Object detection device and object detection method

Abstract

An object detection device (2) is configured to comprise: a distance information acquisition unit (11) that acquires distance information indicating a distance measurement result from a sensor (1) that measures the distance to an object present in an observation target region or the distance to the background of the observation target region; a comparison unit (12) that compares the distance information acquired by the distance information acquisition unit (11) when the object is not present in the observation target region with the distance information acquired by the distance information acquisition unit (11) when object detection processing is performed; and an object detection unit (13) that detects the object on the basis of the result of the distance information comparison performed by the comparison unit (12).

Description

Object detection device and object detection method

[0001] This disclosure relates to an object detection device and an object detection method.

[0002] There are object detection devices that detect objects present in an observation area. As an example of such an object detection device, Patent Document 1 discloses a processing unit that provides a convolutional neural network with two-dimensional image data indicating the distance to objects present in the observation area, and obtains object detection results from the convolutional neural network.

[0003] International Publication No. 2019-194255

[0004] The processing unit disclosed in Patent Document 1 had a problem in that, due to the long distance to the object present in the observation target area, the size of the object on the 2D image data became small, which sometimes resulted in a deterioration of the object detection accuracy.

[0005] This disclosure was made to solve the above-mentioned problems, and aims to provide an object detection device that can suppress the deterioration of object detection accuracy even when the distance to the object in the observation target area is long.

[0006] The object detection device according to this disclosure includes: a distance information acquisition unit that acquires distance information indicating the distance measurement result from a sensor that measures the distance to an object present in the observation target area or the distance to the background of the observation target area; a comparison unit that compares the distance information acquired by the distance information acquisition unit with the distance information acquired by the distance information acquisition unit when an object is not present in the observation target area and performs object detection processing; and an object detection unit that detects an object based on the comparison result of the distance information by the comparison unit.

[0007] According to this disclosure, even if the distance to an object in the observation area is long, the degradation of object detection accuracy can be suppressed.

[0008] This is a configuration diagram showing the object detection device 2 according to Embodiment 1. This is a hardware configuration diagram showing the hardware of the object detection device 2 according to Embodiment 1. This is a hardware configuration diagram of a computer when the object detection device 2 is implemented by software or firmware, etc. Distance measurement result by sensor 1 (θ, φ, z ｔ０ This is an explanatory diagram showing the remeasurement result of the distance by sensor 1 (θ, φ, z). ｔ１ This is an explanatory diagram showing the following: This is an explanatory diagram showing the distance measurement result (θ, φ, ∞) by the sensor when the light from sensor 1 is not reflected by the background. This is a flowchart of the object detection method, which is the processing procedure of the object detection device 2. This is an explanatory diagram showing an example of background depth image generation by background depth image generation unit 12a. This is an explanatory diagram showing an example of object depth image generation by object depth image generation unit 12b. This is a configuration diagram showing the object detection device 2 according to Embodiment 2. This is a hardware configuration diagram showing the hardware of the object detection device 2 according to Embodiment 2.

[0009] To provide a more detailed explanation of this disclosure, the forms for implementing this disclosure will be described below with reference to the attached drawings.

[0010] Embodiment 1. Figure 1 is a configuration diagram showing an object detection device 2 according to Embodiment 1. Figure 2 is a hardware configuration diagram showing the hardware of the object detection device 2 according to Embodiment 1. In Figure 1, the sensor 1 is implemented by, for example, a LiDAR (Light Detection and Ranging), a depth camera, or a stereo camera. The sensor 1 measures the distance to an object present in the observation target area, or the distance to the background of the observation target area. The sensor 1 outputs distance information indicating the distance measurement result to the object detection device 2.

[0011] The object detection device 2 detects objects present in the observation target area based on distance information output from the sensor 1. The object detection device 2 comprises a distance information acquisition unit 11, a comparison unit 12, and an object detection unit 13.

[0012] The distance information acquisition unit 11 is implemented, for example, by the distance information acquisition circuit 21 shown in Figure 2. The distance information acquisition unit 11 acquires distance information from the sensor 1, indicating the measurement result of the distance. The distance information acquisition unit 11 outputs the distance information to the comparison unit 12.

[0013] The comparison unit 12 is implemented, for example, by the comparison circuit 22 shown in Figure 2. The comparison unit 12 includes a background depth image generation unit 12a, an object depth image generation unit 12b, and a comparison processing unit 12c. The comparison unit 12 acquires distance information acquired by the distance information acquisition unit 11 (hereinafter referred to as "first distance information") when an object is not present in the observation target area, and distance information acquired by the distance information acquisition unit 11 (hereinafter referred to as "second distance information") when performing object detection processing. The comparison unit 12 compares the first distance information and the second distance information and outputs the comparison result of the distance information to the object detection unit 13.

[0014] The background depth image generation unit 12a acquires first distance information from the distance information acquisition unit 11. Based on the first distance information, the background depth image generation unit 12a generates a background depth image and outputs the background depth image to the comparison processing unit 12c. The object depth image generation unit 12b acquires second distance information from the distance information acquisition unit 11. Based on the second distance information, the object depth image generation unit 12b generates an object depth image and outputs the object depth image to the comparison processing unit 12c. The comparison processing unit 12c acquires the background depth image from the background depth image generation unit 12a and the object depth image from the object depth image generation unit 12b. The comparison processing unit 12c compares the background depth image and the object depth image and outputs the depth image comparison result to the detection processing unit 13a of the object detection unit 13.

[0015] The object detection unit 13 is implemented, for example, by the object detection circuit 23 shown in Figure 2. The object detection unit 13 includes a detection processing unit 13a. The object detection unit 13 obtains the comparison result of distance information from the comparison unit 12. The object detection unit 13 detects an object based on the comparison result of the distance information. The object detection result by the object detection unit 13 is displayed, for example, on a display device (not shown).

[0016] The detection processing unit 13a obtains the comparison results of the depth images from the comparison processing unit 12c. Based on the comparison results of the depth images, the detection processing unit 13a detects an object.

[0017] In Figure 1, it is assumed that the distance information acquisition unit 11, comparison unit 12, and object detection unit 13, which are components of the object detection device 2, are each implemented by dedicated hardware as shown in Figure 2. That is, it is assumed that the object detection device 2 is implemented by a distance information acquisition circuit 21, a comparison circuit 22, and an object detection circuit 23. The distance information acquisition circuit 21, comparison circuit 22, and object detection circuit 23 can be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0018] The components of the object detection device 2 are not limited to those implemented by dedicated hardware; the object detection device 2 may also be implemented by software, firmware, or a combination of software and firmware. The software or firmware is stored as a program in the computer's memory. The computer refers to the hardware that executes the program, and includes, for example, a CPU (Central Processing Unit), GPU (Graphics Processing Unit), central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, processor, or DSP (Digital Signal Processor).

[0019] Figure 3 is a hardware configuration diagram of a computer when the object detection device 2 is implemented by software or firmware. When the object detection device 2 is implemented by software or firmware, a program that causes the computer to execute the respective processing procedures in the distance information acquisition unit 11, the comparison unit 12, and the object detection unit 13 is stored in memory 31. The computer's processor 32 then executes the program stored in memory 31.

[0020] Furthermore, Figure 2 shows an example in which each component of the object detection device 2 is implemented by dedicated hardware, and Figure 3 shows an example in which the object detection device 2 is implemented by software or firmware, etc. However, this is only one example, and some components of the object detection device 2 may be implemented by dedicated hardware, while the remaining components may be implemented by software or firmware, etc.

[0021] Next, the operation of the object detection device 2 shown in Figure 1 will be explained. When no object is present in the observation target area, sensor 1 measures the distance to the background of the observation target area. When sensor 1 is implemented by LiDAR, as shown in Figure 4, in the direction of light emission from sensor 1 (θ, φ), the distance z from sensor 1 is measured. ｔ０ If a background exists at the position, sensor 1 will measure the distance as (θ, φ, z ｔ０ Figure 4 shows the first distance information (θ, φ, z) measured by the sensor 1, output to the object detection device 2. ｔ０ This is an explanatory diagram showing ). In Figure 4, θ indicates the azimuth direction, φ indicates the elevation angle direction, and z ｔ０ This indicates the distance from sensor 1 to the background at time t0. Sensor 1 measures the distance to the background located in the radial direction (θ, φ) while switching between the azimuth direction θ and the elevation direction.

[0022] After time t0, when the object is not present in the observation target area, the sensor 1 measures again the distance to the background of the observation target area. At time t0, the background that was present in the observation target area disappears, and a new background may appear in the observation target area at time t1 after time t0. In such a case, as shown in FIG. 5, in the radiation direction (θ, φ) of the light from the sensor 1, if a new background appears at a position at a distance z from the sensor 1, the sensor 1 outputs, as the distance measurement result, first distance information indicating (θ, φ, z) to the object detection device 2. FIG. 5 is an explanatory diagram showing the result of re-measuring the distance (θ, φ, z) by the sensor 1. In FIG. 5, θ indicates the azimuth direction, φ indicates the elevation angle direction, and z indicates the distance from the sensor 1 to the background at time t1. The sensor 1 re-measures the distance to the background existing in the radiation direction (θ, φ) while switching between the azimuth direction θ and the elevation angle direction. ｔ１ from the sensor 1 to the object detection device 2. ｔ１ ) as the first distance information indicating. FIG. 5 is an explanatory diagram showing the result of re-measuring the distance (θ, φ, z) by the sensor 1. ｔ１ ) from the sensor 1. In FIG. 5, θ indicates the azimuth direction, φ indicates the elevation angle direction, and z ｔ１ indicates the distance from the sensor 1 to the background at time t1. The sensor 1 re-measures the distance to the background existing in the radiation direction (θ, φ) while switching between the azimuth direction θ and the elevation angle direction.

[0023] As shown in FIG. 6, when the light from the sensor 1 is not reflected by the background, the sensor 1 outputs, as the distance measurement result, distance information indicating (θ, φ, ∞) to the object detection device 2. When the distance from the sensor 1 to the background is, for example, very long, the light from the sensor 1 may not be reflected by the background. Also, when the background is a substance that transmits light, the light from the sensor 1 may not be reflected by the background. FIG. 6 is an explanatory diagram showing the result of measuring the distance (θ, φ, ∞) by the sensor when the light from the sensor 1 is not reflected by the background.

[0024] FIG. 7 is a flowchart showing an object detection method which is a processing procedure of the object detection device 2. The distance information acquisition unit 11 acquires, at time t0 when the object is not present in the observation target area, first distance information indicating the distance measurement result (θ, φ, z) at time t0 from the sensor 1 (step ST1 in FIG. 7). The distance information acquisition unit 11 outputs the first distance information to the comparison unit 12. In a memory (not shown) of the comparison unit 12, the distance measurement result (θ, φ, z) at time t0 is recorded. ｔ０ ) as the first distance information indicating. ｔ０ ) at time t0 is recorded.

[0025] The sensor 1 measures the distance to an object existing in the observation target area when the object detection device 2 performs object detection processing at time t within the period from time t0 to time t1 after time t0. The sensor 1 outputs, as a measurement result of the distance, the second distance information indicating (θ, φ, z ｔ ) to the object detection device 2.

[0026] When the object detection device 2 performs object detection processing, the distance information acquisition unit 11 acquires, from the sensor 1, the second distance information indicating the measurement result of the distance (θ, φ, z ｔ ) at time t (step ST2 in FIG. 7). The distance information acquisition unit 11 outputs the second distance information to the comparison unit 12.

[0027] As shown in FIG. 8, the background depth image generation unit 12a of the comparison unit 12 generates a background depth image based on the measurement result of the distance (θ, φ, z ｔ０ ) at time t0 (step ST3 in FIG. 7). Since the generation process of the background depth image itself is a known technique, a detailed description thereof is omitted. FIG. 8 is an explanatory diagram showing an example of generating a background depth image by the background depth image generation unit 12a. In FIG. 8, the horizontal axis represents the azimuth direction θ, and the vertical axis represents the elevation angle direction φ. The background depth image generation unit 12a outputs the background depth image to the comparison processing unit 12c.

[0028] The object depth image generation unit 12b of the comparison unit 12 acquires the second distance information from the distance information acquisition unit 11. As shown in FIG. 9, the object depth image generation unit 12b generates an object depth image based on the second distance information (step ST4 in FIG. 7). Since the generation process of the object depth image itself is a known technique, a detailed description thereof is omitted. FIG. 9 is an explanatory diagram showing an example of generating an object depth image by the object depth image generation unit 12b. In FIG. 9, the horizontal axis represents the azimuth direction θ, and the vertical axis represents the elevation angle direction φ. The object depth image generation unit 12b outputs the object depth image to the comparison processing unit 12c.

[0029] The comparison processing unit 12c acquires a background depth image from the background depth image generation unit 12a and an object depth image from the object depth image generation unit 12b. The comparison processing unit 12c compares the background depth image and the object depth image (step ST5 in Figure 7). The comparison processing unit 12c outputs the depth image comparison result to the detection processing unit 13a.

[0030] The detection processing unit 13a obtains the comparison results of the depth images from the comparison processing unit 12c. Based on the comparison results of the depth images, the detection processing unit 13a detects objects. The object detection process by the detection processing unit 13a will be described in detail below.

[0031] In the direction of light emission from sensor 1 (θ, φ), the comparison result of the depth image is at distance z ｔ <distance z ｔ０ If it indicates that (in step ST6: YES in Figure 7), the detection processing unit 13a will determine the distance z from sensor 1 in the direction of light emission (θ, φ). ｔ The system determines that an object exists at the position (step ST7 in Figure 7). In this case, the detection processing unit 13a has detected an object that exists in the observation target area. In the direction of light emission from sensor 1 (θ, φ), the comparison result of the depth images is at distance z ｔ ≧distance z ｔ０ If it indicates that (step ST6 in Figure 7: NO), the detection processing unit 13a determines that no object exists in the direction of light emission (θ, φ) (step ST8 in Figure 7). In this case, the detection processing unit 13a will not detect any object present in the observation target area. The detection processing unit 13a determines whether or not an object exists in the direction of emission (θ, φ) while switching between the azimuth direction θ and the elevation direction.

[0032] Here, the comparison results of the depth images are at distance z ｔ <distance z ｔ０ If it indicates that this is the case, the detection processing unit 13a has determined that an object exists in the direction of light emission (θ, φ). However, this is just one example, and the comparison result of the depth images is at distance z ｔ +Threshold < distance z ｔ０If it indicates that such an object exists, the detection processing unit 13a may determine that an object exists in the direction of light emission (θ, φ). The threshold value is greater than 0 and may be stored in the internal memory of the detection processing unit 13a or provided from outside the object detection device 2.

[0033] Subsequently, when the object is not present in the observation area at time t1, the distance information acquisition unit 11 obtains the remeasurement result of the distance at time t1 (θ, φ, z) from the sensor 1. ｔ１ The first distance information is obtained, which shows the distance measurement result (θ, φ, z) at time t1. The distance information acquisition unit 11 outputs the first distance information to the comparison unit 12. The memory of the comparison unit 12 (not shown) stores the distance measurement result (θ, φ, z) at time t1. ｔ１ ) is recorded.

[0034] Sensor 1 measures the distance to an object present in the observation target area at time t' after time t1 when the object detection device 2 performs object detection processing. Sensor 1 uses the distance measurement result to determine (θ, φ, z ｔ The second distance information, which indicates '), is output to the object detection device 2.

[0035] The distance information acquisition unit 11, when the object detection device 2 performs object detection processing, obtains the distance measurement result (θ, φ, z) at time t' from the sensor 1. ｔ The distance information acquisition unit 11 obtains a second distance information indicating '). The distance information acquisition unit 11 outputs the second distance information to the comparison unit 12.

[0036] The background depth image generation unit 12a of the comparison unit 12 generates the remeasurement result of the distance at time t1 (θ, φ, z ｔ１ Based on this, a background depth image is generated. As a result, the background depth image at time t0 is updated to the background depth image at time t1. The background depth image generation unit 12a outputs the updated background depth image to the comparison processing unit 12c.

[0037] The object depth image generation unit 12b of the comparison unit 12 acquires second distance information at time t' from the distance information acquisition unit 11. Based on the second distance information at time t', the object depth image generation unit 12b generates a depth image of the object. The object depth image generation unit 12b outputs the depth image of the object to the comparison processing unit 12c.

[0038] The comparison processing unit 12c obtains the updated background depth image from the background depth image generation unit 12a and the object depth image at time t' from the object depth image generation unit 12b. The comparison processing unit 12c compares the updated background depth image with the object depth image at time t'. The comparison processing unit 12c outputs the depth image comparison result to the detection processing unit 13a.

[0039] The detection processing unit 13a obtains the comparison results of the depth images from the comparison processing unit 12c. Based on the comparison results of the depth images, the detection processing unit 13a detects objects. The object detection process by the detection processing unit 13a will be described in detail below.

[0040] In the direction of light emission from sensor 1 (θ, φ), the comparison result of the depth image is at distance z ｔ '<distance z ｔ１ If it indicates that an object exists in the direction of light emission (θ, φ), the detection processing unit 13a determines that an object exists in the direction of light emission (θ, φ). In the direction of light emission (θ, φ) from sensor 1, the comparison result of the depth images is at distance z ｔ '≧distance z ｔ１ If it indicates that the object is not present, the detection processing unit 13a determines that no object exists in the direction of light emission (θ, φ). The detection processing unit 13a determines whether or not an object exists in the direction of emission (θ, φ) while switching between the azimuth direction θ and the elevation direction.

[0041] In the above embodiment 1, the object detection device 2 is configured to include: a distance information acquisition unit 11 that acquires distance information indicating the distance measurement result from a sensor 1 that measures the distance to an object present in the observation target area or the distance to the background of the observation target area; a comparison unit 12 that compares the distance information acquired by the distance information acquisition unit 11 with the distance information acquired by the distance information acquisition unit 11 when no object is present in the observation target area and when performing object detection processing; and an object detection unit 13 that detects an object based on the comparison result of the distance information by the comparison unit 12. Therefore, the object detection device 2 can suppress the deterioration of object detection accuracy even when the distance to an object present in the observation target area is long.

[0042] In Embodiment 1, the object detection device 2 is configured such that the distance information acquisition unit 11 repeatedly acquires distance information from the sensor 1 when no object is present in the observation target area, and the background depth image generation unit 12a generates a background depth image based on the distance information acquired by the distance information acquisition unit 11 each time distance information is acquired. Therefore, the object detection device 2 can suppress the deterioration of object detection accuracy even when the background of the observation target area changes.

[0043] In the object detection device 2 shown in Figure 1, the comparison processing unit 12c compares the background depth image with the object depth image. The comparison processing unit 12c may also perform filtering on the object depth image and compare the background depth image with the object depth image after filtering. The filter used by the comparison processing unit 12c for filtering is a filter to exclude the detection of objects that are unnecessary for the system. If the object depth image is an object that is unnecessary for the system, the object depth image after filtering will be a depth image that does not contain such an object. Examples of objects that are unnecessary for the system include objects outside the observation target area, objects that are small in size, or objects whose aspect ratio is outside a predetermined aspect ratio range. If the object to be detected is, for example, a human, the aspect ratio of a horizontally elongated object is outside the predetermined aspect ratio range. By performing filtering on the object depth image, the comparison processing unit 12c can avoid detecting objects that are unnecessary for the system.

[0044] In the object detection device 2 shown in Figure 1, the distance information acquisition unit 11 receives first distance information (θ, φ, z) from the sensor 1. ｔ０ ) and second distance information (θ, φ, z ｔ Obtain each of the following and the first distance information (θ, φ, z ｔ０ ) and second distance information (θ, φ, z ｔ Each of these is output to the comparison unit 12. The distance information acquisition unit 11 may convert the Cartesian coordinate system distance information to spherical coordinate system distance information when Cartesian coordinate system distance information is output from the sensor 1 as first distance information and second distance information. The following equations (1) to (3) show the conversion of the first distance information, and the following equations (4) to (6) show the conversion of the second distance information. (x, y, z １ ) is the first distance information in the Cartesian coordinate system, and (x, y, z ２ ) is the second distance information in the Cartesian coordinate system. The distance information acquisition unit 11 acquires the first distance information (θ, φ, z) in the spherical coordinate system. ｔ０ ) and second distance information (θ, φ, z) in spherical coordinate system ｔ ) is output to the comparison unit 12.

[0045]

[0046] Embodiment 2. Embodiment 2 describes an object detection device 2 that includes an object identification unit 14 that acquires information indicating the type of object.

[0047] Figure 10 is a configuration diagram showing the object detection device 2 according to Embodiment 2. In Figure 10, the same reference numerals as in Figure 1 indicate the same or corresponding parts, so a detailed explanation is omitted. Figure 11 is a hardware configuration diagram showing the hardware of the object detection device 2 according to Embodiment 2. In Figure 11, the same reference numerals as in Figure 2 indicate the same or corresponding parts, so a detailed explanation is omitted. The object detection device 2 includes a distance information acquisition unit 11, a comparison unit 12, an object detection unit 13, and an object identification unit 14. The learning model 3 is implemented, for example, by a neural network. During training, the learning model 3 is given training data including depth images of objects and information indicating the type of object, and learns the type of object. During inference, when the learning model 3 is given a depth image of an object, it outputs information indicating the type of object. In Figure 10, the learning model 3 is provided outside the object detection device 2. However, this is just one example, and the learning model 3 may be provided inside the object detection device 2.

[0048] The object identification unit 14 is implemented, for example, by the object identification circuit 24 shown in Figure 11. The object identification unit 14 acquires a depth image of an object from the object depth image generation unit 12b. The object identification unit 14 provides the depth image of the object to the learning model 3 and acquires information indicating the type of object from the learning model 3. The information indicating the type acquired by the object identification unit 14 is displayed, for example, on a display device (not shown).

[0049] In Figure 10, it is assumed that the distance information acquisition unit 11, comparison unit 12, object detection unit 13, and object identification unit 14, which are components of the object detection device 2, are each implemented by dedicated hardware as shown in Figure 11. That is, it is assumed that the object detection device 2 is implemented by a distance information acquisition circuit 21, a comparison circuit 22, an object detection circuit 23, and an object identification circuit 24. Each of the distance information acquisition circuit 21, comparison circuit 22, object detection circuit 23, and object identification circuit 24 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination thereof.

[0050] The components of the object detection device 2 are not limited to those implemented by dedicated hardware; the object detection device 2 may also be implemented by software, firmware, or a combination of software and firmware. When the object detection device 2 is implemented by software or firmware, a program that causes a computer to execute the respective processing procedures in the distance information acquisition unit 11, comparison unit 12, object detection unit 13, and object identification unit 14 is stored in the memory 31 shown in Figure 3. Then, the processor 32 shown in Figure 3 executes the program stored in the memory 31.

[0051] Furthermore, Figure 11 shows an example in which each component of the object detection device 2 is implemented by dedicated hardware, and Figure 3 shows an example in which the object detection device 2 is implemented by software or firmware, etc. However, this is only one example, and some components of the object detection device 2 may be implemented by dedicated hardware, while the remaining components may be implemented by software or firmware, etc.

[0052] Next, the operation of the object detection device 2 shown in Figure 10 will be described. Except for the object identification unit 14, it is the same as the object detection device 2 shown in Figure 1. Therefore, only the operation of the object identification unit 14 will be described here.

[0053] When an object is detected by the object detection unit 13, the object identification unit 14 obtains a depth image of the object from the object depth image generation unit 12b. The object identification unit 14 provides the object depth image to the learning model 3. When the learning model 3 receives the object depth image from the object identification unit 14, it outputs information indicating the type of object to the object identification unit 14. The object identification unit 14 obtains information indicating the type of object from the learning model 3. The information indicating the type obtained by the object identification unit 14 is displayed, for example, on a display device (not shown).

[0054] In the above embodiment 2, the object detection device 2 is configured to include an object identification unit 14 that provides a depth image of an object generated by the object depth image generation unit 12b to a learning model 3 and acquires information indicating the type of object from the learning model 3. Therefore, the object detection device 2 can suppress the deterioration of object detection accuracy even when the background of the observation target area changes, and can also identify the type of object.

[0055] Furthermore, this disclosure allows for free combination of each embodiment, modification of any component in each embodiment, or omission of any component in each embodiment.

[0056] The object detection device of this disclosure can be used as a device for accurately detecting objects located at a long distance from the sensor.

[0057] 1 Sensor, 2 Object detection device, 3 Learning model, 11 Distance information acquisition unit, 12 Comparison unit, 12a Background depth image generation unit, 12b Object depth image generation unit, 12c Comparison processing unit, 13 Object detection unit, 13a Detection processing unit, 14 Object identification unit, 21 Distance information acquisition circuit, 22 Comparison circuit, 23 Object detection circuit, 24 Object identification circuit, 31 Memory, 32 Processor.

Claims

1. An object detection device comprising: a distance information acquisition unit that acquires distance information indicating the distance measurement result from a sensor that measures the distance to an object present in the observation target area, or the distance to the background of the observation target area; a comparison unit that compares the distance information acquired by the distance information acquisition unit with the distance information acquired by the distance information acquisition unit when the object is not present in the observation target area, when performing object detection processing; and an object detection unit that detects the object based on the comparison result of the distance information by the comparison unit.

2. The object detection device according to claim 1, wherein the comparison unit comprises: a background depth image generation unit that generates a background depth image based on distance information acquired by the distance information acquisition unit when the object is not present in the observation target area; an object depth image generation unit that generates a depth image of the object based on distance information acquired by the distance information acquisition unit when performing object detection processing; and a comparison processing unit that compares the background depth image generated by the background depth image generation unit with the depth image of the object generated by the object depth image generation unit, and the object detection unit comprises a detection processing unit that detects the object based on the comparison result of the depth images by the comparison processing unit.

3. The object detection device according to claim 2, characterized in that the distance information acquisition unit repeatedly acquires distance information from the sensor when the object is not present in the observation target area, and the background depth image generation unit generates a background depth image based on the distance information each time distance information is acquired by the distance information acquisition unit.

4. The object detection apparatus according to claim 2 or 3, characterized in that the comparison processing unit performs filtering on the depth image of the object generated by the object depth image generation unit, and compares the background depth image generated by the background depth image generation unit with the depth image of the object after filtering.

5. The object detection device according to any one of claims 2 to 4, characterized in that the distance information acquisition unit converts the orthogonal coordinate distance information acquired from the sensor into spherical coordinate distance information, and outputs the spherical coordinate distance information to the comparison unit.

6. An object detection device according to any one of claims 2 to 5, further comprising an object identification unit that provides a depth image of the object generated by the object depth image generation unit to a learning model and obtains information indicating the type of object from the learning model.

7. An object detection method comprising: a distance information acquisition unit acquires distance information indicating the distance measurement result from a sensor that measures the distance to an object present in the observation target area or the distance to the background of the observation target area; a comparison unit compares the distance information acquired by the distance information acquisition unit with the distance information acquired by the distance information acquisition unit when the object is not present in the observation target area, and an object detection unit detects the object based on the comparison result of the distance information by the comparison unit.

Images