first image based on a terahertz wave from an inspection object; a second camera system configured to capture a second image of the inspection object based on an electromagnetic wave having a wavelength different from the terahertz wave; and a processor configured to process the first image and the second image, wherein the processor detects an inspection area based on the second image and processes information of an area of ​​the first image corresponding to the inspection area.

Description

Technical Field

[0001] The invention relates to a processing system including a camera system. Background Art

[0002] An inspection technique using terahertz waves is known. Japanese Patent Application Laid-Open No. 2004-286716 discloses a method of inspecting illegal drugs (such as narcotics) sealed in a sealed letter.

[0003] When processing images acquired by terahertz waves for inspection, a sufficient amount of information may not be obtained due to the positional relationship between the inspection object and the detection unit or the movement of the inspection object. In addition, when inspecting a person wearing clothes, scattering may occur due to the clothes, or the propagation of terahertz waves may be hindered by the environment. Therefore, sufficient inspection may not be performed. Summary of the invention

[0004] The present invention has been made in view of the above circumstances, and provides a processing system capable of more advantageously performing inspection using terahertz waves.

[0005] According to a first aspect of the present invention, there is provided a processing system comprising: a first camera system configured to capture a first image based on a terahertz wave from an inspection object; a second camera system configured to capture a second image of the inspection object based on an electromagnetic wave having a wavelength different from that of the terahertz wave; and a processor configured to process the first image and the second image, wherein the processor detects an inspection area based on the second image and processes information of an area of ​​the first image corresponding to the inspection area.

[0006] According to a second aspect of the present invention, there is provided a processing system comprising: a first camera system configured to capture a first image based on a terahertz wave from an inspection object; a second camera system configured to capture a second image of the inspection object based on an electromagnetic wave having a wavelength different from that of the terahertz wave; a control unit configured to control the first camera system; and a processor configured to process the second image, wherein the control unit controls the first camera system based on a result of processing the second image by the processor.

[0007] Further features of the present invention will become apparent from the following description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 is a conceptual diagram of a processing system according to a first embodiment;

[0009] Figure 2 is a flowchart of the process according to the first embodiment;

[0010] Figure 3 is a conceptual diagram of a processing system according to a second embodiment;

[0011] Figure 4 is a flowchart of a process according to the second embodiment;

[0012] Figure 5 is a conceptual diagram of a processing system according to a third embodiment;

[0013] Figure 6 is a flowchart of a process according to a third embodiment;

[0014] Figure 7 is a conceptual diagram of a processing system according to a fourth embodiment;

[0015] Figure 8 is a conceptual diagram of a processing system according to a fifth embodiment;

[0016] Fig. 9 is a diagram showing an example of the arrangement of a processing system;

[0017] Fig.10 is a diagram showing an example of the arrangement of a processing system;

[0018] Fig.11A is a diagram showing an example of a model for machine learning;

[0019] Fig. 11B is a diagram showing an example of inspection using a learning model;

[0020] Fig. 12A is a diagram showing an example of a model for machine learning; and

[0021] Fig. 12B is a diagram showing an example of control using a learning model. DETAILED DESCRIPTION

[0022] Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. It should be noted that the following embodiments are not intended to limit the scope of the appended claims. A plurality of features are described in the embodiments. Not all of the plurality of features are essential to the present invention, and a plurality of features may be combined arbitrarily. In addition, throughout the accompanying drawings, the same reference numerals denote the same parts, and their repeated descriptions will be omitted. In the present invention, terahertz waves include electromagnetic waves in a frequency range of 30 GHz to 30 THz. The concept of electromagnetic waves may include visible light, infrared light, and radio waves such as millimeter waves.

[0023] (First embodiment)

[0024] Will refer to Figure 1The processing system 401 according to the first embodiment is described in general terms. The processing system 401 includes a first camera system, a second camera system, and a processor, wherein the first camera system includes a first illumination source 404 and a first camera 402, the second camera system includes a second camera 405, and the processor includes a pre-processing unit 406 and a post-processing unit 407.

[0025] The first camera 402 of the first camera system acquires a first image based on a terahertz wave 403 of a first wavelength radiated from a first illumination source 404. The inspection object 410 is illuminated with the terahertz wave 403 radiated from the first illumination source 404. If the inspection object 410 is a person wearing clothes, the terahertz wave 403 passes through the fibers of the clothes and is reflected by the metal or ceramic held by the inspection object 410. It is known that certain substances (for example, RDX (trimethylene trinitramine) as an explosive) absorb terahertz waves close to 0.8 THz, so the reflected wave is reduced. The first camera 402 acquires a first image based on the reflected wave.

[0026] The second camera 405 of the second camera system acquires a second image from an electromagnetic wave having a different wavelength from the terahertz wave irradiated from the first illumination source 404. As the electromagnetic wave of a different wavelength, visible light, infrared light, or millimeter wave can be used. When infrared light is used, an illumination source (not shown) different from the first illumination source 404 can be prepared. The second image acquired by the second camera 405 is processed by the preprocessing unit 406. The preprocessing unit 406 performs a process of detecting an inspection area from the second image.

[0027] If the second image is acquired by visible light, and the inspection object 410 is a person, the inspection area can be detected by detecting a specific part of the clothes as the inspection area. The inspection area can be specified by creating a model by machine learning and classifying the area of ​​the captured second image by the model. Alternatively, the area can be specified based on information about the shape of the object stored in the database 409. If the second image is acquired by millimeter waves, the part of the image where the intensity distribution is higher than the threshold or the part with a large intensity difference can be detected as the inspection area. If the second image is acquired using infrared light, the part with less radiation of infrared light caused by water or the specific part of the clothes in the image detected by the night vision device can be detected as the inspection area. Even in a dark place or a place where the sight is poor due to weather, the inspection area can be detected using infrared light or millimeter waves. When the inspection area is detected from an image of a person wearing clothes, the unnaturally bulging part of the clothes, the chest of the person, or the pocket part of the clothes can be detected as the inspection area.

[0028] Will be based on Figure 2Describe the inspection of the inspection object 410 by the processor. The pre-processing unit 406 detects the inspection area (steps S422 and S423) from the second image (step S421) acquired by the second camera 405 by the above method. The post-processing unit 407 processes the image data of the information of the area of ​​the first image corresponding to the inspection area detected from the second image (step S425). The first image is an image acquired by the first camera 402 using terahertz waves (step S424), and is an image obtained by looking through clothes, etc. If there is a metal or ceramic object under the clothes, an image can be obtained from the reflected wave. Therefore, the shape of the object can be detected by processing the first image. After the inspection area is detected from the second image, the area corresponding to the inspection area in the first image is selected by comparing the first image and the second image. For the area corresponding to the inspection area detected from the second image, subsequent image processing for the first image is performed.

[0029] When an area corresponding to the inspection area is selected from the first image and image processing is performed, processing can be performed while reducing unnecessary information. Therefore, compared with the processing of the entire image data, the processing load can be reduced and the speed can be increased. Therefore, even if the inspection object 410 is moving, features can be detected multiple times from the first image in a short time with a short moving distance. The determination unit 408 estimates the object under the clothes based on multiple detected features (step S426). The multiple features can be features of a part of the object. The determination unit 408 can classify the shape of the object detected from the first image based on the data in the database 409. The classification can be performed using a model created by machine learning. Due to the movement of the inspection object 410 or the positional relationship between the inspection object and the camera, it is considered that the information of the shape obtained from the image can be information of a part of the object. Even in this case, the estimation accuracy can be improved by classifying the features based on the information of multiple features, accumulating multiple results, and determining based on the accumulated classification results (step S427).

[0030] When the processing system is used in a security monitoring system, the risk of the object detected from the inspection area is determined based on the accumulation of the classification results for the inspection object 410 (step S428). Regarding this determination, the determination based on the cumulative results of the classification can be performed based on a model through machine learning. If it is determined that the inspection object 410 holds a dangerous substance, the inspection object 410 can be notified to the outside that the inspection object 410 holds the dangerous substance. When the inspection object 410 passes through a door arranged with a processing system, the processing system can notify the outside of a warning. When the inspection object 410 puts in a ticket and passes through a ticket gate, the processing system can associate the ticket with the inspection object 410 and notify the inspection object 410 that the inspection object 410 is a surveillance object. If the second image is obtained using visible light, the inspection object 410 can be displayed so that the inspection object 410 can be easily seen by displaying the second image and the first image in a superimposed manner on the monitor. When it is determined to be aborted, the inspection is repeated until the end condition is met. The end condition can be the number of repetitions of the inspection (S429).

[0031] Next, a method of specifying an inspection area from the second image acquired by the second camera 405 using a model (artificial intelligence) created by machine learning will be described in detail.

[0032] Fig.11A 48 is a diagram schematically showing a model for machine learning (i.e., a learning model). In this example, a neural network including an input layer 481, an output layer 483, and at least one intermediate layer 482 is used as a learning model. Image data is input to the input layer 481. In addition, the output layer 483 outputs a feature quantity indicating a partial area of ​​the input image.

[0033] As a learning method of the learning model, supervised data with correct answer labels can be used. That is, a learning model is learned by means such as back propagation using a data set including a set of input image data and a label representing an inspection object area in the image data. The object area includes a person, a bag, a container, etc., but is not limited to these. CNN (Convolutional Neural Network) can be used as a model for learning by deep learning.

[0034] Fig. 11B 48 is a schematic diagram showing a method of specifying an inspection area using a learning model. As input, a visible light image 484 is input. The learning model 485 outputs a feature quantity representing the inspection object area. As an output form, for example, Fig. 11B As shown, the object area in the image 486 is surrounded by a line. Alternatively, coordinate information for image processing may be output.

[0035] When determinations using machine learning or artificial intelligence are made in this manner, accurate determinations can be made.

[0036] In addition, when a specific target object is detected from the first image, a model (artificial intelligence) created by machine learning can be used similarly. In this case, an image of a terahertz wave having the same wavelength as the terahertz wave used by the first camera 402 for shooting is used as supervisory data for learning.

[0037] (Second embodiment)

[0038] In this embodiment, the second camera system is equipped with a second illumination source 411 that radiates terahertz waves. Figure 3 To describe this embodiment. The second illumination source 411 is an illumination source that generates a terahertz wave of a second wavelength different from the first illumination source 404. As described in the first embodiment, a specific substance that absorbs a terahertz wave of a specific wavelength is known. Therefore, a terahertz wave of a first wavelength (about 0.8 THz for RDX as an explosive) that is a wavelength that is easily absorbed by a specific substance is radiated from the first illumination source 404 to the inspection object 410. If the inspection object 410 holds a substance having a characteristic of easily absorbing the terahertz wave of the first wavelength, the reflection of the part holding the substance becomes smaller. On the other hand, when a wavelength that is rarely absorbed by a specific substance (about 0.5 THz when the first wavelength is 0.8 THz) is selected as the terahertz wave of the second wavelength generated by the second illumination source 411, the specific substance reflects the terahertz wave of the second wavelength. For the same inspection area, the difference between the reflected waves from the specific substance can be used to specify the substance.

[0039] Will be based on Figure 4 The processing according to the present embodiment is described. The pre-processing unit 406 detects the high reflection area in the second image acquired by the terahertz wave of the second wavelength as the inspection area (steps S431 and S432). The post-processing unit 407 acquires the first image taken by the first camera 402 based on the terahertz wave of the first wavelength (step S434), and starts to process the image data for the area of ​​the first image corresponding to the inspection area detected from the second image. The post-processing unit 407 can calculate the difference between the information of the inspection area in the second image and the information of the area corresponding to the inspection area of ​​the first image (step S435). By calculating the difference between the two pieces of information, the data of the part where the reflection and absorption in the second image are almost equal to the reflection and absorption in the first image is almost eliminated. However, even by calculating the difference between the two images, the data of the part where the reflection and absorption are different between the first wavelength and the second wavelength will not be eliminated. In this way, the difference in the terahertz wave absorptivity of the substance can be used to perform spectral analysis of the substance in the inspection area. The type of substance can be estimated using spectral analysis. In addition, since the scattering or reflection of clothes is eliminated, unnecessary signals from clothes can be reduced from the obtained image information, and the signal-to-noise ratio of the image can be improved.

[0040] If the inspection object person holds a substance that easily absorbs the first wavelength, the substance detected in the inspection area can be classified based on the difference in absorbance between the first wavelength and the second wavelength (step S436). Regarding classification, when the relationship between a specific substance and a wavelength is maintained in the database 409, the determination unit 408 can be classified based on the database 409. The determination unit 408 can use a model created by machine learning to classify. Using the above method, it can be estimated that the inspection object 410 holds a substance that absorbs a specific wavelength. It is known that there are dangerous substances in substances that absorb terahertz waves of a specific wavelength. The presence of dangerous substances can be estimated by spectral analysis. The detection accuracy can be improved by accumulating multiple spectral analysis results (step S437). Therefore, it is determined that the inspection object 410 may hold a dangerous substance (step S438). Regarding determination, a determination based on the cumulative results of classification can be performed based on a model through machine learning. If it is determined that a dangerous substance is held, the processing system notifies the outside that the inspection object 410 holds a dangerous substance. When the inspection object 410 passes through a door arranged with a processing system, the processing system can notify the outside of a warning. When the person of the inspection object 410 puts in the ticket and passes through the ticket gate, the processing system can associate the ticket with the inspection object 410 and notify the outside that the person is the monitoring object. Regarding the wavelength of the terahertz wave irradiated from the second illumination source 411, multiple illumination sources capable of irradiating terahertz waves of multiple (i.e., three or more) wavelengths can be combined according to the absorption spectrum of the substance to be detected. When it is determined to be aborted, the inspection is repeated until the end condition is met. The end condition can be the number of repetitions of the inspection (S439).

[0041] (Third Embodiment)

[0042] In this embodiment, based on the detection of a specific area in the second image captured by the second camera system, the control unit 412 is controlled to control the first illumination source 404 and the first camera 402 in the first camera system. Figure 5 and Figure 6 This embodiment is described.

[0043] The second camera 405 of the second camera system acquires a second image from an electromagnetic wave having a wavelength different from the terahertz wave radiated from the first illumination source 404. As the electromagnetic wave of different wavelength, visible light, infrared light, or millimeter wave can be used. The second image acquired by the second camera 405 is processed by the preprocessing unit 406. The preprocessing unit 406 detects the inspection area from the second image (steps S452 and S453). As described in the first embodiment, the detection of the inspection area is performed.

[0044] According to the position and range of the inspection area detected from the second image and the state of the inspection area, the conditions when the first camera is shooting are controlled. The conditions include control of the posture of the first camera, control of the gain of the acquired image, and control of the shooting range and viewing angle for zooming or trimming (step S454). The output level (output power) and wavelength of the terahertz wave irradiated from the first illumination source 404 can be changed according to the intensity of the reflected signal from the inspection area or the object object in the inspection area. Through this control, the inspection accuracy can be improved. The first camera system controlled by the control unit 412 acquires the first image based on the terahertz wave of the first wavelength (step S455). The post-processing unit 407 processes the inspection area based on the acquired first image (step S456). Afterwards, the determination unit 408 determines and classifies the object (steps S457, S458 and S459). When the processing system is a security monitoring system, the risk is determined based on the accumulation of the classification results. If it is determined that the inspection object 410 holds a dangerous substance, the processing system notifies the outside that the inspection object 410 holds a dangerous substance. When the inspection object 410 passes through a door where the processing system is arranged, the processing system may notify a warning to the outside. When the inspection object 410 puts in a ticket and passes through the ticket gate, the processing system may associate the ticket with the inspection object 410 and set the inspection object 410 as a monitoring object. When it is determined to be suspended, the inspection is repeated until the end condition is met. The end condition may be the number of repetitions of the inspection (S460).

[0045] The photographing of the first camera 402 can be controlled by a model (artificial intelligence) created by machine learning from the second image acquired by the second camera 405. This method will be described in detail.

[0046] Fig. 12A 48 is a diagram schematically showing a model for machine learning (i.e., a learning model). In this example, a neural network including an input layer 481, an output layer 483, and at least one intermediate layer 482 is used as a learning model. Image data is input to the input layer 481. In addition, the output layer 483 can output the classification result of the object of the input image.

[0047] As a learning method for the learning model, supervised data with correct answer labels can be used. That is, a learning model is learned by means such as back propagation using a data set including a set of input image data and a label representing an inspection object area in the image data. Learning by deep learning can be performed using CNN (convolutional neural network) as a model.

[0048] Data for object classification is selected according to the purpose of control. In order to control zoom, supervisory data with a label indicating that the object is small or large, or has an appropriate size is used. In order to control gain, supervisory data with a label indicating that the exposure of the object is underexposed, properly exposed, or overexposed is used. In order to control switching of the wavelength used by the first camera 402, supervisory data that associates the input image with an appropriate wavelength band is used. Alternatively, in order to control the output of the first illumination source 404, supervisory data classified according to the transmittance of the terahertz wave output from the first illumination source 404 is used. These supervisory data are only examples and are not limited to these. Learning can be performed by deep learning without using supervisory data. In this case, learning can be performed by a means for evaluating the control result according to the output for the input.

[0049] Fig. 12B 4 is a schematic diagram showing a method of controlling the first camera 402 using a learning model. As an input, a visible light image 487 is input. The learning model 488 can output information indicating whether there is an object having low sensitivity to terahertz waves in the wavelength range detected by the first camera 402. Based on the result, control is performed to increase the output of the first illumination source 404, thereby obtaining an image 489.

[0050] When determination using machine learning or artificial intelligence is performed in this manner, detection accuracy of the target object using the first camera 402 can be made higher.

[0051] In addition, when a specific target object is detected from the first image, a model (artificial intelligence) created by machine learning can be used similarly. In this case, an image of a terahertz wave having the same wavelength as the terahertz wave used by the first camera 402 for shooting is used as supervisory data for learning.

[0052] (Fourth embodiment)

[0053] In this embodiment, an environment monitoring unit 413 configured to monitor the humidity around the processing unit is provided. Figure 7 This embodiment is described. Terahertz waves are easily absorbed by water vapor. Terahertz waves of a longer wavelength are hardly affected by water vapor. Therefore, an environment monitoring unit 413 is provided to measure humidity, and the camera system is controlled so as to be hardly affected by the surrounding environment.

[0054] More specifically, if the environment monitoring unit 413 detects that the humidity becomes high, the wavelength of the terahertz wave 403 radiated from the first illumination source 404 is switched to a longer wavelength than the currently used wavelength. Depending on the humidity, the wavelength can be switched to a wavelength that is hardly affected by water vapor (a region that exists near a wavelength of 1.2 mm or 0.75 mm, where the attenuation of the atmosphere is particularly small). When the wavelength of the terahertz wave becomes longer, the resolution of the image taken by the camera decreases. However, the influence of water vapor can be reduced and the inspection can be continued.

[0055] (Fifth Embodiment)

[0056] In this embodiment, photographing is performed using terahertz waves of different wavelengths. A second image is acquired using terahertz waves of a second wavelength longer than the wavelength when the first image is photographed, and the inspection area is detected from the second image. The inspection area may be detected as an area including an object of a predetermined shape using a model created by machine learning, or an area where the spectrum of a reflected wave of a predetermined wavelength changes may be detected as the inspection area.

[0057] Will refer to Figure 8 This embodiment is described. Based on the inspection area detected from the second image, image data of the area of ​​the first image corresponding to the inspection area is processed. The first image captured by the terahertz wave of the first wavelength generated by the first illumination source 404 is shooting 1, and the second image captured by the terahertz wave of the second wavelength generated by the illumination source 411 is shooting 2. Since the image of shooting 1 is acquired using a terahertz wave having a wavelength shorter than that in shooting 2, the resolution is high and the amount of information is large. Therefore, in the image acquired by the terahertz wave, the shape of each object held by the inspection object 410 is clear. However, since a short-wavelength terahertz wave is used, the depth of field is shallow, and the shooting is sensitive to changes in the posture of the inspection object 410. More specifically, according to the posture of the inspection object 410, a partial shape is acquired as the shape of the object held by the inspection object 410. On the other hand, in the image obtained by shooting 2, since the wavelength of the terahertz wave is longer, the resolution is low, and the shape of each object is not clear compared to shooting 1. However, since a long wavelength terahertz wave is used, the depth of field is deep, and the shooting is insensitive to changes in the posture of the inspection object 410. More specifically, regardless of the posture of the inspection object 410, the overall shape of the object held by the inspection object 410 is acquired. When the low-resolution shot 2 is processed to specify the position of the object held by the inspection object 410, and the data of the shot 1 is processed based on the detected inspection area, the processing load can be reduced, and processing can be performed at a higher speed. Therefore, even if the inspection object 410 is moving, the features of the inspection object 410 can be detected multiple times in a short time and at a short moving distance, and the object under the clothes can be estimated based on the detected features.

[0058] In addition, when the difference between shot 1 and shot 2 using two different wavelengths of terahertz waves is calculated, the reflection of the clothes is eliminated, and the noise in the obtained image information can be reduced. More specifically, since scattering is the main component of the reflection from the entire clothes, the intensity difference is small, and the shooting is insensitive to the posture change of the inspection object 410 (random noise is added to the acquired image as a whole). Therefore, when the differential image between shot 1 and shot 2 is calculated, the signal of the clothes is eliminated. In addition, when the difference is calculated, an image based on the difference in the terahertz wave absorption rate of the substance through which the terahertz wave passes can be obtained. Therefore, the shape of an object containing a substance other than metal or ceramic as a component can be detected based on the difference between the first image and the second image.

[0059] By classifying the shapes of the objects detected from the shot 1, the determination unit 408 estimates the objects in the inspection area. If the inspection object 410 moves, the shape of the object obtained from the image is generally local. Therefore, the determination accuracy can be improved by accumulating multiple classification results and determining based on the accumulated classification results. In the case of a security monitoring system, the risk is determined based on the accumulation of the classification results. If it is determined that the inspection object 410 holds a dangerous substance, the processing system notifies the inspection object 410 that the dangerous substance is held. When the inspection object 410 passes through a door where the processing system is arranged, the processing system can notify a warning to the outside. When the inspection object 410 puts in a ticket and passes through the ticket gate, the processing system can associate the ticket with the inspection object 410 and set the inspection object 410 as a monitoring object.

[0060] (Sixth embodiment)

[0061] Will refer to Fig. 9 and Fig.10 Describe application examples of the processing system. Fig. 9An example is shown in which a first illumination source 404 for a terahertz wave of a first wavelength and a second illumination source 411 of a second wavelength different from the first wavelength are arranged on one side of a doorway 414 of a vehicle or the like. A first camera 402 configured to shoot based on a terahertz wave of a first wavelength, a second camera 405-1 configured to shoot based on one of visible light, infrared light, and a millimeter wave, and a second camera 405-2 configured to shoot based on a terahertz wave of a second wavelength are arranged on the other side of the doorway 414. When the camera and the illumination light source are combined, the processing for inspection described in the first to fifth embodiments can be combined. The inspection object 410 can be tracked by the second camera 405-1, and the posture and viewing angle of the first camera 402 can be controlled. When the wavelength of the terahertz wave used for shooting by the second camera 405-2 configured to shoot based on the terahertz wave is set according to the absorption rate of the substance, spectral analysis can be performed. In addition, when the second cameras 405-1 and 405-2 are used to detect the inspection area, the processing load for the first image captured by the first camera 402 can be reduced. In addition, the difference in the absorbance of the material for the wavelength of the terahertz wave can be used to detect the shape of an object containing a material other than metal or ceramic as a component. In this embodiment, as the second camera 405, a camera for visible light, infrared light or millimeter wave and a camera for the terahertz wave of the second wavelength are used. However, only one of the camera for visible light, infrared light or millimeter wave and the camera for the terahertz wave of the second wavelength can be used as the second camera. The illumination source and the camera can be buried in the wall surface, ceiling or floor surface inconspicuously. The illumination source and the camera can be arranged on the left and right sides of the doorway 414. When the illumination source and the camera are arranged near the doorway 414, the overlap of multiple inspection objects 410 can be reduced, and the inspection accuracy can be improved.

[0062] Will refer to Fig.10 An example is described below in which a processing system is arranged near a ticket gate 415 installed at a ticket gate of a station. A first illumination source 404 for a terahertz wave of a first wavelength and a second illumination source 411 of a second wavelength different from the first wavelength are arranged on one side of the ticket gate 415. A first camera 402 configured to shoot based on a terahertz wave of a first wavelength, a second camera 405-1 configured to shoot based on one of visible light, infrared light, and a millimeter wave, and a second camera 405-2 configured to shoot based on a terahertz wave of a second wavelength are arranged on the other side of the ticket gate 415. When the processing system is arranged near the ticket gate 415, overlap of a plurality of inspection objects 410 can be reduced, and inspection accuracy can be improved.

[0063] The operation of the processing system may be started based on the detection of the movement of the inspection object 410, the opening / closing of the door, the insertion of the ticket into the ticket inspection machine 415, etc. by a sensor provided separately from the processing system. A plurality of first cameras and second cameras may be provided. By using a plurality of cameras, the detection accuracy may be improved, the number of inspection objects may be increased, and the inspection area may be expanded.

[0064] While the present invention has been described with reference to exemplary embodiments, it is to be understood that the invention is not limited to the disclosed exemplary embodiments.The scope of the following claims is to be accorded the broadest interpretation so as to encompass all such modifications and equivalent structures and functions.

Claims

1. A processing system comprising: a first camera system configured to capture a first image based on a first terahertz wave from an inspection object, the first terahertz wave having a frequency in a frequency range of 30 GHz to 30 THz; a second imaging system configured to capture a second image of the inspection object based on a second terahertz wave having a longer wavelength than the first terahertz wave, the second terahertz wave having a frequency in a frequency range of 30 GHz to 30 THz; as well as a processor configured to process the first image and the second image, wherein the processor detects the inspection area based on the second image and processes information of an area of ​​the first image corresponding to the inspection area, wherein the processor calculates a difference between the information of the area of ​​the first image corresponding to the inspection area and the information of the area of ​​the second image corresponding to the inspection area, and Therein, the processor classifies the material detected in the inspection area based on the difference.

2. The processing system according to claim 1, wherein: The processing system further comprises: A control unit is configured to control the first camera system based on the detection of the examination region.

3. The processing system according to claim 2, wherein: The control unit controls at least one of a posture of a camera of the first imaging system, a gain of the camera, a photographing range of the camera, an output of a terahertz wave, and a wavelength of the terahertz wave.

4. The processing system according to claim 2, wherein: The processing system further comprises: an environment monitoring unit configured to measure the surrounding environment, The control unit controls at least one of the wavelength and the output of the radiated terahertz wave according to the output of the environment monitoring unit.

5. The processing system according to claim 1, wherein: The processing system also includes a plurality of illumination sources for photographing.

6. The processing system according to claim 1, wherein: At least one of the first camera system and the second camera system includes a plurality of cameras for photographing.

7. The processing system according to claim 1, wherein: The capturing of the first image, the capturing of the second image, and the above-described processing are repeated.

8. The processing system of claim 1, wherein: The detection of the inspection region is based on one of the shape of the inspection region and information of a spectrum detected from the inspection region.

9. The processing system of claim 1, wherein: The processing system further comprises: A determination unit is configured to determine a risk regarding the inspection object based on the output of the processor.

10. The processing system according to any one of claims 1 to 9, wherein: The processor detects an inspection area in the second image based on a learning model learned using at least one of deep learning and machine learning.

11. A processing system comprising: a first camera system configured to capture a first image based on a first terahertz wave from an inspection object, the first terahertz wave having a frequency in a frequency range of 30 GHz to 30 THz; a second imaging system configured to capture a second image of the inspection object based on a second terahertz wave having a longer wavelength than the first terahertz wave, the second terahertz wave having a frequency in a frequency range of 30 GHz to 30 THz; a control unit configured to control the first camera system; as well as a processor configured to process the second image, The control unit controls the first camera system based on the result of processing the second image by the processor, wherein the processor detects the inspection area based on the second image and processes information of an area of ​​the first image corresponding to the inspection area, wherein the processor calculates a difference between the information of the area of ​​the first image corresponding to the inspection area and the information of the area of ​​the second image corresponding to the inspection area, and Therein, the processor classifies the material detected in the inspection area based on the difference.

12. The processing system according to claim 11, wherein: The control unit controls the conditions under which the first camera system captures the first image based on the result of the processing of the second image by the processor, and The conditions include one or more conditions selected from a posture of a camera of the first camera system, a zoom of a camera of the first camera system, a viewing angle of a camera of the first camera system, a gain for the first image, and a cropping range for the first image.

13. The processing system of claim 11, wherein: The processing system also includes an illumination source configured to radiate terahertz waves, Wherein, the control system controls at least one of a wavelength and an output of the terahertz wave radiated from the illumination source based on a result of the processing.

14. The processing system of claim 11, wherein: The processing system also includes a plurality of illumination sources for photographing.

15. The processing system of claim 11, wherein: At least one of the first camera system and the second camera system includes a plurality of cameras for photographing.

16. The processing system of claim 11, wherein: The processing system further comprises: A determination unit is configured to determine a risk regarding the inspection object based on the output of the processor.

17. A processing system according to any one of claims 11 to 16, wherein: The processor processes the second image based on a learning model learned using at least one of deep learning and machine learning.

Images