JP2016217899A - Object identification device, object identification system, object identification method, and object identification program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object identification device capable of identifying an object in a short time using hyperspectral data.SOLUTION: An object identification device 10 includes a storage section 11 for storing object information including hyperspectral data corresponding to a prescribed object and polarization information indicating the polarization direction of light from the prescribed object received by an image sensor that has acquired the hyperspectral data, and an extraction section 12 for extracting object information corresponding to the polarization information that was acquired when the hyperspectral data of the object to be identified was acquired from the storage section 11 and extracting the object information corresponding to the acquired hyperspectral data from the extracted object information.SELECTED DRAWING: Figure 12

Description

  The present invention relates to an object identification device, an object identification system, an object identification method, and an object identification program, and more particularly to an object identification device, an object identification system, an object identification method, and an object identification program using a polarization hyperspectral camera device.

  There is a hyperspectral camera that can acquire hyperspectral information, which is a spectrum that has been split into several tens of bands (types). A hyperspectral camera is a camera that can simultaneously acquire spatial information and hyperspectral information by acquiring hyperspectral information for each pixel.

  In the verification process using general hyperspectral information, hyperspectral information including wavelength information corresponding to pixel spatial information is verified for each pixel with spectral data stored in the spectrum library. By comparing with the spectrum data, it is identified what object the object indicated by each pixel is.

  However, the method of collating hyperspectral information and spectral data pixel by pixel requires a lot of time because collation is performed by the number of acquired pixels.

  As a method of solving the above problem, there is a method of clustering acquired hyperspectral data and collating each clustered cluster with each spectrum data stored in a spectrum library. Spectral data verification is not performed on all the acquired pixels, so that the time until all verification processing is completed is shortened. After the collation and the object corresponding to each cluster are identified, the information of the object corresponding to the cluster corresponding to the pixel is reflected in each pixel of the acquired hyperspectral data.

  The identification method using clustering is suitable for the purpose of grasping the general distribution tendency of the spectrum such as the distribution of minerals on the ground surface and the distribution of forest tree species.

  In addition to the identification method using clustering, Patent Document 1 describes an object identification method that does not perform matching processing on all pixels. In Patent Document 1, any one of a plurality of end members is removed from spectral data recorded in each pixel of a hyperspectral image, and the spectral data of the pixel after removal is compared with spectral data in a database. An image analysis device for specifying an object recorded in a pixel is described.

  Patent Document 2 describes an image processing apparatus using a remote sensing image that recognizes an object shown in an image with high accuracy.

JP 2010-237865 A JP 2010-237865 A

  The identification method using the above clustering has a problem that the identification result varies depending on, for example, the accuracy and method of clustering, and the matching and identification methods. In addition, for example, the clustering process itself requires a long time.

  Accordingly, an object of the present invention is to provide an object identification device, an object identification system, an object identification method, and an object identification program that can identify an object in a short time using hyperspectral data.

  The object identification device according to the present invention stores object information including hyperspectral data corresponding to a predetermined object and polarization information indicating the polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. And the object corresponding to the hyperspectral data acquired from the extracted object information by extracting the object information corresponding to the polarization information acquired at the time of acquiring the hyperspectral data of the object to be identified. And an extraction unit for extracting information.

  The object identification device according to the present invention stores object information including hyperspectral data corresponding to a predetermined object and polarization information indicating the polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. A storage unit, a camera unit for acquiring hyperspectral data of the identification target object based on light from the identification target object, and an object corresponding to the polarization information acquired when the hyperspectral data of the identification target object is acquired And an extraction unit that extracts information from the storage unit and extracts object information corresponding to the hyperspectral data acquired by the camera unit from the extracted object information.

  An object identification system according to the present invention is an object identification system including a polarization filter on which reflected light reflected by an object to be identified enters, a camera, an object identification device, and a display device. The camera includes a polarization filter. The camera includes a camera unit that acquires hyperspectral data of an object to be identified based on transmitted light. The object identification device receives hyperspectral data corresponding to a predetermined object and an image sensor that acquires the hyperspectral data. A storage unit that stores object information including polarization information indicating the polarization direction of light from a predetermined object, an angle detection unit that detects a rotation angle of the polarization filter, and a polarization direction corresponding to the detected rotation angle Hyper-spectrum obtained by extracting object information corresponding to polarization information from the storage unit, and by the camera unit from the extracted object information It includes an extraction unit for extracting the object information corresponding to the over data, display device, and displaying the extracted object information.

  The object identification method according to the present invention stores object information including hyperspectral data corresponding to a predetermined object and polarization information indicating the polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. An object identification method executed in an object identification apparatus including a storage unit that extracts object information corresponding to polarization information acquired when hyperspectral data of an object to be identified is acquired from the storage unit, and the extracted object Object information corresponding to hyperspectral data acquired from the information is extracted.

  The object identification program according to the present invention stores object information including hyperspectral data corresponding to a predetermined object and polarization information indicating the polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. An object identification program executed in a computer including a storage unit that extracts from the storage unit object information corresponding to polarization information acquired when hyperspectral data of an object to be identified is acquired. And a second extraction process for extracting object information corresponding to the hyperspectral data acquired from the extracted object information.

  According to the present invention, an object can be identified in a short time using hyperspectral data.

It is explanatory drawing which shows the example of 1st Embodiment of the object identification system by this invention. It is a block diagram which shows the structural example of 1st Embodiment of the object identification system by this invention. It is explanatory drawing which shows the hardware structural example of the object identification apparatus by this invention. It is a flowchart which shows the operation | movement of the spectrum library construction process with polarization information by the object identification apparatus 200 in 1st Embodiment. It is explanatory drawing which shows the data processing by the object identification apparatus 200 at the time of construction of the spectrum library with polarization information. It is a flowchart which shows the operation | movement of the object identification process by the object identification apparatus 200 in 1st Embodiment. It is explanatory drawing which shows the data processing by the object identification apparatus 200 at the time of collation with the acquired hyperspectral data and the data registered in the spectrum library 204 with polarization information. It is explanatory drawing which shows the example of 2nd Embodiment of the object identification system by this invention. It is explanatory drawing which shows the other example of 2nd Embodiment of the object identification system by this invention. It is a block diagram which shows the structural example of 2nd Embodiment of the object identification system by this invention. It is explanatory drawing which shows the example of 3rd Embodiment of the object identification system by this invention. It is a block diagram which shows the outline | summary of the object identification device by this invention. It is a block diagram which shows the other outline | summary of the object identification apparatus by this invention. It is a block diagram which shows the outline | summary of the object identification system by this invention.

  First, hyperspectral imaging techniques are described below.

  In the hyperspectral imaging technology, a camera for hyperspectral imaging is installed in a ground-mounted camera, an aircraft, or a satellite. The installed camera takes pictures from the ground or above.

  A spectroscopic device exists behind the lens of the camera for hyperspectral imaging. The spectroscopic device splits light corresponding to any one of visible light, near infrared light, and far infrared light for each predetermined wavelength width. In addition, the range of the wavelength of light corresponding to any of visible light, near infrared light, or far infrared light is, for example, 300 nm to 1200 nm.

  The imaging device of the camera for hyperspectral imaging receives the dispersed light, and converts the received light from an optical signal to an electrical signal. The imaging device creates an image having a spatial axis in the horizontal direction and a wavelength axis in the vertical direction based on the converted electrical signal, and outputs the created image. An image output from the image sensor corresponds to a hyperspectral image.

  An image corresponding to a predetermined spatial axis shows only a one-dimensional light / dark pattern. That is, the imaging device corresponds to a line sensor arranged in a line in a line in the spatial axis direction. Therefore, a camera for hyperspectral imaging installed in a ground-mounted camera acquires data by a mirror scan method, for example. In addition, cameras for hyperspectral imaging installed on aircraft and artificial satellites acquire data by, for example, a push bloom method.

  In order to identify the substance represented by each pixel (pixel) of the hyperspectral image, it is required to grasp in advance how the substance distributed in the observation target region is acquired as a spectrum. In the spectrum library, spectrum information corresponding to each substance is accumulated.

  In addition, the spectral library summarizes spectral characteristics such as the spectrum measured by a spectrophotometer in the laboratory or the actual environment, or the end component (end member) spectrum obtained from the hyperspectral image, and information other than the spectral characteristics. It is saved in the state.

  The information other than the spectral characteristics is, for example, the substance name, observation conditions, wavelength axis, spectral resolution, and observation error. Information other than the spectral characteristics is stored in the spectral library as spectral metadata.

  Examples of the existing spectral library include libraries distributed by research institutions such as USGS (US Geologic Survey), NASA JPL (Jet Propulsion Laboratory), and JHU (Johns-Hopkins University). The library distributed by each research institution includes hundreds to thousands of spectra measured by each research institution.

  There are mainly three methods for collating the acquired hyperspectral data with the spectral data included in the spectral library.

  The first method is a method in which hyperspectral data of all pixels is collated with spectral data included in a spectral library, and a spectrum indicated by spectral data having a high degree of coincidence is determined as a pixel spectrum. An object corresponding to the determined spectrum is identified as an object indicated by the pixel.

  In the first method, since the matching process is performed for each pixel, it takes a lot of time to finish the matching process for all the pixels. That is, when the first method is used, the hyperspectral data of each pixel of a two-dimensional image having a spatial axis in the horizontal direction and a wavelength axis in the vertical direction is collated with the spectral data contained in the spectral library. As a result, the object indicated by each pixel is identified. In the first method, since the matching process is performed by the number of acquired pixels, it takes a lot of time to finish the matching process for all the pixels.

  In the second method, clustering is performed for each pixel having similar data in the entire acquired hyperspectral data, and representative hyperspectral data of each cluster is collated with spectral data included in the spectrum library. It is a technique. As a result of the collation, the spectrum indicated by the spectrum data having a high degree of coincidence is determined as the spectrum of all pixels included in the cluster. An object corresponding to the determined spectrum is identified as an object indicated by all the pixels included in the cluster.

  The second method is suitable for use in classifying natural objects distributed over a wide area such as forests and rocks. However, it is not suitable for use in detecting and identifying artifacts hidden in natural objects. Further, the second method lacks immediacy because it takes time for clustering.

  The third method is to extract pixels having hyperspectral data whose characteristics are different from the hyperspectral data of surrounding pixels (that is, abnormal point pixels), and the hyperspectral data of the abnormal point pixels are included in the spectrum library. This is a method of collating with existing spectral data. As a result of the collation, the spectrum indicated by the spectrum data having a high degree of coincidence is determined as the spectrum of the pixel at the abnormal point. An object corresponding to the determined spectrum is identified as an object indicated by the pixel of the abnormal point.

  In the third method, the pixels with which the spectral data is collated are narrowed down. However, even in the third method, it takes time to extract abnormal point pixels. In addition, since the hyperspectral data of the abnormal point pixels is collated with the entire spectrum data included in the spectrum library, the third method also lacks immediacy.

  That is, all of the first to third methods require a lot of time. Since a lot of time is required, there is a problem in that the above-described method is not applied to an immediate detection process of an object that immediately detects and identifies an artifact or an object detection process that detects and identifies an artificial object in a state close to immediate.

  Each embodiment of the present invention that solves the above problems will be described below.

Embodiment 1. FIG.
[Description of configuration]
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is an explanatory diagram showing an example of a first embodiment of an object identification system according to the present invention. The example shown in FIG. 1 corresponds to an example in which the present invention is applied to a ground-mounted hyperspectral camera.

  The light reflected by the material includes not only polarized light. Objects, especially artifacts, often reflect a specific polarization when reflecting light.

  The object identification apparatus included in the object identification system 100 according to the present embodiment using the above-described property uses a search method that performs two-stage narrowing, that is, narrowing by “specific polarization” and narrowing by “specific spectrum”. By using a search method that performs two-stage narrowing, the object identification device shortens the time until the target spectrum is extracted.

  The object identification device includes a spectrum library that stores wavelength information and polarization angle information together. When the wavelength information and the polarization angle information are stored together in the spectrum library, the time required for the collation process and the identification process of “spectrum acquired using specific polarization” is shortened. That is, the object identification device can shorten the time required to search for the target spectrum and the time required to identify the object. In addition, since a plurality of pieces of information are used, the object identification device can improve the accuracy of the matching process and the identification process.

  In addition, when the user wants to search for spectrum data having specific polarization angle information acquired in advance and to identify an object, the object identification device can obtain polarization angle information corresponding to arbitrary spectrum data registered in the spectrum library. May be used to rotate the polarizing filter.

  The hyperspectral data based on the acquired arbitrary polarization is highly likely to match the spectral data corresponding to the polarization angle information used for rotating the polarization filter. That is, when the object identification device rotates the polarization filter using the polarization angle information corresponding to the spectrum data, the target spectrum data is easily searched and the corresponding object is identified.

  FIG. 2 is a block diagram showing a configuration example of the first embodiment of the object identification system according to the present invention. As shown in FIG. 2, the object identification system 100 includes an object identification device 200, a polarization filter rotation device 300, an angle detection device 301, a polarization filter 400, a hyperspectral camera 500, a scan mirror 600, and a display device. 700.

  The object identification device 200 includes a polarization angle control unit 201, a data input / output unit 202, a camera control unit 203, a spectrum library with polarization information 204, and a mirror control unit 205.

  The polarization angle control unit 201 has a function of outputting a control signal for controlling the rotation operation of the polarization filter 400 by the polarization filter rotation device 300 via the control line. The polarization angle control unit 201 has a function of acquiring and recording polarization angle information when acquiring hyperspectral data. The polarization angle control unit 201 receives the polarization angle information acquired by the angle detection device 301 via the signal line.

  The data input / output unit 202 has a function of recording the two characteristic indexes of the polarization angle and the spectrum of the object together by recording the polarization angle information together with the hyperspectral data when the hyperspectral library is constructed. The data input / output unit 202 registers the polarization angle information input from the polarization angle control unit 201 and the hyperspectral data input from the camera control unit 203 together with the spectrum library 204 with polarization information.

  In addition, the data input / output unit 202 has a function of identifying and identifying an object using two indexes of a polarization angle and an object spectrum. The accuracy of the object identification process is increased by using two indexes of the polarization angle and the spectrum.

  The data input / output unit 202 extracts spectrum data corresponding to the polarization angle information input from the polarization angle control unit 201 from the spectrum library 204 with polarization information. Next, the data input / output unit 202 identifies and identifies the object by comparing the extracted spectral data with the input hyperspectral data.

  The data input / output unit 202 has a function of acquiring polarization angle information stored in the hyperspectral library corresponding to an object that the user wants to detect and identify. The polarization angle control unit 201 rotates the polarization filter 400 that is a polarizing plate via the polarization filter rotation device 300 based on the acquired polarization angle information. By acquiring hyperspectral data in a state where the rotation angle of the polarizing filter 400 is set to a predetermined value in advance, the object is efficiently detected.

  The camera control unit 203 has a function of outputting a control signal for controlling a parameter related to shooting of the camera 503 and a shooting signal of the camera 503 via a control line. Further, the camera control unit 203 has a function of receiving captured hyperspectral data from the camera 503 via a signal line.

  The spectrum library with polarization information 204 has a function of storing polarization angle information and hyperspectral data. When the spectrum library with polarization information 204 is constructed, hyperspectral data and information on the rotation angle of the polarization filter 400 are acquired.

  The polarization angle information corresponding to the acquired rotation angle is registered in the spectrum library 204 with polarization information together with the acquired hyperspectral data. As a result, the object identification device 200 including the spectrum library with polarization information 204 can identify hyperspectral data using the polarization angle information.

  In the present embodiment, the rotation angle of the polarization filter 400 indicated by the polarization angle information registered in the spectrum library with polarization information 204 is an angle at which the reflected light from the object is transmitted through the polarization filter 400 best. Note that the rotation angle of the polarization filter 400 indicated by the polarization angle information may be an angle other than the angle at which the reflected light from the object is transmitted through the polarization filter 400 best.

  As described above, a spectrum library including polarization angle information is constructed. In the collation processing and identification processing using the polarization angle information and the spectrum data in the present embodiment, the polarization information-added spectrum library 204 in which data is accumulated as described above is used.

  The mirror control unit 205 has a function of outputting a control signal for controlling the imaging angle of the scan mirror 600 to the scan mirror 600 via a control line. The scan mirror 600 rotates based on the output control signal.

  The hyperspectral camera 500 includes a lens 501, a spectroscope 502, and a camera 503. The lens 501 is connected to the spectroscope 502, and the spectroscope 502 is connected to the camera 503. The lens 501, the spectroscope 502, and the camera 503 operate as a unit.

  The camera 503 is communicably connected to the camera control unit 203 via a control line and a signal line. The camera 503 transmits the captured hyperspectral data to the camera control unit 203 via a signal line.

  The camera 503 is equipped with a CCD (Charge-Coupled Device) image sensor or a CMOS (Complementary Metal Oxide Semiconductor) image sensor, which is a two-dimensional image sensor. The light receiving sensor 504 shown in FIG. 2 corresponds to the mounted image sensor.

  The image sensor mounted on the camera 503 outputs an image having a horizontal axis on the horizontal axis and a wavelength axis on the vertical axis. That is, the image sensor corresponds to a line sensor with respect to a predetermined space axis. Even if the image output from the image sensor has only one-dimensional spatial information, the hyperspectral camera 500 can acquire the two-dimensional spatial information by using the scan mirror 600.

  The polarization filter 400, the polarization filter rotating device 300, and the angle detection device 301 are attached to the front end of the lens 501 of the hyperspectral camera 500, respectively.

  The polarizing filter 400 includes a surface fixed to the lens 501 and a rotatable surface. That is, the polarizing filter 400 has a double structure.

  The polarization filter rotating device 300 has a function of rotating the rotatable surface of the polarization filter 400. The polarization filter rotation device 300 is communicably connected to the polarization angle control unit 201 via a control line. The polarization filter rotating device 300 rotates the rotatable surface of the polarization filter 400 based on the control signal received from the polarization angle control unit 201 via the control line.

  The angle detection device 301 has a function of detecting the rotation angle from the reference angle of the rotatable surface of the polarizing filter 400. The angle detection device 301 is communicably connected to the polarization angle control unit 201 via a signal line. The angle detection device 301 transmits information on the detected rotation angle to the polarization angle control unit 201 via a signal line.

  The scan mirror 600 is communicably connected to the mirror control unit 205 via a control line. The scan mirror 600 rotates within a preset shooting angle range. Note that when the data to be registered in the spectrum library is created, the scan mirror 600 may not be used.

  The mirror control unit 205 can rotate the scan mirror 600 around the center. By rotating the scan mirror 600, the object identification device 200 can acquire two-dimensional spatial information. That is, if the scan mirror 600 is used, the fixedly installed hyperspectral camera 500 can acquire three-dimensional spectrum data having a two-dimensional space axis and a one-dimensional wavelength axis. The three-dimensional spectrum data is also called a hyperspectral cube.

  The display device 700 has a function of displaying the result of object identification processing by the object identification device 200 and the like.

  Hereinafter, a specific example of the hardware configuration of the object identification device 200 in the present embodiment will be described. FIG. 3 is an explanatory diagram showing a hardware configuration example of the object identification device according to the present invention.

  The object identification device 200 shown in FIG. 3 includes a CPU (Central Processing Unit) 211, a main storage unit 212, a communication unit 213, and an auxiliary storage unit 214. Moreover, you may provide the input part 215 for a person to operate, and the output part 216 for showing the process result or progress of a process content to a person.

  The main storage unit 212 is used as a data work area or a temporary data save area. The main storage unit 212 is, for example, a RAM (Random Access Memory).

  The communication unit 213 has a function of inputting and outputting data to and from peripheral devices via a wired or wireless network (information communication network).

  The auxiliary storage unit 214 is a storage device such as a hard disk device. The auxiliary storage unit 214 is, for example, a ROM (Read Only Memory).

  The input unit 215 has a function of inputting data and processing instructions. The input unit 215 is, for example, a keyboard or a mouse.

  The output unit 216 has a function of outputting data. The output unit 216 is a display device such as a liquid crystal display device or a printing device such as a printer.

  As shown in FIG. 3, in the object identification device 200, each component is connected to a system bus 217.

  The auxiliary storage unit 214 stores, for example, programs for realizing the polarization angle control unit 201, the data input / output unit 202, the camera control unit 203, and the mirror control unit 205 shown in FIG.

  The main storage unit 212 is used as a storage area of the spectrum library with polarization information 204, for example.

  The polarization angle control unit 201, the camera control unit 203, and the mirror control unit 205 transmit a control signal via the communication unit 213. Also, the polarization angle control unit 201 and the camera control unit 203 receive data via the communication unit 213.

  The object identification device 200 may be realized by hardware. For example, the object identification device 200 may be mounted with a circuit including hardware components such as an LSI (Large Scale Integration) in which a program for realizing the function shown in FIG. 2 is incorporated.

  Further, the object identification device 200 may be realized by software by causing the CPU 211 illustrated in FIG. 3 to execute a program that provides the functions of each component illustrated in FIG.

  When implemented by software, the CPU 211 loads a program stored in the auxiliary storage unit 214 to the main storage unit 212 and executes it, thereby controlling the operation of the object identification device 200, thereby realizing each function by software. Is done.

[Description of operation]
Hereinafter, the operation of the object identification device 200 of the present embodiment will be described with reference to FIGS.

  First, the operation of the object identification device 200 when the polarization information-added spectrum library 204 is constructed will be described with reference to FIGS. FIG. 4 is a flowchart showing the operation of the polarization information-added spectrum library construction process by the object identification device 200 in the first embodiment. FIG. 5 is an explanatory diagram showing data processing by the object identification device 200 when constructing a spectrum library with polarization information.

  The target sample reflects the light from the light source shown in FIG. 5 (step S101). The light emitted from the light source is light having a uniform wavelength, such as light emitted from a halogen lamp. The light reflected by the sample includes polarized light.

  The light reflected from the sample enters the polarizing filter 400 (step S102). The polarization filter 400 removes a polarization component that vibrates in a polarization direction different from the polarization direction of the polarization filter 400 from the incident light.

  The polarizing filter 400 is rotated by the polarizing filter rotating device 300. The polarization filter rotation device 300 sets the rotation angle of the polarization filter 400 so that the reflected light from the sample is most effectively transmitted through the polarization filter 400.

  The angle detection device 301 detects the rotation angle of the polarizing filter 400 when the reflected light from the sample is best transmitted through the polarizing filter 400. Next, the angle detection device 301 converts the detected rotation angle into an angle signal. The angle detection device 301 transmits the converted angle signal to the polarization angle control unit 201 via the signal line (step S103).

  The reflected light from the sample from which the polarization component that vibrates in a polarization direction different from the polarization direction of the polarization filter 400 is removed enters the lens 501 (step S104). The light collected by the lens 501 reaches the spectroscope 502.

  The reflected light is split by the spectroscope 502 (step S105). Each split light reaches the light receiving sensor 504 (step S106).

  The light receiving sensor 504 converts each received light from an optical signal to an electrical signal. Next, the light receiving sensor 504 creates hyperspectral data based on the converted electrical signal. The camera 503 transmits the created hyperspectral data to the camera control unit 203 (step S107).

  As shown in FIG. 5, the rotation angle information indicated by the angle signal transmitted in step S103 and the hyperspectral data transmitted in step S107 are merged and registered in the spectrum library 204 with polarization information.

  That is, the data input / output unit 202 registers the polarization angle information corresponding to the rotation angle indicated by the angle signal in the polarization information-added spectrum library 204 as the metadata of the corresponding hyperspectral data (step S108).

  The object identification device 200 repeatedly performs the processing from step S101 to step S108 for each target sample. The hyperspectral library including the polarization angle information is constructed by repeatedly performing the process shown in FIG. After registering all the hyperspectral data of the target sample, the object identification device 200 ends the polarization information-added spectrum library construction process.

  Note that the display device 700 displays to the user that the data has been registered correctly, that the data has not been registered correctly for some reason, or that the process has been completed in the spectrum library construction process with polarization information. May be.

  Next, the operation of the object identification device 200 when identifying an object by collating the acquired hyperspectral data with the data registered in the polarization information-added spectrum library 204 will be described with reference to FIGS. explain.

  FIG. 6 is a flowchart showing the operation of the object identification process by the object identification device 200 in the first embodiment. FIG. 7 is an explanatory diagram showing data processing by the object identification device 200 when collating the acquired hyperspectral data with the data registered in the polarization information-added spectrum library 204.

  The subject shown in FIG. 7 reflects the light from the light source shown in FIG. 7 (step S201). The light source is, for example, the sun or a halogen lamp. The light reflected by the subject includes polarized light.

  The light including the polarized light reflected from the subject is reflected by the scan mirror 600 (step S202). The light reflected by the scan mirror 600 enters the polarizing filter 400 (step S203).

  When light from the light source enters the polarizing filter 400, the polarization angle control unit 201 transmits a control signal including information on the rotation angle to the polarization filter rotating device 300. The polarization filter rotation device 300 that has received the control signal rotates the polarization filter 400 to an appropriate angle. A suitable angle is a value having a predetermined width.

  The angle detection device 301 measures the rotation angle of the polarizing filter 400. Next, the angle detection device 301 converts the measured rotation angle into an angle signal. The angle detection device 301 transmits the converted angle signal to the polarization angle control unit 201 via the signal line (step S204).

  The polarization filter 400 removes a polarization component that vibrates in a polarization direction different from the polarization direction of the polarization filter 400 from the incident light. That is, light having only a polarization component that vibrates in the same polarization direction as the polarization direction of the polarization filter 400 enters the lens 501 and is condensed (step S205).

  The light collected by the lens 501 enters the spectroscope 502. The reflected light is split by the spectroscope 502 (step S206). Each split light reaches the light receiving sensor 504 (step S207).

  The light receiving sensor 504 converts each received light from an optical signal to an electrical signal. Next, the light receiving sensor 504 creates hyperspectral data based on the converted electrical signal. The camera 503 transmits the created hyperspectral data to the camera control unit 203 (step S208).

  Next, the data input / output unit 202 uses the polarization angle information of the polarization filter 400 transmitted from the angle detection device 301 and the hyperspectral data transmitted from the camera 503 to perform a matching process and an identification process for identifying a subject. I do.

  As shown in FIG. 7, the data input / output unit 202 collates the transmitted polarization angle information of the polarization filter 400 with the polarization angle information registered in the spectrum library 204 with polarization information. As a result of the collation, the data input / output unit 202 extracts only the spectral data corresponding to the polarization direction indicated by the polarization angle information associated with the polarization direction of the polarization filter 400 indicated by the polarization angle information (step S209).

  Next, as shown in FIG. 7, the data input / output unit 202 collates the extracted spectrum data with the transmitted hyperspectral data for each pixel. As a result of the collation, the data input / output unit 202 sets the spectrum indicated by the spectrum data having a high degree of coincidence among the extracted spectrum data as the spectrum of the collated pixel (step S210). That is, the data input / output unit 202 assumes that the substance indicated by the pixel is a substance corresponding to a spectrum with a high degree of coincidence.

  The data input / output unit 202 performs collation processing and identification processing between the extracted spectrum data and the transmitted hyperspectral data for all pixels. After completing the matching process and the identification process for all the pixels, the object identification device 200 ends the object identification process.

  Note that the display device 700 may display to the user the result of the object identification process or the completion of the object identification process.

[Description of effects]
The object identification apparatus according to the present embodiment is extracted from a spectrum library, which is a target to be verified with, for example, a hyperspectral cube, when the data input / output unit 202 performs filtering using polarization information indicating the rotation angle of the polarization filter. Spectral data can be reduced. That is, the object identification device can perform the matching process and the identification process at a higher speed.

  In addition, the object identification system of the present embodiment can be easily obtained by setting the rotation angle of the polarization filter in advance by the polarization filter rotation device 300 using the polarization information of the spectrum data to be extracted from the hyperspectral library with polarization information. The target object can be identified by extracting the spectrum data.

Embodiment 2. FIG.
[Description of configuration]
Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 8 is an explanatory diagram showing an example of the second embodiment of the object identification system according to the present invention.

  As shown in FIG. 8, the object identification system 100 is mounted on a flying object. The object identification device 200 in this embodiment is connected to a hyperspectral sensor mounted on a flying object. Note that the flying object shown in FIG. 8 is, for example, an aircraft or a helicopter.

  FIG. 9 is an explanatory diagram showing another example of the second embodiment of the object identification system according to the present invention. FIG. 9 shows an object identification system 100 mounted on an aircraft.

  In the example shown in FIG. 9, the hyperspectral sensor mounted on the aircraft operates as a push bloom sensor in the traveling direction of the aircraft. The push bloom sensor is a sensor that performs scanning only in one direction (traveling direction) by arranging light detection elements for one line. Since scanning is performed on the image plane, no scanning optical system is required in the push bloom sensor.

  FIG. 10 is a block diagram showing a configuration example of the second embodiment of the object identification system according to the present invention. Since scanning is performed only in the traveling direction, the object identification system 100 of this embodiment does not include a scan mirror as shown in FIG.

Embodiment 3. FIG.
[Description of configuration]
Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 11 is an explanatory diagram showing an example of the third embodiment of the object identification system according to the present invention.

  As shown in FIG. 11, the object identification systems 100 to 101 are mounted on a ship. The object identification device 200 of the present embodiment is connected to a hyperspectral sensor mounted on a ship.

  The object identification device 200 according to the present embodiment is different from the ground-installed object identification device 200 according to the first embodiment in that it includes a shake correction unit. The fluctuation correcting unit has a function of absorbing the fluctuation of the ship with respect to the entire sensor.

  Next, the outline of the present invention will be described. FIG. 12 is a block diagram showing an outline of an object identification device according to the present invention. The object identification device 10 according to the present invention includes object information including hyperspectral data corresponding to a predetermined object and polarization information indicating the polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. The storage unit 11 (for example, the spectrum library with polarization information 204) to be stored and the object information corresponding to the polarization information acquired at the time of acquiring the hyperspectral data of the object to be identified are extracted from the storage unit 11, and the extracted object information And an extraction unit 12 (for example, a data input / output unit 202) that extracts object information corresponding to hyperspectral data acquired from the data.

  With such a configuration, the object identification device can identify an object in a short time using hyperspectral data.

  In addition, the object identification device 10 uses a polarization filter that is incident before light from the object to be identified received by the image sensor that acquires hyperspectral data is received by the image sensor, using polarization information included in the object information. A control unit (for example, a polarization angle control unit 201) that controls the above may be provided.

  With such a configuration, the object identification device can efficiently identify the target object.

  The object identification device 10 also stores a data storage unit (for example, data input / output) that stores object information including hyperspectral data corresponding to a predetermined object and polarization information acquired at the time of hyperspectral data acquisition in the storage unit 11. Part 202).

  With such a configuration, the object identification device can construct a hyperspectral library including polarization information.

  FIG. 13 is a block diagram showing another outline of the object identification device according to the present invention. The object identification device 20 according to the present invention includes object information including hyperspectral data corresponding to a predetermined object and polarization information indicating the polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. A storage unit 21 for storing (for example, a spectrum library with polarization information 204), a camera unit 22 (for example, a camera 503) for acquiring hyperspectral data of the object to be identified based on light from the object to be identified; Object information corresponding to the polarization information acquired when hyperspectral data of the object to be identified is acquired is extracted from the storage unit 21, and the object corresponding to the hyperspectral data acquired by the camera unit 22 from the extracted object information. And an extraction unit 23 (for example, a data input / output unit 202) that extracts information.

  With such a configuration, the object identification device can identify an object in a short time using hyperspectral data.

  The object identification device 20 controls the polarization filter that is incident before the light from the object to be identified received by the camera unit 22 is received by the camera unit 22 using the polarization information included in the object information. (E.g., a polarization angle control unit 201).

  With such a configuration, the object identification device can efficiently identify the target object.

  The object identification device 20 also stores a data storage unit (for example, data input / output) that stores object information including hyperspectral data corresponding to a predetermined object and polarization information acquired at the time of hyperspectral data acquisition in the storage unit 21. Part 202).

  With such a configuration, the object identification device can construct a hyperspectral library including polarization information.

  FIG. 14 is a block diagram showing an outline of an object identification system according to the present invention. An object identification system 30 according to the present invention is an object identification system including a polarizing filter 40 on which reflected light reflected by an object to be identified is incident, a camera 50, an object identification device 60, and a display device 70. Reference numeral 50 includes a camera unit 51 (for example, a camera 503) that acquires hyperspectral data of an object to be identified based on light transmitted through the polarizing filter 40. The object identification device 60 is a hyper corresponding to a predetermined object. A storage unit 61 (for example, a spectrum library 204 with polarization information) that stores object information including spectrum data and polarization information indicating the polarization direction of light from a predetermined object received by the image sensor that has acquired the hyperspectral data; An angle detector 62 (for example, an angle detector 301) that detects the rotation angle of the polarizing filter 40; The object information corresponding to the polarization information indicating the polarization direction corresponding to the rotation angle is extracted from the storage unit 61, and the object information corresponding to the hyperspectral data acquired by the camera unit 51 is extracted from the extracted object information. The display device 70 (for example, the display device 700) displays the extracted object information.

  With such a configuration, the object identification system can identify an object in a short time using hyperspectral data.

  Further, the object identification device 60 may include a control unit (for example, a polarization angle control unit 201) that controls the polarization filter 40 using polarization information included in the object information.

  With such a configuration, the object identification device can efficiently identify the target object.

  INDUSTRIAL APPLICABILITY The present invention can be suitably applied to the use of detection and identification of a vehicle or person camouflaged with grass or snow in the defense field, or the detection of a person disguised with a face mask. In addition, the present invention can be used for detection and identification of offshore floating objects such as extraction of floating persons on the ocean and offshore floating rings, and detection of persons in situations where it is difficult to visually detect persons such as during disasters. And can be suitably applied to identification applications. In addition, the present invention can be suitably applied to the use of identification of a disguised food or detection of a foreign substance in food.

10, 20, 60, 200 Object identification device 11, 21, 61 Storage unit 12, 23, 63 Extraction unit 62 Angle detection unit 22, 51 Camera unit 30, 100-101 Object identification system 40, 400 Polarization filter 50, 503 Camera 70, 700 Display device 201 Polarization angle control unit 202 Data input / output unit 203 Camera control unit 204 Spectrum library with polarization information 205 Mirror control unit 211 CPU
212 Main storage unit 213 Communication unit 214 Auxiliary storage unit 215 Input unit 216 Output unit 217 System bus 300 Polarizing filter rotation device 301 Angle detection device 500 Hyperspectral camera 501 Lens 502 Spectroscope 504 Light reception sensor 600 Scan mirror

Claims (10)

A storage unit for storing object information including hyperspectral data corresponding to a predetermined object; and polarization information indicating a polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data;
The object information corresponding to the hyperspectral data acquired from the extracted object information is extracted from the storage unit, the object information corresponding to the polarization information acquired when acquiring the hyperspectral data of the object to be identified An object identification apparatus comprising: an extraction unit that extracts information. A control unit that controls a polarization filter that is incident before light from an object to be identified received by an image sensor that acquires hyperspectral data is received by the image sensor, using polarization information included in the object information. The object identification device according to claim 1. The object identification device according to claim 1, further comprising: a data storage unit that stores object information including hyperspectral data corresponding to a predetermined object and polarization information acquired at the time of acquiring the hyperspectral data in a storage unit. . A storage unit for storing object information including hyperspectral data corresponding to a predetermined object; and polarization information indicating a polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data;
A camera unit for acquiring hyperspectral data of the object to be identified based on light from the object to be identified;
The hyperspectral data obtained by extracting the object information corresponding to the polarization information acquired when the hyperspectral data of the identification target object is acquired from the storage unit and acquired by the camera unit from the extracted object information An object identification apparatus comprising: an extraction unit that extracts the object information corresponding to 5. The object according to claim 4, further comprising: a control unit configured to control a polarization filter that is incident before light from the identification target object received by the camera unit is received by the camera unit, using polarization information included in the object information. Identification device. The object identification device according to claim 4 or 5, further comprising: a data storage unit that stores object information including hyperspectral data corresponding to a predetermined object and polarization information acquired at the time of acquiring the hyperspectral data in a storage unit. . An object identification system including a polarizing filter on which reflected light reflected by an object to be identified enters, a camera, an object identification device, and a display device,
The camera
A camera unit for acquiring hyperspectral data of the object to be identified based on light transmitted through the polarizing filter;
The object identification device includes:
A storage unit for storing object information including hyperspectral data corresponding to a predetermined object; and polarization information indicating a polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data;
An angle detector for detecting a rotation angle of the polarizing filter;
The object information corresponding to the polarization information indicating the polarization direction corresponding to the detected rotation angle is extracted from the storage unit, and the hyperspectral data acquired by the camera unit is extracted from the extracted object information. An extraction unit for extracting the corresponding object information,
The display device
An object identification system characterized by displaying the extracted object information. The object identification system according to claim 7, wherein the object identification device includes a control unit that controls the polarization filter using polarization information included in the object information. An object including a storage unit that stores object information including hyperspectral data corresponding to a predetermined object and polarization information indicating a polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. An object identification method executed in an identification device,
Extracting the object information corresponding to the polarization information acquired at the time of acquiring hyperspectral data of the object to be identified from the storage unit;
The object identification method, wherein the object information corresponding to the hyperspectral data acquired from the extracted object information is extracted. A computer including a storage unit that stores object information including hyperspectral data corresponding to a predetermined object and polarization information indicating a polarization direction of light from the predetermined object received by the image sensor that has acquired the hyperspectral data. An object identification program executed in
In the computer,
A first extraction process for extracting the object information corresponding to the polarization information acquired when the hyperspectral data of the object to be identified is acquired from the storage unit; and the hyperspectral data acquired from the extracted object information The object identification program for performing the 2nd extraction process which extracts the said object information corresponding to.

Images