JP2022021535A - Object detection device, system, and method - Google Patents

Abstract

To provide a technique capable of detecting an object in various environment.SOLUTION: An object detection device includes: a first image acquisition unit for acquiring a first image in which a predetermined space is photographed by a first imaging device that photographs an image by detecting light from an object in a predetermined first wavelength band; a second image acquisition unit for acquiring a second image in which the predetermined space is photographed by a second imaging device that photographs an image by detecting light from an object in a second wavelength band that is different from the first wavelength band; a determination unit for determining a usage image used for object detection from the first image and the second image; and a detection processing unit for detecting a predetermined target object that is a detection target included in the usage image.SELECTED DRAWING: Figure 2

Description

The present disclosure relates to a technique for detecting an object.

Patent Document 1 discloses an object detection device. The object detection device of Patent Document 1 detects an object included in an image acquired by a photographing device. The photographing device is a near-infrared camera or a visible light camera. That is, the object detection device of Patent Document 1 detects an object included in an image acquired by a near-infrared camera or a visible light camera.

Japanese Unexamined Patent Publication No. 2020-027405

The object detection device disclosed in Patent Document 1 acquires an image from an imaging device of either a near-infrared camera or a visible light camera, and detects an object in the image. For example, when the object detection device acquires an image of a visible light camera, a situation may occur in which an object that could be detected from an image of a near-infrared camera cannot be detected. On the contrary, by using the image of the near-infrared camera, a situation may occur in which an object that could be detected from the image of the visible light camera cannot be detected.

One object of the present disclosure is to provide a technique that enables detection of an object in various environments.

The object detection device according to one aspect of the present disclosure acquires a first image obtained by capturing a predetermined space by a first image pickup device that captures an image by detecting light from an object in a predetermined first wavelength band. A second image that acquires a second image of the space by a first image acquisition unit and a second image pickup device that captures an image by detecting light from an object in a second wavelength band different from the first wavelength band. An acquisition unit, a determination unit that determines a use image to be used for object detection from the first image and the second image, and a detection processing unit that detects a predetermined target object included in the use image. Have.

According to one aspect of the present disclosure, it is possible to provide a technique that enables detection of an object in various environments.

It is a schematic block diagram of an object detection system. It is a block diagram of an object detection device. It is a flowchart which shows a series of processing by a determination unit 13 and a detection processing unit 14. It is a flowchart of a rule-based object detection process. It is a flowchart of a rule-based object detection process. It is a flowchart of the area division processing. It is a figure which shows the use image of the object detection system. It is a figure which shows the hardware composition of the object detection apparatus. It is a flowchart which shows the series of processing by the determination unit 13 and the detection processing unit 14 in the modification.

Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of an object detection system. The object detection system is a system that detects a predetermined object to be detected (hereinafter, also referred to as "object") in the space 90, and is visible to the object detection device 10, the TOF (Time-of-Flit) sensor 81, and the TOF (Time-of-Flit) sensor 81. It has an optical camera 82 and a monitor 83. The TOF sensor 81 and the visible light camera 82 are installed in the space 90 for detecting the target object. The space 90 is, for example, a multimodal environment in which many types of edge devices are installed in the same space. People 91 and 92 exist in the space 90 as examples of objects. The TOF sensor 81 and the visible light camera 82 are connected to the object detection device 10 by a wired or wireless communication line.

The TOF sensor 81 is an imaging device provided with a near-infrared light source and captures an image in space 90 (hereinafter, also referred to as “near-infrared image”) by detecting light reflected by an object from the light source. .. As an example, the TOF sensor 81 also captures an image showing the distance to an object for each pixel (hereinafter, also referred to as a “distance image”). A thermography camera may be used instead of the TOF sensor 81 for capturing a near-infrared image. The thermography camera detects the intensity of far-infrared rays incident from an object and determines the intensity of far-infrared rays, that is, the temperature of the object for each pixel. Take a temperature distribution image showing the corresponding color. This temperature distribution image may be used instead of the near infrared image obtained by the TOF sensor 81.

The visible light camera 82 is an image pickup device that captures an image in space 90 (hereinafter, also referred to as “visible light image”) by detecting visible light incident from an object.

The object detection device 10 is, for example, a device that automatically selects suitable data from sensor data acquired in a space 90 in a multimodal environment and realizes good detection of a target object. The object detection device 10 is a device that determines an image to be used for detecting an object (hereinafter, also referred to as a "utilized image") from a near-infrared image and a visible light image, and detects a target object included in the used image. The object detection device 10 determines the image to be used based on, for example, one or a plurality of a near-infrared image, a visible light image, and a distance image.

The object detection device 10 is realized by, for example, a computer executing a software program. The configuration and arrangement of the computer are not particularly limited. For example, the computer may be an edge computer owned and managed by the user himself / herself, or may be a cloud computer provided on the cloud. Details of the configuration and operation of the object detection device 10 will be described later.

The monitor 83 is a display device that displays the detection result or the like by the object detection device 10 on the screen.

FIG. 2 is a block diagram of the object detection device. The object detection device 10 includes a near-infrared image acquisition unit 11, a near-infrared image acquisition unit 12, a determination unit 13, a detection processing unit 14, and a user interface (UI) unit 15.

The near-infrared image acquisition unit 11 acquires near-infrared image data from the TOF sensor 81. The near-infrared image acquisition unit 11 may also acquire distance image data from the TOF sensor 81. The visible light image acquisition unit 12 acquires the data of the visible light image from the visible light camera 82.

The determination unit 13 determines the image to be used from the near-infrared image and the visible light image based on any one or a plurality of the near-infrared image, the visible light image, and the distance image. The detection processing unit 14 analyzes the used image and detects the target object included in the used image. Details of a series of processes by the determination unit 13 and the detection processing unit 14 will be described later.

The user interface unit 15 is an interface unit for the object detection device 10 to present various information to the user and to acquire various information from the user. For example, the user interface unit 15 may generate a screen to be displayed on the monitor 83 of the touch panel display and acquire information input by a touch operation on the screen of the monitor 83.

FIG. 3 is a flowchart showing a series of processes by the determination unit 13 and the detection processing unit 14. First, the determination unit 13 calculates the depth of the space 90 in step 101. The method for calculating the depth of the space 90 is not particularly limited, but in the present embodiment, the depth of the space 90 is calculated based on the distance image. For example, the largest distance value (pixel value) in the distance image may be the depth of the space 90. As another method, for example, a near-infrared image may be used to estimate the distance (depth) from the focal position of the TOF sensor 81 and the degree of blurring of the near-infrared image. When a thermography camera is used instead of the TOF sensor 81, a distance image cannot be obtained, so this method is effective. The depth referred to here does not necessarily mean a distance in the horizontal direction, and may be a distance in another direction such as a vertical direction. Next, in step 102, the determination unit 13 determines whether or not the depth of the space 90 exceeds a predetermined threshold value. The threshold value may be set arbitrarily, for example, 5 meters.

If the depth of the space 90 exceeds the threshold value in the determination in step 102, the determination unit 13 determines the near-infrared image as the image to be used in step 103. Subsequently, the detection processing unit 14 detects the target object in the space 90 by executing the rule-based object detection processing in step 104. The rule-based object detection process is a process of detecting an object according to a preset rule. The rule-based object detection process does not need to prepare and learn learning data, and is a relatively simple method as compared with the learning-based object detection process described later. The details of the rule-based object detection process will be described later.

On the other hand, if the depth of the space is equal to or less than the threshold value in the determination in step 102, the determination unit 13 determines the visible light image as the image to be used in step 105. Subsequently, the detection processing unit 14 detects the target object in the space 90 by executing the learning-based object detection process in step 106. In the learning-based object detection process, learning such as deep learning is performed using the image data collected in advance as learning data to generate a prediction model, and the detection processing unit 14 is based on the prediction model from the used image. Detect the target object.

After step 104 or after step 106, the user interface unit 15 presents to the user in step 107 by displaying the detection result by the detection processing unit 14 on the screen of the monitor 83.

4 and 5 are flowcharts of the rule-based object detection process. The near-infrared image according to the present embodiment is assumed to be an image in which a plurality of frame images are connected in chronological order. Different processing is performed separately for a predetermined initial image and subsequent subsequent images. The initial image does not have to be one, and can be arbitrarily provided, such as being repeatedly provided at time intervals. For example, initial images may be provided at regular time intervals, or initial images may be provided at regular frame number intervals such as every 30 frames and every 60 frames.

FIG. 4 shows the processing for the initial image, and FIG. 5 shows the processing for the succeeding image.

Referring to FIG. 4, when the initial image is input in step 201, the detection processing unit 14 converts the initial image into gray scale in step 202. Subsequently, the detection processing unit 14 executes the preprocessing in step 203.

The preprocessing includes the area division process of step 204 and the labeling of step 205. In the area division process of step 204, the detection processing unit 14 identifies the area of the target object in the initial image. The details of the area division process will be described later. In the labeling of step 205, the detection processing unit 14 assigns a unique identification number for identifying the region of the target object identified in step 204. The area of the target object identified and given the identification number by this preprocessing becomes the tracking area to be tracked in the subsequent image.

After that, in step 206, the detection processing unit 14 saves the frame information including the identification number and the feature amount of the region of the target object identified from the initial image. The feature amount of the region is not particularly limited, but may be, for example, circularity, area, shading, center of gravity, and the like. Further, in step 207, the user interface unit 15 outputs the information obtained from the initial image by the series of processes of steps 201 to 206.

Referring to FIG. 5, when the succeeding image is input in step 301, the detection processing unit 14 converts the succeeding image into gray scale in step 302. Subsequently, the detection processing unit 14 executes the tracking process in step 303.

The tracking process includes calculation of the motion vector in step 304 and calculation of the motion vector for each tracking area in step 305. In the motion vector calculation in step 304, the detection processing unit 14 calculates a motion vector representing the movement amount in block units with respect to the image before the succeeding image to be processed. In calculating the movement vector for each tracking area in step 305, the detection processing unit 14 uses the motion vector calculated in step 304 to move the movement amount of each tracking area with respect to the image before the succeeding image to be processed. Calculate the motion vector representing.

After that, in step 306, the detection processing unit 14 saves the frame information including the identification number and the motion vector of the tracking area in the subsequent image. Further, in step 307, the user interface unit 15 outputs the information obtained from the subsequent image by the series of processes of steps 301 to 306.

FIG. 6 is a flowchart of the area division process. Here, as an example, it is assumed that the shape of the target object is not linear.

First, in step 401, the detection processing unit 14 acquires the initial image converted to grayscale in step 202 described above. Subsequently, in step 402, the detection processing unit 14 performs edge detection on the grayscale initial image, and generates an image representing the detected edge (hereinafter referred to as “edge image”). The edge is a portion where the change in shade is large, and tends to occur in the contour of the object. Further, the detection processing unit 14 applies a high-pass filter to the edge image to remove noise in step 403, and binarizes the edge image from which noise has been removed by the high-pass filter in step 404. Edges are emphasized in the binarized edge image.

The binarized edge image is subjected to two types of processing, steps 405 to 407 and steps 408 to 410.

In step 405, the detection processing unit 14 divides the binarized edge image into regions separated by pixel values. Subsequently, in step 406, the detection processing unit 14 extracts the feature amount of each region obtained in step 405. The feature amount of the region is not particularly limited, but may be, for example, circularity, area, shading, center of gravity, and the like. Further, in step 407, the detection processing unit 14 excludes a region having a feature quantity different from that of the target object from the region obtained in step 405 based on the feature quantity extracted in step 406, and the region of the target object. Narrow down the candidates.

On the other hand, in step 408, the detection processing unit 14 detects a straight line from the binarized edge image. Subsequently, in step 409, the detection processing unit 14 divides the linear region (hereinafter referred to as “straight line region”) detected in step 409 from the binarized edge image. Further, in step 410, the detection processing unit 14 extracts the feature amount of the linear region obtained in step 410. The feature amount in the linear region is not particularly limited, but may be, for example, circularity, area, shading, center of gravity, and the like.

After the two processings for the binarized edge image are completed, in step 411, the detection processing unit 14 excludes the linear region obtained in step 410 from the regions obtained in step 407 and remains. The area is the area of the target object. Then, in step 412, the detection processing unit 14 outputs the information of the region of the target object obtained by the series of processing of steps 401 to 411.

In the learning-based object detection process described above, the visible light image according to the present embodiment is assumed to be an image in which a plurality of frame images are continuous in time series, and a step is taken with respect to the region of the target object detected by the prediction model. The tracking process of 303 may be performed to calculate the motion vector of the region of the target object, and the motion vector may be saved as frame information in the same manner as in step 306.

FIG. 7 is a diagram showing a usage image of the object detection system. FIG. 7 illustrates a system for monitoring the living room. When the object monitoring system detects the intrusion of a suspicious person into the living room, it outputs the information of the detection result. Therefore, in this case, the person is the target object. A terminal device such as a personal computer or a smartphone can be connected to the object detection device 10, and the detection result can be confirmed in real time on the screen.

The TOF sensor 81 and the visible light camera 82 are installed so as to be able to photograph the living space 90, and the object detection device 10 is constructed on the cloud. The user 93 connects to the object detection device 10 using the personal computer 84 and confirms the detection result in real time.

On the screen of the personal computer 84, the place monitored by the system as the monitoring place 85 is the living room, and the visible light image taken by the visible light camera 82 is the image as the used image 86. It is displayed. Further, the screen shows level 1 as the abnormality detection level 87, the time-series object detection accuracy 88 of the near-infrared image and the visible light image, and the total real-time object detection accuracy 89. Further, the near-infrared image taken by the TOF sensor 81 and the visible light image taken by the visible light camera 82 may be switchably displayed on the screen of the personal computer 84.

FIG. 8 is a diagram showing a hardware configuration of the object detection device. FIG. 8 shows, as an example, the hardware configuration of the object detection device 10 by an edge computer.

The object detection device 10 has a processor 31, a main memory 32, a storage device 33, a communication device 34, an input device 35, and a display device 36 as hardware, and these are connected to the bus 37.

The storage device 33 stores data so that it can be written and read, and various data are recorded in the storage device 33. The processor 31 reads the data stored in the storage device 33 into the main memory 32, and executes the processing of the software program using the main memory 32. The processor 31 realizes the near-infrared image acquisition unit 11, the visible light image acquisition unit 12, the determination unit 13, the detection processing unit 14, and the user interface unit 15 shown in FIG. The communication device 34 transmits the information processed by the processor 31 via a communication network including wired, wireless, or both, and transmits the information received via the communication network to the processor 31. The received information is used by the processor 31 for software processing. The input device 35 is a device that receives information by operation input by an operator such as a keyboard and a mouse, and the input information is used for software processing by the processor 31. For example, the learning data is input to the storage device 33 via the communication device 34 and the input device 35. The display device 36 is a device that displays image or text information on a display screen as a result of software processing by the processor 31.

The embodiments described above are examples for the purposes of the present disclosure and are not intended to limit the scope of the invention to those embodiments. One of ordinary skill in the art can implement various other embodiments without departing from the scope of the invention.

In the above-described embodiment, as shown in FIG. 3, the object detection device 10 selects either a near-infrared image or a visible light image as a utilization image depending on the depth of the space 90. However, as a modification, either a near-infrared image or a visible light image may be selected as a utilization image depending on the distance to the object instead of the depth of the space 90. At that time, when a plurality of objects exist in the space 90, which of the near-infrared image and the visible light image should be used may be selected for each object or for each region in the space 90.

Hereinafter, a modified example will be described. The configurations of the object detection system and the object detection device in the modified example are basically the same as those of the present embodiment shown in FIGS. 1 and 2.

FIG. 9 is a flowchart showing a series of processes by the determination unit 13 and the detection processing unit 14 in the modified example. First, the determination unit 13 calculates the distance to the object existing in the space 90 in step 501. The method for calculating the distance to the object is not particularly limited, but in the present embodiment, the distance to the object is calculated based on the distance image. For example, a region in a distance image that is closer than a preset depth may be used as an object, and the distance (pixel value) may be used as the distance to the object. The number of objects is not limited to one, and if there are a plurality of objects, the distance to each object may be calculated and the following processing may be executed for each object. Further, the distance here does not necessarily mean a distance in the horizontal direction, and may be a distance in another direction such as a vertical direction.

Next, in step 502, the determination unit 13 determines whether or not the distance to the object exceeds a predetermined threshold value.

If the distance to the object exceeds the threshold value in the determination in step 502, the determination unit 13 determines the near-infrared image as the image to be used in step 503. Subsequently, the detection processing unit 14 detects the target object in the space 90 by executing the rule-based object detection processing in step 504. Rule-based object detection processing The same as described above.

On the other hand, if the distance to the object is equal to or less than the threshold value in the determination in step 502, the determination unit 13 determines the visible light image as the image to be used in step 505. Subsequently, the detection processing unit 14 detects the target object in the space 90 by executing the learning-based object detection process in step 506. The learning-based object detection process is the same as described above.

After step 504 or after step 506, the user interface unit 15 presents to the user in step 507 by displaying the detection result by the detection processing unit 14 on the screen of the monitor 83.

As described above, the object detection device 10 according to the present embodiment captures a predetermined space 90 by a first imaging device (TOF sensor 81) that captures an image by detecting light from an object in the first wavelength band. A first image acquisition unit (near infrared image acquisition unit 11) that acquires the first image, and a second image pickup device that captures an image by detecting light from an object in a second wavelength band different from the first wavelength band. The second image acquisition unit (visible light image acquisition unit 12) that acquires the second image obtained by capturing the space 90 by the (visible light camera 82), and the image to be used for object detection are determined from the first image and the second image. It has a determination unit 13 and a detection processing unit 14 that detects a predetermined target object included in the image to be detected. As a result, the image to be used is determined from the first image and the second image having different properties, and the object is detected from the used image, so that good detection of the object in various environments becomes possible.

Further, in the present embodiment, in the object detection device 10, the determination unit 13 acquires the depth of the space 90 or the distance to the object in the space 90, and determines the image to be used based on the depth or the distance. As a result, the first image or the second image is used depending on the environment in which the object to be detected is placed, so that the image suitable for the environment enables good detection of the object.

Further, in the present embodiment, the first image is an image in a predetermined first wavelength band, the second image is an image in a second wavelength band having a shorter wavelength than the first wavelength band, and the determination unit 13 is used. If the depth of the space 90 is equal to or less than a predetermined threshold value, the first image is selected, and if the depth of the space 90 is larger than the threshold value, the second image is selected. As a result, the first image with a relatively long wavelength is used in an environment where the object to be detected is placed at a relatively short distance, and in an environment where the object to be detected may be placed at a relatively long distance. Since a second image having a relatively short wavelength is used, it is possible to detect an object well by using an image with light in a wavelength band suitable for a distance.

Alternatively, in the present embodiment, the first image is an image in a predetermined first wavelength band, the second image is an image in a second wavelength band having a shorter wavelength than the first wavelength band, and the determination unit 13 is used. If the distance to the object is equal to or less than a predetermined threshold value, the first image may be selected, and if the distance to the object is larger than the threshold value, the second image may be selected. As a result, the first image with a relatively long wavelength is used when the object is located at a relatively short distance, and the second image with a relatively short wavelength is used when the object is located at a relatively long distance. An image with light in a suitable wavelength band according to the situation enables good detection of an object.

Further, in the object detection device 10 of the present embodiment, the detection processing unit 14 detects the target object from the first image on a rule basis when the first image is selected, and when the second image is selected, the second image. Detects the target object on a learning basis from. As a result, for the detection of objects at a relatively short distance from images taken at a long wavelength, the object is detected based on a relatively simple rule base, and objects at a relatively long distance from images taken at a short wavelength are detected. As for the detection, since the object is detected by a relatively precise learning base, the object can be detected with a desired accuracy according to the type of the image.

Further, in the object detection device 10 of the present embodiment, the first image is a plurality of images connected in time series, and the detection processing unit 14 detects an edge from a predetermined initial image in the first image and is based on the edge. By region division, the region of the target object is specified from the initial image, and the motion vector of the region of the target object is calculated in the subsequent images connected after the initial image.

Further, the detection processing unit 14 extracts a region having a feature amount possessed by the target object from the region generated by the region division based on the edge, and excludes the region having the feature amount not possessed by the target object to exclude the target object. Identify the area of. As a result, the area is narrowed down so as to leave the area having the feature amount possessed by the target object and exclude the area having the feature amount not possessed by the target object, so that the area of the target object can be satisfactorily specified.

Further, in the object detection device 10 of the present embodiment, the first image is an image by near infrared rays, and the second image is an image by visible light. For near-infrared images where the image data is relatively small and it is relatively difficult to collect sufficient training data, the object is detected on a rule basis, and the image data is relatively abundant and the training data can be collected relatively easily. Since the object is detected based on the learning base for the optical image, it is possible to detect the object satisfactorily with a desired accuracy by a suitable method according to the type of the image.

Alternatively, in the object detection device 10 of the present embodiment, the first image may be a temperature distribution image by far infrared rays, and the second image may be an image by visible light. In this case as well, the temperature distribution image by far infrared rays, which has relatively little image data and is relatively difficult to collect sufficient training data, detects the object on a rule basis, and the image data is relatively abundant and the training data is collected. Since the visible light image that can be collected relatively easily is detected by the learning base, it is possible to detect the object satisfactorily with a desired accuracy by a suitable method according to the type of the image.

Further, in the object detection device 10 of the present embodiment, image attribute information indicating whether the image to be used is an image taken by the first image pickup device or an image taken by the second image pickup device, and the first image pickup device and the second image pickup device. It also has a user interface unit that displays observation location information indicating the location where the device is installed and a usage image on the screen.

Further, in the object detection device 10 of the present embodiment, the first image pickup device includes a light source in the first wavelength band and captures the first image by detecting the light reflected from the object by the light from the light source. , A distance image showing the distance to the object for each pixel is further taken based on the time until the light from the light source is reflected by the object and returned, and the first image acquisition unit further acquires the distance image and the determination unit. 13 determines the image to be used from the first image and the second image based on the distance image. Since the image to be used is determined based on the distance image, good detection of objects at various distances becomes possible.

10 ... Object detection device, 11 ... Near infrared image acquisition unit, 12 ... Visible light image acquisition unit, 13 ... Judgment unit, 14 ... Detection processing unit, 15 ... User interface unit, 31 ... Processor, 32 ... Main memory, 33 ... Storage device, 34 ... communication device, 35 ... input device, 36 ... display device, 37 ... bus, 81 ... TOF sensor, 82 ... visible light camera, 83 ... monitor, 84 ... personal computer, 85 ... monitoring location, 86 ... use Image, 87 ... Abnormality detection level, 88 ... Object detection accuracy, 89 ... Object detection accuracy, 90 ... Space, 91, 92 ... Person, 93 ... User

Claims (13)

A first image acquisition unit that acquires a first image captured in a predetermined space by a first image pickup device that captures an image by detecting light from an object in a predetermined first wavelength band.
A second image acquisition unit that acquires a second image of the space taken by a second image pickup device that captures an image by detecting light from an object in a second wavelength band different from the first wavelength band.
A determination unit that determines a usage image to be used for object detection from the first image and the second image,
A detection processing unit that detects a predetermined target object included in the image to be detected, and a detection processing unit.
Object detection device with. The determination unit acquires the depth of the space or the distance to the object in the space, and determines the image to be used based on the depth or the distance.
The object detection device according to claim 1. The first image is an image in a predetermined first wavelength band, and the second image is an image in a second wavelength band having a shorter wavelength than the first wavelength band.
The determination unit selects the first image if the depth of the space is equal to or less than a predetermined threshold value, and selects the second image if the depth of the space is larger than the threshold value.
The object detection device according to claim 2. The first image is an image in a predetermined first wavelength band, and the second image is an image in a second wavelength band having a shorter wavelength than the first wavelength band.
The determination unit selects the first image if the distance to the object is equal to or less than a predetermined threshold value, and selects the second image if the distance to the object is larger than the threshold value.
The object detection device according to claim 2. When the first image is selected, the detection processing unit detects the target object from the first image on a rule basis, and when the second image is selected, the target object is learned from the second image. To detect,
The object detection device according to claim 3 or 4. The first image is a plurality of images connected in chronological order, and is a plurality of images.
The detection processing unit
An edge is detected from a predetermined initial image in the first image, and a region of a target object is specified from the initial image by region division based on the edge.
In the subsequent images following the initial image, the motion vector of the region of the target object is calculated.
The object detection device according to claim 5. The detection processing unit extracts a region having a feature amount possessed by the target object from a region generated by the region division based on the edge, and excludes a region having a feature amount not possessed by the target object. Identify the area of the target object,
The object detection device according to claim 6. The first image is an image by near infrared rays, and the second image is an image by visible light.
The object detection device according to claim 5. The first image is a temperature distribution image by far infrared rays, and the second image is an image by visible light.
The object detection device according to claim 5. Image attribute information indicating whether the image to be used is an image taken by the first image pickup device or an image taken by the second image pickup device, and a place where the first image pickup device and the second image pickup device are installed. It further has a user interface unit that displays the observation location information to be shown and the used image on the screen.
The object detection device according to claim 1. The first image pickup apparatus includes a light source in the first wavelength band, and captures the first image by detecting the light reflected from the object by the light from the light source, and the light from the light source is the light from the light source. Further, a distance image showing the distance to the object for each pixel is taken based on the time until the object reflects and returns.
The first image acquisition unit further acquires the distance image and obtains the distance image.
The determination unit determines the image to be used from the first image and the second image based on the distance image.
The object detection device according to claim 1. A first image pickup device that captures an image by detecting light from an object in a predetermined first wavelength band.
A second imaging device that captures an image by detecting light from an object in a second wavelength band different from the first wavelength band.
The first image of a predetermined space is acquired by the first image pickup device, the second image of the space is captured by the second image pickup device, and the object is detected from the first image and the second image. An object detection device that determines a usage image to be used and detects a predetermined target object included in the usage image.
Object detection system with. The computer
An image is taken by detecting light from an object in a predetermined first wavelength band. A first image in which a predetermined space is photographed is acquired by a first image pickup device.
A second image of the space is acquired by a second image pickup device that captures an image by detecting light from an object in a second wavelength band different from the first wavelength band.
The image to be used for object detection is determined from the first image and the second image, and the image to be used is determined.
Detects a predetermined target object included in the used image, which is a detection target.
An object detection method that does that.

Images