Multi-camera device

Abstract

To provide a multi-camera apparatus capable of satisfying the required distance accuracy and imaging cycle that change according to the situation using a single multi-camera apparatus.SOLUTION: The multi-camera apparatus includes: an imaging timing switching unit (synchronous / asynchronous switching unit 32) that recognizes the outside world by using a parallax obtained by using plural images obtained from any set of two or more of plural cameras and performs sync / async switching among the plural cameras according to the desired distance accuracy and imaging cycle; and a parallax calculation unit 31 that calculates the parallax by using the plural images and the amount of deviation in imaging timing between the plural cameras.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a multi-camera device.

Background Art

[0002] As the background art of this technical field, there is Japanese Patent Application Laid-Open No. 2011-205388 (Patent Document 1). In this publication, as a problem, it is described that "it is possible to more easily detect a deviation in the shooting timing of a stereo camera." As a solution, "a presentation signal whose luminance periodically changes at a predetermined frequency is input to the generation unit 91 to the images obtained by shooting with the two left and right cameras constituting the stereo camera. The generation unit 91 generates left and right capture signals indicating the temporal change in the luminance of the presentation signal from the supplied left and right images. When the frequency of the presentation signal is higher than the Nyquist frequency of the frame rate of the camera, the filter processing unit 93 extracts left and right aliased signals of a specific frequency from the capture signal by filter processing. The matching processing unit 94 detects the phase shift between the left and right aliased signals, and the shift amount calculation unit 95 detects the deviation in the shooting timing of the camera from the phase shift of the aliased signal."

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] As a multi-camera device for detecting an object and measuring a distance, a stereo camera that measures a distance by triangulation is known. As indexes indicating the performance of the device, there are distance accuracy and processing cycle, but these performances are in a trade-off relationship.

[0005] Synchronous stereo cameras that synchronize multiple cameras and capture images at the same time are excellent in distance accuracy, while asynchronous stereo cameras (see Patent Document 1) that capture images at different times with multiple cameras are excellent in processing cycle. However, since the required distance accuracy and processing cycle vary depending on the situation, there has been a problem that when only one of them is adopted, the required performance cannot be satisfied depending on the scene.

[0006] An object of the present invention is to provide a multi-camera device that can satisfy a required distance accuracy and imaging cycle that change according to the situation with a single multi-camera device.

Means for Solving the Problems

[0007] In order to achieve the above object, the present invention is a multi-camera device that recognizes the external world using a parallax obtained using a plurality of images obtained from any combination of two or more of a plurality of cameras, and according to a desired distance accuracy and imaging cycle, an imaging timing switching unit that performs synchronous / asynchronous switching of the plurality of cameras, and a parallax calculation unit that calculates the parallax using the plurality of images and the amount of deviation in imaging timing between the plurality of cameras.

Effects of the Invention

[0008] According to the present invention, it is possible to provide a multi-camera device that can satisfy the required performance by appropriately switching the synchronization / asynchronization of the cameras even when the required distance accuracy and imaging cycle change.

[0009] Problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Modes for Carrying Out the Invention

[0011] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0012] [Example 1] Example 1 will be described with reference to FIGS. 1 to 7.

[0013] FIG. 1 shows the configuration of the multi-camera device. The multi-camera device 10 is composed of a multi-camera 12 constituted by a plurality of cameras 11, a memory 13, a CPU 14, an image processing unit 15, and an external output unit 16. The images captured by the multi-camera 12 (each camera 11) are stored in the memory 13. For the stored images, processing such as object detection is performed in the image processing unit 15, and the obtained results are output to the outside through the external output unit 16.

[0014] In this embodiment, an in-vehicle multi-camera device shown in FIG. 2 will be described. The multi-camera device 10 is mounted on a vehicle (hereinafter also referred to as the host vehicle, etc.) 1, captures images of the surroundings of the host vehicle, and detects a target object using a parallax obtained using a plurality of images obtained from any combination of two or more of the plurality of cameras 11, etc., to recognize the surroundings (the outside world) of the host vehicle, and the obtained information is output to a control command unit 2 as a controller. The control command unit 2 determines the operations (operating states) of a brake (not shown), an engine 3, a steer 4, etc., which are control means of the host vehicle, based on the input information (information on the detected target object, etc.), and controls the vehicle 1 (acceleration / deceleration control and steering control). Since the vehicle 1 travels in various scenes such as highways and shopping streets, in order to always satisfy the required performance (specifically, the required performance of distance accuracy and imaging cycle (processing cycle)), it is necessary to switch between synchronous and asynchronous according to this embodiment.

[0015] Next, the image processing unit 15 will be described with reference to FIG. 3. The image processing unit 15 of this embodiment includes a parallax calculation unit 31 and a synchronous / asynchronous switching unit 32.

[0016] First, in the synchronous / asynchronous switching unit 32, according to the required distance accuracy and imaging cycle, it is determined whether to capture images of the multi-camera 12 (each camera 11) synchronously or asynchronously (specifically, a synchronous flag or an asynchronous flag is output) (described later). This result is transmitted to the multi-camera 12, controls the imaging timing of the camera 11, and is used to correct the parallax of the asynchronous stereo camera in the parallax calculation unit 31. That is, the synchronous / asynchronous switching unit 32 has a function as an imaging timing switching unit that performs synchronous / asynchronous switching of the plurality of cameras 11 according to the desired distance accuracy and imaging cycle.

[0017] The parallax calculation unit 31 is composed of blocks as shown in FIG. 4. The parallax calculation unit 31 receives at least two images 1 and 2 from the multi-camera 12 as inputs in the parallax calculation processing unit 41 and executes parallax calculation processing. In the parallax calculation processing, the position where the same point as a specific point in Image 1 is imaged in Image 2 is searched for. The displacement of the coordinate position of the point obtained as a result of this search is called parallax and is converted into a distance. The calculated parallax (distance) has its correction amount calculated by a flag (synchronization flag or asynchronous flag) indicating the synchronous / asynchronous state set in the multi-camera 12 by the synchronization / asynchronous switching unit 32 (in the parallax correction amount calculation processing unit 43), and the correction is executed by the parallax correction processing unit 42 to obtain the final parallax.

[0018] The explanation of this parallax correction processing is shown in FIG. 5. In the synchronous stereo camera shown in the upper part of FIG. 5, parallax is calculated for the target object 51 based on the image obtained from the left camera 52 at a certain time t and the image obtained from the right camera 53 at the same time t. On the other hand, in the asynchronous stereo camera shown in the lower part of FIG. 5, parallax is calculated based on the image obtained from the left camera 52 at a certain time t and the image obtained from the right camera 54 at a different time t+1. When the sensor or the target object is moving, since the distances to the target object 51 when these two images are captured are different, it is not possible to calculate the distance from the simple parallax, and it is necessary to correct the calculated parallax by the correction 55 due to time change. That is, it is necessary to calculate (correct) the parallax using the two images and the amount of deviation in imaging timing (corresponding to the correction 55 due to time change) between the left and right cameras. Since errors are included in the correction, generally the distance accuracy of the asynchronous stereo camera is inferior to that of the synchronous stereo camera.

[0019] The difference in imaging periods at this time is shown in Fig. 6. When the imaging period of the camera is set as Δt, the imaging period of the synchronous stereo camera shown in the upper part of Fig. 6 is Δt. On the other hand, in the asynchronous stereo camera, as shown in the lower part of Fig. 6, when imaging is performed with a shift of exactly half of the imaging period, the imaging period is Δt / 2. Since a new image cannot be obtained for the time of the imaging period after an image is acquired once, the shorter the imaging period, the higher the responsiveness of the system can be improved. In addition, when an object whose shape changes moment by moment like a person is targeted, the shorter the imaging period, the less deformation there is between images. When a moving object is targeted, the shorter the imaging period, the less the amount of movement between images, so there are advantages such as easier tracking.

[0020] Next, the synchronous / asynchronous switching unit will be described with reference to Fig. 7. The synchronous / asynchronous switching unit 32 determines whether to perform imaging of each camera 11 of the multi-camera 12 synchronously or asynchronously according to the required distance accuracy and imaging period. As described above, when distance accuracy is required, a synchronous flag is output, and when it is necessary to shorten the imaging period, an asynchronous flag is output.

[0021] First, in S71, a determination is made based on the type of the target object for which the control means (operating state) of the host vehicle is to be determined. If the target object is a vehicle, it is assumed that there are no sudden changes in behavior or shape, so a synchronous stereo camera is desirable. On the other hand, if the target object is not a vehicle (e.g., a pedestrian), it is assumed that there are sudden changes in behavior or shape, so it is necessary to shorten the imaging period, and an asynchronous stereo camera is desirable. Therefore, in S71, if the target object is a vehicle, the process proceeds to S72, and if the target object is not a vehicle, the process proceeds to S76 to output an asynchronous flag.

[0022] Next, in S72, a determination is made based on the behavior of the host vehicle. Here, a determination is made as to whether the host vehicle is going straight or turning, based on vehicle information such as the steering angle. When the host vehicle is going straight, the surrounding environment is less likely to change rapidly. However, when turning, the direction of the sensor changes significantly, so the surrounding environment is more likely to change rapidly. Therefore, it is necessary to shorten the imaging cycle, and an asynchronous stereo camera is desirable. Thus, in S72, if the host vehicle is going straight, the process proceeds to S73. If the host vehicle is not going straight (i.e., turning), the process proceeds to S76 to output an asynchronous flag.

[0023] Next, in S73, a determination is made using the surrounding environment based on the object detection result. Here, a determination is made as to whether there is a blind spot area around the host vehicle (in other words, whether there is a blind spot where a sudden object may jump out). When there are few blind spots around the host vehicle and the view of the surroundings is good, the possibility of a sudden object appearing is low. However, when there are many objects and blind spots and the view of the surroundings is poor, the possibility of a sudden object jumping out from the blind spot is high, and an asynchronous stereo camera that can shorten the imaging cycle is desirable. Thus, in S73, if the view of the surroundings is good, the process proceeds to S74. If the view of the surroundings is poor, the process proceeds to S76 to output an asynchronous flag.

[0024] Next, in S74, a determination is made based on the driving route. Here, a determination is made as to whether the host vehicle is going straight or turning, based on map information and the like. The driving route of the host vehicle is estimated using map information and the result of white line detection by the sensor, and it is determined whether the host vehicle will go straight or turn in the future for the same reason as in S72. When the possibility of turning is high, switch to an asynchronous stereo camera. That is, in S74, if the estimated driving route is a straight road, the process proceeds to S75 to output a synchronous flag. If the estimated driving route is not a straight road (i.e., a turning road), the process proceeds to S76 to output an asynchronous flag.

[0025] In this embodiment, multiple conditions are treated with the same weight. In practice, weights may be assigned to each determination so that important determinations are prioritized, or each determination may be scored and the final decision may be made based on the total score.

[0026] (Operational Effects of Example 1) As a multi-camera device that detects an object and measures distance, a stereo camera that measures distance by triangulation is known. As indices indicating the performance of the device, there are distance accuracy and processing cycle, and these performances are in a trade-off relationship.

[0027] In order to improve the ranging accuracy, it is desirable to synchronize a plurality of cameras and capture images at the same time to remove the influence of movement and deformation of the target object. At this time, the processing cycle of the entire device becomes equal to the imaging cycle of a single camera.

[0028] On the other hand, a device that captures images at different times with a plurality of cameras is called an asynchronous stereo camera. In the above-mentioned Patent Document 1, it is described that "the matching processing unit 94 detects the phase shift of the left and right folded-back signals, and the shift amount calculation unit 95 detects the shift of the camera shooting timing from the phase shift of the folded-back signals." The shift amount of the measured distance due to imaging at different times is estimated, and the ranging result is corrected. The ranging accuracy at this time is inferior to that of a synchronous stereo camera, but since images are captured at different times with a plurality of cameras, the processing cycle of the entire device becomes shorter than the imaging cycle of a single camera.

[0029] Although the synchronous stereo camera is excellent in distance accuracy and the asynchronous stereo camera is excellent in processing cycle, since the required distance accuracy and processing cycle vary depending on the situation, there is a problem that when only one of them is adopted, the required performance cannot be satisfied depending on the scene.

[0030] The multi-camera device 10 of this example described above is a multi-camera device 10 that recognizes the external world using parallax obtained using a plurality of images obtained from any combination of two or more of a plurality of cameras, and according to a desired distance accuracy and imaging cycle, an imaging timing switching unit (synchronous / asynchronous switching unit 32) that performs synchronous / asynchronous switching of the plurality of cameras, and a parallax calculation unit 31 that calculates the parallax using the plurality of images and the shift amount of the imaging timing between the plurality of cameras.

[0031] Further, the desired distance accuracy and imaging cycle are determined according to at least one of the type of the target object for determining the control means of the host vehicle, the presence or absence of a blind spot area around the host vehicle (whether there is a blind spot where a sudden jump-out may occur), map information, or vehicle information such as the steering angle, depending on whether the host vehicle is going straight or turning.

[0032] According to this embodiment, by switching the synchronization / asynchronization of the multi-camera device 10 according to the required distance accuracy and processing cycle according to the situation, a multi-camera device 10 that satisfies the required performance in various scenes can be realized.

[0033] [Embodiment 2] In Embodiment 1, an embodiment of switching the synchronization / asynchronization of the camera was described. In Embodiment 2, an embodiment of smoothly changing the amount of deviation in imaging timing between cameras in an asynchronous stereo camera will be described in addition to the switching of the synchronization / asynchronization of the camera.

[0034] FIG. 8 shows an explanatory diagram of the parallax calculation unit in Embodiment 2. The parallax calculation unit 31 of this embodiment is configured to include a required performance calculation unit 81 and an imaging timing deviation amount calculation unit 82 in addition to the above-described Embodiment 1.

[0035] The required performance calculation unit 81 calculates the required performance that the multi-camera device 10 should satisfy (specifically, the required performance of distance accuracy and imaging cycle). An explanatory diagram of the required performance calculation unit 81 is shown in FIG. 9. 91, 92, 93, and 94 in FIG. 9 are the required performances calculated from the same viewpoints as S71, S72, S73, and S74 in FIG. 7, respectively. In this embodiment, instead of a simple switch, in order to smoothly change the amount of deviation in imaging timing, the required performance is calculated not by a simple presence / absence determination but by scoring so as to smoothly change according to the scene.

[0036] In the required performance calculation unit 91 based on the target type, the required performance is calculated according to the type of the target object. For example, considering a sensor capable of determining the types of vehicles and pedestrians, since pedestrians are more likely to have rapid movement changes compared to vehicles, it is desirable to increase the deviation amount of the imaging timing and raise the response speed. Additionally, since it is desirable that the response speed is higher when the time to collision TTC [s] is shorter, the target imaging period Δt_a [s] is set according to a table set for each type of the target object and TTC. Although vehicles and pedestrians are cited as examples of types, other types with different possibilities of movement changes, such as trucks having fewer rapid movement changes, may be added.

[0037] In the required performance calculation unit 92 based on the blind spot information, the required performance is calculated so that collision avoidance can be achieved when an object jumps out from the blind spot. When determining the target imaging period Δt_b [s], assuming the number of frames n [frame] required to detect the target object and perform a collision determination, the distance to the blind spot of the target object z [m], the vehicle speed v_car [m / s], and the collision-avoidable distance z_th [m], since it is only necessary to pass through the collision-avoidable distance between the jump-out and the collision-avoidance determination, the required imaging period may be determined to satisfy the following mathematical formula. At this time, for n, it may be changed according to the distance to the target object, such as increasing it when the target object is far away to suppress false braking, or decreasing it when the target object is in the vicinity to widen the allowable range of brake implementation because the risk is high.

[0038] [Equation 1] TIFF0007717524000001.tif735

[0039] In the required performance calculation unit 93 based on the host vehicle behavior, a method of linearly changing the required distance accuracy and the required imaging period according to the angular velocity ω [rad / s] of the host vehicle can be considered. Since a larger ω indicates a sharper turn and it is necessary to increase the response speed of the sensor, the target imaging period Δt_c [s] can be expressed, for example, by the following mathematical formula. As long as there is a relationship such that Δt_c becomes smaller when ω is larger, other mathematical formulas or a form of referring to a table storing values according to ω may also be acceptable.

[0040] [Number 2] TIFF0007717524000002.tif1118

[0041] In the required performance calculation unit 94 based on the surrounding environment, the required imaging period Δt_d [s] for the entire scene is changed. For example, based on GPS information, the surrounding map information is received. If it is a highway, since the view around is good, Δt_d is increased to reduce the responsiveness and improve the distance accuracy. If it is an environment such as a residential area where the view around is poor and various objects are expected to pop out, Δt_d is decreased to increase the responsiveness. Similar to the required performance calculation unit 92 based on the target type, not only highways and residential areas, but also if it is an environment with different required imaging periods, switching of three or more types may be performed. Also, instead of GPS, the determination of the above environment may be made by scene understanding using image recognition.

[0042] The required performance calculated from the above-mentioned multiple viewpoints is integrated by the required performance integrated determination unit 95. At the same time, when the required performance cannot be satisfied, priorities are assigned according to the urgency of the vehicle's control, and integration is performed so as to satisfy those with higher priorities. The flowchart of the integration process (that is, the calculation process of the final target imaging period) of the required performance integrated determination unit 95 is shown in FIG. 10. Specifically, since Δt_a and Δt_b are for setting the required imaging period and there are target objects assuming a collision, the time to collision TTC_a and TTC_b when a collision occurs can be calculated. TTC_a and TTC_b are compared with the threshold (S101, S102), and TTC_a and TTC_b are compared with each other (S103). When these are less than the threshold, they are prioritized over the required performance for the entire scene, and the target imaging period Δt_a [s] or Δt_b [s] is adopted (S104, S105). On the other hand, when these are greater than or equal to the threshold, since there is a time margin until the collision avoidance control is implemented even if an object pops out, the final required imaging period (= final target imaging period) is calculated by weighted averaging (weights w_a, w_b, w_c, w_d) with the required imaging period calculated by other means, and adopted (S106).

[0043] The required distance accuracy and the required imaging period are sent from the above to the imaging timing deviation amount calculation unit 82 (FIG. 8), and the imaging timing deviation amount calculation unit 82 calculates the imaging timing deviation amount that satisfies the received required performance. That is, the imaging timing deviation amount calculation unit 82 has a function as an imaging timing determination unit that determines the deviation amount of the imaging timing among the plurality of cameras 11 according to the desired distance accuracy and imaging period. The calculated (determined) imaging timing deviation amount is transmitted to the multi-camera 12, controls the imaging timing of the camera 11, and is sent to the parallax correction amount calculation processing unit 43, where it is used for parallax correction processing in the asynchronous stereo camera.

[0044] (Operation and Effect of Example 2) The multi-camera device 10 of the present embodiment described above includes, instead of the imaging timing switching unit, an imaging timing determination unit (imaging timing deviation amount calculation unit 82) that determines the deviation amount of the imaging timing among the plurality of cameras according to the desired distance accuracy and imaging period. Further, it further includes a required performance calculation unit 81 that calculates the required imaging period (final) for determining the deviation amount of the imaging timing among the plurality of cameras.

[0045] According to the present embodiment, by smoothly changing the imaging timing deviation amount of the multi-camera device 10 according to the required distance accuracy and processing period, a multi-camera device 10 with properties more suitable for various scenes can be realized compared to Example 1.

[0046] [Example 3] In Example 3, an embodiment when the cameras of the multi-camera device are used for different purposes will be described. In this example, the multi-camera device 10 is configured by a total of three cameras, one surveillance camera and two vehicles equipped with cameras, and an example of controlling the two vehicles will be described.

[0047] Fig. 11 shows the overview of the system. The base camera 111 that constitutes the surveillance camera installed in the environment captures images at regular intervals and transmits the imaging timing and images to the (in-vehicle) cameras 112 and 113 mounted on vehicles 7 and 8. Since the camera 112 mounted on the vehicle 7 moving straight ahead needs to detect and measure the distance to the preceding vehicle 9 far ahead of the front of its own vehicle, the required performance for distance measurement accuracy is high. Therefore, the camera 112 captures images at the same timing as the base camera 111 and processes them as a synchronous stereo camera to meet the required performance. On the other hand, since the field of view range of the camera 113 mounted on the vehicle 8 during turning changes greatly depending on the time due to the change in the direction of its own vehicle, the required response performance is high. Therefore, the camera 113 shifts the imaging timing from the base camera 111 and processes it as an asynchronous stereo camera to meet the required performance. That is, the multi-camera device 10 of the present embodiment is based on the imaging timing of at least one base camera 111 among the plurality of cameras, and the cameras 112 and 113 other than the base camera 111 are provided with a plurality of imaging timing determination units that determine the amount of deviation of the imaging timing according to the desired distance accuracy and imaging cycle for each camera.

[0048] Thus, even when images are aggregated from a plurality of independent cameras and image processing is used for different purposes respectively, it is possible to meet the required performance individually set for each camera.

[0049] As for the setting of the timing, in addition to communicating and synchronizing the imaging times between the cameras, the base camera 111 can be made to emit (project) an infrared light flash at the imaging timing, and the other cameras 112 and 113 can use the infrared light flash reflected in the captured images to adjust the imaging timing (as a reference). Means such as this can be considered.

[0050] (Function and effect of Embodiment 3) The multi-camera device 10 of the present embodiment described above is based on the imaging timing of at least one base camera 111 among the plurality of cameras, and cameras 112 and 113 other than the base camera 111 are provided with an imaging timing determination unit that determines the amount of deviation of the imaging timing according to the desired distance accuracy and imaging cycle for each camera.

[0051] Also, the base camera 111 transmits its own imaging timing and images to cameras 112 and 113 other than the base camera 111.

[0052] In addition, the base camera 111 projects an infrared light flash at its own imaging timing, and cameras 112 and 113 other than the base camera 111 are provided with an imaging timing determination unit that determines the amount of deviation of the imaging timing according to the desired distance accuracy and imaging cycle for each camera based on the timing of the captured infrared light flash.

[0053] According to the present embodiment, a multi-camera device 10 can be realized in which a plurality of cameras installed for different purposes such as a surveillance camera and an in-vehicle camera adjust their imaging timings along their respective purposes, so that each camera satisfies different required performances.

[0054] Note that the present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Also, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Further, for a part of the configuration of each embodiment, addition, deletion, or replacement with other configurations is possible.

[0055] Further, each of the above configurations may be configured such that some or all of them are configured by hardware or are realized by a program being executed by a processor. Also, the control lines and information lines show those considered necessary for explanation, and not necessarily all control lines and information lines on the product are shown. In reality, it may be considered that almost all configurations are interconnected.

Explanation of Signs

[0056] 1: Vehicle (own vehicle), 10: Multi-camera device, 11: Camera, 12: Multi-camera, 13: Memory, 14: CPU, 15: Image processing unit, 16: External output unit, 31: Parallax calculation unit, 32: Synchronous / asynchronous switching unit (imaging timing switching unit), 81: Required performance calculation unit (Example 2), 82: Imaging timing deviation amount calculation unit (imaging timing determination unit) (Example 2)

Claims

1. A multi-camera device that recognizes the external world using parallax obtained from a plurality of images obtained from any combination of two or more of a plurality of cameras, an imaging timing switching unit that switches between synchronization and asynchronization of the plurality of cameras according to a desired distance accuracy and an imaging period, a parallax calculation unit that calculates the parallax using the plurality of images and the amount of deviation in imaging timing between the plurality of cameras, and the desired distance accuracy and imaging period are determined by the type of target object that determines the control means of the host vehicle, wherein when the target object is a vehicle, the imaging timing switching unit synchronizes the plurality of cameras, and when the target object is not a vehicle, the plurality of cameras are asynchronous. A multi-camera device characterized by this.

2. A multi-camera device that recognizes the external world using parallax obtained from a plurality of images obtained from any combination of two or more of a plurality of cameras, an imaging timing switching unit that switches between synchronization and asynchronization of the plurality of cameras according to a desired distance accuracy and an imaging period, a parallax calculation unit that calculates the parallax using the plurality of images and the amount of deviation in imaging timing between the plurality of cameras, and the desired distance accuracy and imaging period are determined by the presence or absence of a blind spot area around the host vehicle, wherein when there are few blind spots around the host vehicle and the view around is good, the imaging timing switching unit synchronizes the plurality of cameras, and when there are blind spots around the host vehicle and the view around is poor, the plurality of cameras are asynchronous. A multi-camera device characterized by this.

3. A multi-camera device that recognizes the external world using parallax obtained from a plurality of images obtained from any combination of two or more of a plurality of cameras, an imaging timing switching unit that switches between synchronization and asynchronization of the plurality of cameras according to a desired distance accuracy and an imaging period, a parallax calculation unit that calculates the parallax using the plurality of images and the amount of deviation in imaging timing between the plurality of cameras, and the desired distance accuracy and imaging period are determined by whether the host vehicle is going straight or turning, determined from at least one of map information or vehicle information. The imaging timing switching unit synchronizes the plurality of cameras when the host vehicle is going straight, and asynchronizes the plurality of cameras when the host vehicle is turning. A multi-camera device characterized by this.

4. The multi-camera device according to any one of Claims 1 to 3, characterized in that it is provided with an imaging timing determination unit that determines the amount of deviation in imaging timing between the plurality of cameras according to a desired distance accuracy and imaging cycle, instead of the imaging timing switching unit.

5. The multi-camera device according to Claim 4, further comprising a required performance calculation unit that calculates a required imaging cycle for determining the amount of deviation in imaging timing between the plurality of cameras. A multi-camera device characterized by this.

6. The multi-camera device according to any one of Claims 1 to 3, based on the imaging timing of at least one base camera among the plurality of cameras, wherein cameras other than the base camera are provided with an imaging timing determination unit that determines the amount of deviation in imaging timing according to a desired distance accuracy and imaging cycle for each camera. A multi-camera device characterized by this.

7. The multi-camera device according to Claim 6, characterized in that the base camera transmits its own imaging timing and images to cameras other than the base camera.

8. The multi-camera device according to Claim 6, wherein the base camera projects an infrared light flash at its own imaging timing, and cameras other than the base camera are provided with an imaging timing determination unit that determines the amount of deviation in imaging timing according to a desired distance accuracy and imaging cycle for each camera, based on the timing of the imaged infrared light flash. A multi-camera device characterized by this.

Images