JP2016176697A - Distance measurement device, distance measurement program, method for distance measurement, and vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine whether information on a measured distance is valid.SOLUTION: The distance measurement device includes: an imaging unit capturing image data on a subject; a distance measuring unit 1051 acquiring information on a distance from the image data to the subject; a specific-scene recognition unit 1052 determining whether the image data corresponds to a specific scene from neutral network information with image data on the previously captured specific scene learned as teaching data; and a distance information validity determination unit 1053 determining whether the distance information is valid based on the result of determination by the specific-scene recognition unit 1052.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a distance measuring device, a distance measuring program, a distance measuring method, and a vehicle.

  The vehicle driving support system detects an object in front of the vehicle with an imaging device, measures the distance to the detected object, and performs notification to the driver and vehicle motion control. The driving support system includes, for example, a radar-type distance measuring device combined with a monocular imaging device, a stereo camera-type distance measuring device that calculates parallax from a stereo image acquired by a pair of imaging devices, and the like. Yes.

  Regarding vehicle driving support systems, a technology for recognizing a light source of a target vehicle from an image and recognizing a preceding / oncoming vehicle (see, for example, Patent Document 1), or an obstacle is recognized and the obstacle is identified from its movement pattern. A technique has been proposed (see, for example, Patent Document 2).

  However, the distance measuring device has a reduced accuracy of the measured distance information in a specific situation such as in a dark place or when the contrast of the object is low.

  An object of this invention is to provide the distance measuring apparatus which can judge whether the measured distance information is effective.

  The present invention provides an imaging unit that acquires image data of a subject, a distance measurement unit that acquires distance information from the image data to the subject, and neural network information in which image data of a specific scene captured in advance is learned as teacher data A specific scene recognition unit that determines whether or not the image data corresponds to a specific scene, and a distance information validity determination unit that determines whether or not the distance information is valid based on a result determined by the specific scene recognition unit. It is characterized by having.

  According to the present invention, it can be determined whether or not the measured distance information is valid.

It is a perspective view of the stereo camera which shows embodiment of the distance measuring apparatus which concerns on this invention. It is a front view which shows the example of the vehicle carrying the stereo camera of FIG. It is a block diagram of the stereo camera of FIG. It is a functional block diagram of a distance measurement program. It is a schematic diagram explaining the distance measurement principle of the stereo camera of FIG. It is a schematic diagram which shows the parallax of the stereo camera of FIG. It is a schematic diagram which shows the example of the image pick-up element of the stereo camera of FIG. It is a schematic diagram which shows the example of the parallax of the stereo camera of FIG. It is a schematic diagram which shows the concept of a neural network. It is a schematic diagram which shows the example of the image data of a specific scene. It is a schematic diagram which shows another example of the image data of a specific scene. It is a schematic diagram which shows another example of the image data of a specific scene. It is a schematic diagram which shows another example of the image data of a specific scene. It is a figure which shows the example of the image data which the stereo camera of FIG. 1 acquires. It is a flowchart of a distance measurement program. It is a flowchart which shows the specific scene recognition block of a distance measurement program. It is a flowchart which shows another example of a distance measurement program.

  Hereinafter, embodiments of a distance measuring device, a distance measuring program, and a distance measuring method according to the present invention will be described with reference to the drawings.

● Distance measuring device (1) ●
A stereo camera 100 that is an embodiment of a distance measuring device according to the present invention will be described.

  As shown in FIG. 1, in the stereo camera 100, a pair of imaging lenses 12a and 12b is exposed to the outside from a housing 110. As shown in FIG. 2, the stereo camera 100 is mounted in the vicinity of the inner rear view mirror 52 on the inner side (vehicle compartment side) of the windshield 51 of the vehicle 50. The mounting position of the stereo camera 100 is not limited to the above-described position, and may be any position that can detect a situation outside the vehicle ahead of the traveling direction of the vehicle 50. In the present embodiment, the width direction of the vehicle 50 (the left-right direction on the paper surface) is the vehicle width direction, and the direction passing through the paper surface is the vehicle traveling direction.

  The stereo camera 100 is installed on an object whose distance to the measurement object changes. The stereo camera 100 is not limited to the vehicle 50, and can also be installed on a moving object such as a ship or a railway, or a fixed object such as a building in the case of FA (Factory Automation). The stereo camera 100 obtains a distance from the image data including the imaged subject to the subject (measurement target) as described later (acquires distance information). The measurement target of the stereo camera 100 is another moving body, a person, an animal, or the like, and when the stereo camera 100 is installed on the moving body, it is a fixed object in the traveling direction of the moving body.

Stereo Camera Block As shown in the block diagram of FIG. 3, the stereo camera 100 includes cameras 100-1 and 100-2 and a processor 103. The stereo camera 100 includes an SDRAM (Synchronous Dynamic Random Access Memory) 104 and a ROM (Read Only Memory) 105. In addition, the stereo camera 100 includes a temperature sensor 106, a vehicle information acquisition unit 107, and a built-in memory 108.

  The cameras 100-1 and 100-2 acquire image data of a pair of subjects used for distance measurement as a pair of imaging units. The cameras 100-1 and 100-2 include imaging lenses 12a and 12b and CCDs (Charge Coupled Devices) 101-1 and 101-2. The cameras 100-1 and 100-2 include F / E (Front / End) -IC (Integrated Circuit) 102-1 and 102-2.

  The imaging lenses 12a and 12b allow light to pass through and form subject images on the light receiving surfaces of the CCDs 101-1 and 101-2. The imaging lenses 12a and 12b have the same shape and the same focal length, and their optical axes are parallel to the vehicle traveling direction. In the stereo camera 100, the interval between the optical axes of the imaging lenses 12a and 12b is referred to as a baseline length D.

  The CCDs 101-1 and 101-2 are examples of image sensors that convert subject images formed by the imaging lenses 12a and 12b into electrical image data and output the electrical image data to the F / E-ICs 102-1 and 102-2. is there. The image sensor is not limited to the CCDs 101-1 and 101-2, and various photoelectric conversion elements such as CMOS (Complementary Metal Oxide Semiconductor) can be used.

  The F / E-ICs 102-1 and 102-2 perform predetermined processing on the image data generated by the CCDs 101-1 and 101-2 and convert the image data into digital signals. The image data converted into the digital signal is transferred to the processor 103.

  The processor 103 performs image processing on the image data received from the cameras 100-1 and 100-2. The processor 103 includes a CCD signal processing block 103-1, a memory controller block 103-2, a CPU (Central Processing Unit) block 103-3, and an external I / F block 103-4. These blocks are connected to each other via a bus line.

  The CCD signal processing block 103-1 performs signal processing such as white balance adjustment and γ adjustment on the digital image data input from the F / E-ICs 102-1 and 102-2, as well as VD signals and HD signals. Is output.

  The memory controller block 103-2 performs processing for storing the image data processed by the CCD signal processing block 103-1 in the SDRAM 104.

  The CPU block 103-3 is an example of a computer that can execute a program. The CPU block 103-3 reads a distance measurement program stored in the ROM 105 and executes a distance measurement method described below. The distance measurement program is stored in the ROM 105 in advance. The distance measurement program may be downloaded via an external memory or a network and read into the ROM 105.

  The external I / F block 103-4 controls the operation of the temperature sensor 106 and the vehicle information acquisition unit 107 connected to the outside of the processor 103.

  The temperature sensor 106 acquires temperature information inside the stereo camera 100. The vehicle information acquisition unit 107 is connected to the vehicle control device of the vehicle 50 and acquires vehicle information such as vehicle speed and yaw rate. The built-in memory 108 stores temperature information acquired from the temperature sensor 106, vehicle information acquired from the vehicle information acquisition unit 107, and the like.

  The cameras 100-1 and 100-2 have a vertical drive signal (VD signal) / horizontal drive signal (HD signal) / clock signal sent from the CCD signal processing block 103-1 to the F / E-ICs 102-1 and 102-2. Controlled by (CLK signal). The timing at which the CCDs 101-1 and 101-2 acquire the image signal is synchronized with the VD signal. In the stereo camera 100, the pair of cameras 100-1 and 100-2 generate image data based on the interval of the VD signals, and the generated image data is stored in the SDRAM 104.

  By executing the distance measurement program stored in the ROM 105, the CPU block 103-3 realizes the functional blocks shown in FIG. The CPU block 103-3 realizes a distance measurement block 1051, a specific scene recognition block 1052, and a distance information validity determination block 1053. The function of each functional block and the flow of the distance measurement method executed by the distance measurement program will be described later.

Distance measurement process The principle of distance measurement executed by the distance measurement block 1051 will be described. An example in which the distance A between the subject 111 and an arbitrary point (for example, the main point) of the lenses 113a and 113b is measured by a pair of cameras 112a and 112b having the same configuration of the optical system shown in FIG. The distance measurement block 1051 corresponds to a distance measurement unit.

  The pair of cameras 112a and 112b includes lenses 113a and 113b and line sensors 115a and 115b, which are examples of imaging elements, and are both attached to the base 116. The cameras 112a and 112b are both arranged with the optical axis parallel to the vehicle traveling direction. Further, the optical axes of the cameras 112a and 112b are arranged apart from each other by the base line length D in the vehicle width direction.

  As shown in FIG. 5 and FIG. 6, the subject image 114a of the subject 111 obtained through the lens 113a and the subject image 114b of the subject 111 obtained through the lens 113b are lined with the same point on the subject 111 shifted by the parallax Δ. Reach the sensors 115a and 115b, respectively. The light reaching the line sensors 115a and 115b is received by a plurality of image sensors (pixels) of the line sensors 115a and 115b and converted into electric signals as described above.

  When the focal length of the lenses 113a and 113b is f, the following expression (1) is established when A >> f from the base line length D of the lenses 113a and 113b and the distance A between the subject 111 and the lenses 113a and 113b.

  A = Df / Δ (1)

  Since the base length D and the focal length f of the lenses 113a and 113b are known, the distance A to the subject 111 can be calculated by detecting the parallax Δ from the equation (1) and the principle of triangulation. Since the line sensors 115a and 115b are disposed near the center of the field of view of the lenses 113a and 113b having high resolution, focusing is performed near the center of the field of view.

  As shown in FIG. 7, the stereo camera 100 includes CCDs 101-1 and 101-2 in which areas A and a of a plurality of image sensors are configured in two directions (vehicle width direction and vehicle height direction) perpendicular to the vehicle traveling direction. The distance measuring principle explained above is two-dimensionally developed and realized.

  In the stereo camera 100, the distance measurement block 1051 calculates the parallax Δ of the CCDs 101-1 and 101-2 for each area. In order to obtain the parallax Δ, the distance measurement block 1051 obtains a position difference dA from the area A of the CCD 101-2 corresponding to the area a of the CCD 101-1 as shown in FIG. . In the distance measurement block 1051, the position where the difference dA is the smallest is set as the parallax of the area.

  The distance measurement block 1051 can acquire the parallax Δ in units of pixels by expanding the parallax acquisition range for each area to the entire pixels of the CCDs 101-1 and 101-2. The distance measurement block 1051 can obtain the distance A in units of pixels by substituting the parallax Δ into the above equation (1).

Specific scene recognition processing The specific scene recognition block 1052 determines whether or not the acquired image data corresponds to a specific scene based on neural network information obtained by learning image data of a specific scene captured in advance as teacher data. To do. The specific scene recognition block 1052 corresponds to a specific scene recognition unit.

  As shown in FIG. 9, the deep learning structure of the neural network used in the specific scene recognition block 1052 outputs image data as the input layer I and the recognition rate of the scene to be learned as the output layer O. The hidden layer H between the input layer I and the output layer O is characterized by having a large number of layers. Such a large number of hidden layers O is referred to as a deep learning structure. Each of the input layer I, the output layer O, and the hidden layer O has a node n that models a neuron.

  The image data of the specific scene to be learned in advance by the specific scene recognition block 1052 is a situation (scene) in which it is difficult to acquire the distance A from the subject 111 because the stereo camera 100 is difficult to acquire the parallax Δ, for example. Further, the image data of the specific scene is a scene in which the accuracy of the distance A from the subject 111 is low because the stereo camera 100 cannot acquire the parallax Δ with high accuracy, for example. In addition, it is difficult to acquire the distance A in all of the image data of the specific scene, and in addition, it is a scene that the vehicle 50 wants to avoid erroneously traveling when the distance A is erroneously recognized. The specific scene includes image data regarding a situation in which the accuracy of the distance information is reduced. The image data of the specific scene includes, as a subject, an object (obstruction) that hinders the acquisition of distance information as shown below, for example.

  FIG. 10 is an example of image data of a specific scene, and is image data for learning a scene having the fence 600 in the specific scene recognition block 1052. The image data including the fence 600 as a subject is difficult to acquire the parallax Δ due to the low contrast and the presence of objects with a high frequency and a narrow interval between the meshes of the fence 600.

  FIG. 11 is another example of image data of a specific scene, and is image data for learning a scene where the specific scene recognition block 1052 has the shutter 700. Image data including the shutter 700 as a subject has a low contrast, so it is difficult to obtain the parallax Δ.

  FIG. 12 is still another example of image data of a specific scene, and is image data for learning a scene having the fence 800 in the specific scene recognition block 1052.

  FIG. 13 shows still another example of image data of a specific scene, which is image data for causing a specific scene recognition block 1052 to learn a scene having a shutter (blocking door) 700 being opened and closed. Since image data including the shutter 700 in the middle of opening and closing as a subject may recognize a far subject such as the vehicle 900 from the gap at the time of opening and closing, it is difficult to obtain the parallax Δ.

  By learning a large number of image data of various specific scenes as described above, the specific scene recognition block 1052 can determine whether or not it is a specific scene from the image data acquired in the distance measurement block 1051. When the specific scene recognition block 1052 learns a large amount of image data of a specific scene including those having different image brightness, it is determined whether or not the image data acquired in a dark place is also a specific scene. be able to. In addition, when the specific scene recognition block 1052 learns a large amount of image data of specific scenes including obstacles such as the fence 600 that differ in the type and way of reflection, image data acquired in various scenes is also acquired. It can be determined whether or not the scene is a specific scene.

  The specific scene recognition block 1052 also learns the image data 200 of the scene obtained during normal driving as shown in FIG. 14 together with the image data of the specific scene as described above. By learning the image data of the scene obtained during such normal running, the specific scene recognition block 1052 improves the recognition rate between image data that is not a specific scene and image data of the specific scene.

Distance information validity judgment process The distance information validity judgment block 1053 is based on the result of the recognition process of whether or not the specific scene recognition block 1052 is the image data of the specific scene, and the distance information acquired from the image data is valid. It is determined whether or not. The distance information validity determination block 1053 corresponds to a distance information validity determination unit.

  The distance information validity determination block 1053 indicates that the distance information is not valid for the image data recognized by the specific scene recognition block 1052 as the image data of the specific scene in which the accuracy of the distance A from the subject 111 is low. to decide. When the distance information validity determination block 1053 determines that the distance information is not valid, the distance information validity determination block 1053 outputs information indicating the fact (scene NG information) to the vehicle control device of the vehicle 50.

  The distance information validity determination block 1053 determines that the distance information is valid for the image data recognized by the specific scene recognition block 1052 as not being image data of the specific scene. If the distance information validity determination block 1053 determines that the distance information is valid, the distance information validity determination block 1053 outputs information indicating that fact (scene OK information) to the vehicle control device of the vehicle 50.

  The vehicle control device that has received the scene OK information or the scene NG information from the distance information validity determination block 1053 operates the driving support system based on these information. In the vehicle control device, when the scene NG information is received, it is determined that the distance information used for the driving support system cannot be acquired, and the operation of the driving support system is stopped, or the distance information cannot be acquired. To notify the driver.

  The distance information validity determination block 1053 may determine whether or not the distance information is valid based on environment information that is information related to the environment of the stereo camera 100 in addition to the result determined by the specific scene recognition block 1052. The environmental information includes, for example, the elapsed time after the stereo camera 100 is activated, the temperature information inside the stereo camera 100 acquired from the temperature sensor 106, the luminance value of the image data, the vehicle speed of the vehicle 50 acquired from the vehicle information acquisition unit 107, and the like. It is done.

  When the elapsed time after the activation of the stereo camera 100 is short, the vehicle 50 on which the stereo camera 100 is mounted is also immediately after starting, and is often located in a garage or a parking lot. When the vehicle 50 is in such a place, the image data acquired by the stereo camera 100 is highly likely to be the image data of the specific scene described above, and therefore it is highly likely that the distance information is not effective. Further, even when the temperature inside the stereo camera 100 is high, the distance information may not be effective because the measurement accuracy of the distance information may be affected. Further, it is difficult to acquire distance information as described above even when the luminance value of the acquired image data is low.

  The distance information validity determination block 1053 can further improve the accuracy of the distance information validity determination process by determining whether the distance information is valid based on the environmental information as described above.

● Distance measurement method (1) ●
An example of the distance measurement method performed by each functional block described above, which is realized by the stereo camera 100 by causing the CPU block 103-3 to execute the distance measurement program, will be described with reference to the flowchart shown in FIG.

  The distance measurement block 1051 waits to receive a VD signal in order to take a timing of image signals transmitted from the cameras 100-1 and 100-2 (15-1).

  When the VD signal is received, the distance measurement block 1051 performs distance measurement processing for acquiring distance information using image data stored in the SDRAM 104 from the cameras 100-1 and 100-2 (15-2).

  The distance information validity determination block 1053 acquires vehicle information (elapsed time after startup, vehicle speed information) input from the vehicle information acquisition unit 107 via the external I / F block 103-4 (15-3). Based on the acquired vehicle information, the distance information validity determination block 1053 determines whether or not the vehicle is immediately after startup (the elapsed time after startup is within a predetermined time) (15-4).

  When the vehicle is not immediately after starting (15-4: NO), the distance information validity determination block 1053 determines whether the vehicle is stopped (the vehicle speed is equal to or lower than a predetermined speed) based on the acquired vehicle information. (15-9).

  When the vehicle is immediately after starting (15-4: YES), or when the vehicle is not immediately after starting (15-4: NO) and is stopped (15-9: YES), the specific scene recognition block 1052 is Then, the scene of the image data subjected to the distance measurement process is recognized (15-5).

  After recognizing the scene of the acquired image data, the specific scene recognition block 1052 determines whether the scene of the image data acquired based on the image data of the specific scene given as teacher data of the neural network corresponds to the specific scene. (15-6).

  The process for determining whether or not the scene corresponds to the specific scene is performed according to the flowchart shown in FIG. In the specific scene recognition block 1052, it is determined whether or not the acquired image data corresponds to the traveling scene shown in FIG. 14 (16-1).

  In the specific scene recognition block 1052, when it is determined that the acquired image data does not correspond to the traveling scene (16-1: NO), the acquired image data is converted into the specific scene image data shown in FIGS. It is determined whether or not this is true (16-2).

  When it is determined that the acquired image data is a traveling scene (16-1: YES), or when it is determined that it is not a specific scene (16-3: NO), the acquired image data can acquire distance information. It is determined that the scene is a scene (scene OK) (16-2).

  When it is determined that the scene is a specific scene (16-3: YES), it is determined that the acquired image data is a scene (scene NG) where it is difficult to acquire distance information (16-4).

  After the specific scene recognition process is completed for the acquired image data, the distance information validity determination block 1053 determines whether the distance information is valid based on the recognition result (15-7).

  If the result of the distance information validity determination process (15-6) is scene NG (15-7: YES), the result of the distance measurement process (15-2) based on the determined image data is not reliable. It is determined that the distance information is invalid (15-8). If the scene information is OK (15-7: NO) as a result of the distance information validity determination process (15-6), the result of the distance measurement process (15-2) based on the determined image data is reliable. It is determined that the distance information is valid (15-10).

  According to the distance measurement method described above, it is possible to determine whether the distance information is valid by determining whether the scene is a specific scene from the acquired image data. For this reason, in the driving support system of the vehicle 50 using the acquired distance information, driving support can be performed by appropriately recognizing a situation such as immediately after starting or during a stop.

  In addition, according to the distance measurement method described above, whether or not the distance information is valid is determined based on the vehicle information together with the specific scene recognition process when determining whether or not the distance information is valid. The reliability of the determination can be further improved.

  In the embodiment described above, the specific scene recognition block 1052 performs the specific scene recognition processing on the image data acquired by both the camera 100-1 and the camera 100-2, but the image data acquired by either one of them. You may process about. The specific scene recognition block 1052 performs a specific scene recognition process on the image data acquired by both the camera 100-1 and the camera 100-2 by taking the logical sum of the recognition results of both images, thereby recognizing accuracy. Can be improved.

● Distance measurement method (2) ●
Another example of the distance measurement method realized by causing the CPU block 103-3 to execute the distance measurement program and performed by each functional block described above will be described with reference to the flowchart shown in FIG.

  The distance measurement block 1051 waits to receive the VD signal in order to take the timing of the image signals transmitted from the cameras 100-1 and 100-2 (17-1).

  When the VD signal is received, the distance measurement block 1051 performs distance measurement processing for acquiring distance information using image data stored in the SDRAM 104 from the cameras 100-1 and 100-2 (17-2).

  The distance information validity determination block 1053 acquires temperature information input from the temperature sensor 106 via the external I / F block 103-4 (17-3). The distance information validity determination block 1053 determines the state of the internal temperature of the stereo camera 100 based on the acquired temperature information. For example, the distance information validity determination block 1053 determines whether or not the change in the acquired temperature information exceeds a predetermined range (20 ° C.) (17-4).

  When the temperature information change falls below the predetermined range (17-4: NO), the distance information validity determination block 1053 indicates that the result of the distance measurement process (17-2) based on the determined image data is reliable. It judges, distance information judges that it is effective, and ends processing (17-9).

    When the temperature information change exceeds a predetermined range (17-4: YES), the specific scene recognition block 1052 recognizes the scene of the image data subjected to the distance measurement process (17-5).

  After recognizing the scene of the acquired image data, the specific scene recognition block 1052 determines whether the scene of the image data acquired based on the image data of the specific scene given as teacher data of the neural network corresponds to the specific scene. (17-6).

  The process of determining whether or not the scene corresponds to the specific scene is performed according to the flowchart described above with reference to FIG.

  After the specific scene recognition process is completed for the acquired image data, the distance information validity determination block 1053 determines whether the distance information is valid based on the recognition result (17-7).

  If the result of the distance information validity determination process (17-6) is scene NG (17-7: YES), the result of the distance measurement process (17-2) based on the determined image data is not reliable. It is determined that the distance information is invalid (17-8). In the case of scene OK (17-7: NO) as a result of the distance information validity determination process (17-6), the result of the distance measurement process (17-2) based on the determined image data is reliable. It is determined that the distance information is valid (17-9).

  According to the distance measurement method described above, it is possible to determine whether the distance information is valid by determining whether the scene is a specific scene from the acquired image data. For this reason, in the driving support system of the vehicle 50 using the acquired distance information, driving support can be performed by appropriately recognizing a situation such as immediately after starting or during a stop.

  In a situation where the outside air temperature is high, such as under the hot summer sun, and immediately after startup, there may be a large difference between the temperature information just stored in the built-in memory 108 and the acquired temperature information. In this case, the distance measurement block 1051 may cause an error in the distance measurement result. Although the distance measurement result can be corrected in accordance with the temperature change, it may be difficult to correct the distance measurement result in the case of a precise device such as the stereo camera 100.

  According to the distance measurement method described above, it is possible to improve the effectiveness of the distance information by determining that the distance information is not effective for such a distance measurement result that is difficult to correct.

100 Stereo Camera 100-1 Camera 100-2 Camera 101-1 CCD
101-2 CCD
102-1 F / E-I / C
102-2 F / E-I / C
103 processor 103-1 CCD signal processing block 103-2 memory controller block 103-3 CPU block 103-4 external I / F block 104 SDRAM
105 ROM
106 Temperature Sensor 107 Vehicle Information Acquisition Unit 108 Built-in Memory 111 Subject 116 Base 110 Case 200 Image Data 50 Vehicle 51 Front Glass 52 Inner Rear View Mirror 600 Fence 700 Shutter 1051 Distance Measurement Block 1052 Specific Scene Recognition Block 1053 Distance Information Validity Judgment Block

JP2013-092975A JP 2004-145660 A

Claims (18)

An imaging unit for acquiring image data of a subject;
A distance measuring unit for acquiring distance information from the image data to the subject;
A specific scene recognition unit that determines whether the image data corresponds to the specific scene based on neural network information in which image data of the specific scene captured in advance is learned as teacher data;
A distance information validity determination unit that determines whether or not the distance information is valid based on a result determined by the specific scene recognition unit;
A distance measuring device. The specific scene includes image data in a situation where the accuracy of the distance information is reduced.
The distance measuring device according to claim 1. The specific scene includes an obstacle that is an object that hinders the acquisition of the distance information.
The distance measuring device according to claim 1 or 2. The distance information validity determination unit determines whether or not the distance information is valid based on environmental information that is information related to the environment of the distance measurement device in addition to the result of the determination by the scene recognition unit.
The distance measuring device according to claim 1. The environmental information is an elapsed time after the distance measuring device is activated.
The distance measuring device according to claim 4. The environmental information is temperature information of the distance measuring device.
The distance measuring device according to claim 4 or 5. The environmental information is a luminance value of the image data.
The distance measuring device according to claim 3. The distance measuring device is mounted on a vehicle,
The environmental information is a vehicle speed acquired from the vehicle.
The distance measuring device according to claim 4. The imaging unit is a stereo camera including a pair of imaging lenses and a pair of imaging elements that receive light of the subject that has passed through the imaging lens.
The distance measuring device according to claim 1. Obtaining image data of the subject;
Obtaining distance information from the image data to the subject;
Determining whether the image data corresponds to the specific scene based on neural network information in which image data of the specific scene captured in advance is learned as teacher data;
Determining whether the distance information is valid based on a result of determining whether or not the specific scene is applicable;
A distance measurement program that causes a computer to execute. The specific scene includes image data in a situation where the accuracy of the distance information is reduced.
The distance measurement program according to claim 10. The specific scene includes an obstacle that is an object that hinders the acquisition of the distance information.
The distance measurement program according to claim 10 or 11. In the step of determining whether or not the image data corresponds to the specific scene, in addition to the result of determining whether or not the image data corresponds to the specific scene, based on environmental information that is information about the environment of the distance measuring device Determining whether the distance information is valid;
The distance measurement program according to any one of claims 10 to 12. Obtaining image data of the subject;
Obtaining distance information from the image data to the subject;
Determining whether the image data corresponds to the specific scene by neural network information in which image data of the specific scene captured in advance is learned as teacher data;
Determining whether the distance information is valid based on a result of determining whether or not the specific scene is applicable;
Perform distance measurement method. The specific scene includes image data in a situation where the accuracy of the distance information is reduced.
The distance measuring method according to claim 14. The specific scene includes an obstacle that is an object that hinders the acquisition of the distance information.
The distance measurement program according to claim 14 or 15. In the step of determining whether or not the image data corresponds to the specific scene, in addition to the result of determining whether or not the image data corresponds to the specific scene, based on environmental information that is information about the environment of the distance measuring device Determining whether the distance information is valid;
The distance measuring method according to claim 14. It has a distance measuring device that measures the distance to the object in front of the traveling direction of the vehicle,
The said distance measuring device is a vehicle which is the distance measuring device in any one of Claims 1 thru | or 9.

Images