JP7700715B2 - Object recognition device and program - Google Patents

Description

The present invention relates to an object recognition device and a program.

There are known object recognition devices that recognize the distance from the vehicle to an object based on an image captured by an imaging device. For example, the object recognition device described in Patent Document 1 recognizes the actual distance from the vehicle to the object as the average of a distance calculated based on a horizontal edge that indicates the bottom edge of the object in the image and a distance calculated based on a vertical edge that indicates the left and right edges of the object in the image.

JP 2011-180020 A

Depending on the driving conditions of the vehicle, the accuracy of the distance calculated based on the horizontal and vertical edges may decrease due to a decrease in the accuracy of recognizing objects in the image. In this case, the accuracy of the actual distance may decrease. Here, the same problem occurs when calculating the actual distance using two different methods other than using the horizontal and vertical edges to calculate the actual distance.

The present invention has been made in consideration of the above circumstances, and its main objective is to provide an object recognition device and program that can properly determine whether the reliability of the actual distance from the vehicle to an object has decreased.

The present invention provides an object recognition device that acquires an image captured by an imaging device and calculates a distance from a vehicle to an object present in the image,
a first calculation unit that calculates a distance from the host vehicle to the object as a first distance based on a bottom position that is a position of a bottom end portion of the object in the image;
a second calculation unit that calculates a distance from the host vehicle to the object as a second distance based on a size of the object in the image;
an actual distance calculation unit that calculates an actual distance from the host vehicle to the object based on the first distance and the second distance;
a reliability determination unit that determines a reliability of the actual distance based on a distance difference between the first distance and the second distance;
Equipped with.

In a configuration in which the actual distance from the vehicle to an object is calculated using a first distance calculated based on the bottom edge position of the object in the image and a second distance calculated based on the size of the object in the image, the accuracy of the actual distance can be improved compared to when the actual distance is calculated using only one of these distances. However, if the accuracy of at least one of the first distance and the second distance is reduced, it is conceivable that the accuracy of the actual distance will be reduced.

Therefore, in the present invention, the reliability of each actual distance is determined based on the distance difference between the first distance and the second distance. In this case, if the distance difference between the first distance and the second distance is large, the accuracy of one of these distances may be reduced, and by understanding this distance difference, the reliability of the actual distance can be appropriately determined.

FIG. 1 is a configuration diagram of a driving assistance system. FIG. 2 is a diagram showing an image captured of a scene ahead in the traveling direction of the host vehicle; 11 is a flowchart showing the procedure of an image recognition process. 10 is a flowchart showing the processing procedure of image recognition processing according to a second embodiment.

First Embodiment
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment in which an object recognition device according to the present invention is applied to a vehicle driving assistance system will be described with reference to the drawings.

As shown in FIG. 1, the driving assistance system 10 according to this embodiment includes a radar device 21, an imaging device 22, a vehicle speed sensor 23, a steering angle sensor 24, an angle sensor 25, a receiving device 26, an ECU 30, and a controlled device 40. In this embodiment, the ECU 30 corresponds to the driving assistance device.

The radar device 21 is, for example, a known millimeter wave radar that transmits high-frequency signals in the millimeter wave band. The radar device 21 is, for example, attached to the front of the vehicle with its optical axis facing forward, and detects the position of an object within a detection range that is within a predetermined detection angle. Specifically, the radar device 21 transmits search waves at a predetermined cycle and receives the reflected waves by multiple antennas. The radar device 21 calculates the distance to the object based on the transmission time of the search wave and the reception time of the reflected wave. The radar device 21 also calculates the relative speed based on the frequency of the reflected wave reflected by the object, which has changed due to the Doppler effect. In addition, the radar device 21 calculates the direction of the object based on the phase difference of the reflected waves received by the multiple antennas. The radar device 21 outputs detection data such as the distance to the object, the relative speed, and the direction of the object to the ECU 30.

The imaging device 22 is a monocular camera installed in only one vehicle. The imaging device 22 is, for example, a CCD camera, a CMOS image sensor, a near-infrared camera, etc. The imaging device 22 is attached, for example, at a predetermined height in the center of the vehicle width direction, and captures an image of an area extending in a predetermined angular range forward of the vehicle from a bird's-eye view. The imaging device 22 sequentially outputs the captured images to the ECU 30. Note that the imaging device 22 may be a stereo camera installed in multiple locations in the vehicle.

The vehicle speed sensor 23 is a sensor that detects the traveling speed of the host vehicle, and outputs a traveling speed signal corresponding to the traveling speed of the host vehicle to the ECU 30. The steering angle sensor 24 is a sensor that detects the steering angle of the steering wheel, and outputs a steering angle signal corresponding to the change in steering angle to the ECU 30. The angle sensor 25 includes a gyro sensor and an acceleration sensor attached to the body of the host vehicle, and detects the pitch angle of the host vehicle from the angular velocity detected by the gyro sensor and the acceleration detected by the acceleration sensor. The angle sensor 25 outputs a detected pitch angle signal of the host vehicle to the ECU 30.

The receiving device 26 is a receiving device for a positioning signal from a satellite positioning system, for example a GPS receiving device. The receiving device 26 receives road information about the road on which the vehicle is traveling and the surrounding roads, and outputs the received road information to the ECU 30. The road information includes at least road gradient information.

The functions provided by the ECU 30 can be provided by software recorded in a physical memory device and a computer that executes the software, by software alone, by hardware alone, or by a combination of these. For example, when the ECU 30 is provided by an electronic circuit that is hardware, it can be provided by a digital circuit including a large number of logic circuits, or an analog circuit. For example, the ECU 30 executes a program stored in a non-transitory tangible storage medium that serves as a storage unit that the ECU 30 has. The program includes programs for each arithmetic process, which will be described later. When the program is executed, a method corresponding to the program is performed. The storage unit is, for example, a non-volatile memory. The programs stored in the storage unit can be updated, for example, via a network such as the Internet.

Based on the input information, the ECU 30 performs object recognition processing to recognize objects present around the vehicle. In the object recognition processing, at least the distance from the vehicle to the object is recognized. Objects include four-wheeled vehicles, motorbikes, bicycles, people (pedestrians), etc. The object recognition processing will be described later.

The ECU 30 performs driving assistance control based on the results of the object recognition process. In this embodiment, the ECU 30 performs a collision prevention operation to prevent a collision with an object as driving assistance control.

The ECU 30 determines that the condition for performing collision prevention operation is met when the collision prediction time TTC (Time to Collision) for an object in front of the vehicle falls below a predetermined threshold. The collision prediction time TTC is an evaluation value that indicates how many seconds it will take for the vehicle to collide with an object if traveling at the current vehicle speed. The smaller the collision prediction time TTC, the higher the risk of collision, and the larger the collision prediction time TTC, the lower the risk of collision. The collision prediction time TTC is, for example, a value obtained by dividing the relative distance of the object to the vehicle by the relative speed of the object to the vehicle.

The ECU 30 operates the controlled devices 40, including the warning device 41, accelerator device 42, brake device 43, and steering device 44, as a collision prevention operation. The warning device 41 is a device that issues a warning to the driver of the vehicle, and is, for example, a device that issues an auditory notification such as a speaker or buzzer installed in the vehicle cabin, or a device that issues a visual notification such as a display or warning light. The accelerator device 42 is an engine or motor as a vehicle power source, and is a device that applies driving force to the vehicle. The brake device 43 is a device that brakes the vehicle by operating a hydraulic actuator or the like. The steering device 44 is a device that steers the vehicle by operating a steering motor or the like. In this embodiment, the ECU 30 brakes the vehicle by the brake device 43 as a collision prevention operation.

The object recognition process performed by the ECU 30 will be described below. As part of the object recognition process, the ECU 30 performs image recognition processing to acquire an image captured by the imaging device 22 and calculate the distance from the vehicle to objects present in the image. In this embodiment, the ECU 30 includes a type recognition unit 31, a setting unit 32, a first calculation unit 33, a second calculation unit 34, and an actual distance calculation unit 35. The image recognition process will be described below with reference to FIG. 2. FIG. 2 is an example of an image G captured in front of the vehicle in the direction of travel. A person 50 is captured in the image G as an object.

The type recognition unit 31 recognizes the type of object in the image. In this embodiment, the type recognition unit 31 extracts a rectangular area surrounding an object in the image, and calculates a type score indicating the likelihood of the object type for the image in the rectangular area. The type recognition unit 31 predetermines multiple type recognition targets (e.g., buses, trucks, ordinary cars, motorcycles, bicycles, adults, children, etc.), and calculates a type score for each image in the rectangular area using learning data that defines the recognition targets. The form of the learning data is arbitrary, but it is preferable to use learning data generated by deep learning, for example. The type score may be expressed as the probability that each image in the rectangular area corresponds to a certain type. In FIG. 2, the type recognition unit 31 extracts a rectangular area R surrounding a person 50 in the image G, and calculates the type score of the person 50 as "child: 85%", "adult: 10%", and "other objects: 5%". In this case, the type recognition unit 31 recognizes the type of person 50 as a child because the type score indicates the highest probability of being a child. Note that instead of recognizing the type of object based on the type score, the type recognition unit 31 may recognize the type of object by performing template matching processing on the image within the rectangular area.

The type recognition unit 31 may, for example, calculate a motion vector for each pixel in the image, and extract a rectangular area surrounding the object by considering a group of pixels with motion vectors of the same direction and magnitude as pixels that constitute the same object. A motion vector is a vector that indicates the direction and magnitude of change over time in each pixel that constitutes an object in the image.

The setting unit 32 sets a size reference value that is a reference value for the actual size of an object in an image. The size reference value is determined in advance for each type of object. In this embodiment, the setting unit 32 sets the height width value of an object corresponding to the type of object as the size reference value. The setting unit 32 associates the height width value with 1.7 m for adults, 1.2 m for children, 2.0 m for regular cars, and 3.5 m for large vehicles such as buses. In FIG. 2, the setting unit 32 sets the size reference value of a person 50 recognized as a child to 1.2 m. Note that the setting unit 32 may set the width value of an object corresponding to the type of object instead of the height width value of the object as the size reference value.

The first calculation unit 33 calculates the distance from the vehicle to the object as the first distance based on the bottom position, which is the position of the bottom end of the object in the image. In this embodiment, as shown in FIG. 2, the first calculation unit 33 calculates the first distance based on the number of pixels L corresponding to the distance between the bottom end position of the person 50 and the point at infinity FOE of the imaging device 22. When the number of pixels L is small, the first calculation unit 33 calculates the first distance to be longer than when the number of pixels L is large. Note that the first calculation unit 33 may calculate the first distance based on the number of pixels corresponding to the distance between the bottom end position of the person 50 and the bottom end position of the image G instead of the number of pixels L. In this case, when the number of pixels corresponding to the distance between the bottom end position of the person 50 and the bottom end position of the image G is large, the first calculation unit 33 may calculate the first distance to be longer than when the number of pixels is small.

The second calculation unit 34 calculates the distance from the vehicle to the object as the second distance based on the size of the object in the image and the size reference value of the object set by the setting unit 32. If the size reference value of the object is the same, the second calculation unit 34 calculates the second distance to be shorter when the size of the object in the image is large than when the size of the object in the image is small. Also, if the size of the object in the image is the same, the second calculation unit 34 calculates the second distance to be shorter when the size reference value of the object is small than when the size reference value of the object is large. In this embodiment, as shown in FIG. 2, the second calculation unit 34 uses the number of pixels H corresponding to the distance from the top end position to the bottom end position of the rectangular region R as the size of the object in the image. Note that the second calculation unit 34 may use the number of pixels corresponding to the distance from the left end position to the right end position of the rectangular region R instead of the number of pixels H as the size of the object in the image.

The actual distance calculation unit 35 calculates the actual distance from the vehicle to the object based on the first distance and the second distance. The actual distance is used, for example, to calculate the collision prediction time TTC in the collision prevention operation. In this embodiment, the actual distance calculation unit 35 calculates the average value of the first distance and the second distance as the actual distance.

In this configuration, the first distance calculated based on the bottom edge position of the object in the image and the second distance calculated based on the size of the object in the image are used to calculate the actual distance from the vehicle to the object, and the accuracy of the actual distance can be improved compared to when the actual distance is calculated using only one of these distances. However, depending on the driving conditions of the vehicle, the accuracy of at least one of the first distance and the second distance may decrease, resulting in a decrease in the accuracy of the actual distance.

Therefore, in this embodiment, the ECU 30 includes a reliability determination unit 36 to determine the reliability of the actual distance. The reliability determination unit 36 determines the reliability of the actual distance based on the distance difference between the first distance and the second distance. Here, the distance difference between the first distance and the second distance is the absolute value of the difference between the first distance and the second distance.

Figure 3 shows the procedure for the image recognition process performed by the ECU 30. This process is repeated at a predetermined interval.

In step S10, the image captured by the imaging device 22 and driving condition information are acquired. The driving condition information is information related to the pitch angle of the vehicle and information related to the gradient of the road on which the vehicle is traveling. The information related to the gradient of the road on which the vehicle is traveling is, for example, road information acquired from the receiving device 26. The information related to the pitch angle of the vehicle is, for example, the pitch angle signal of the angle sensor 25. In this embodiment, step S10 corresponds to the "acquisition unit."

In step S11, the first distance D1 is calculated. In this embodiment, the first distance D1 is calculated based on the number of pixels corresponding to the distance between the bottom edge position of the object in the image and the point at infinity FOE of the imaging device 22.

In step S12, a type recognition process is performed. In this embodiment, the type recognition process is a process in which a rectangular area surrounding an object in an image is extracted, and the type of object is recognized based on the type score of the image in the rectangular area. In step S13, it is determined whether or not the type of object in the image has been recognized by the type recognition process. If a negative determination is made in step S13, the process proceeds to step S22. On the other hand, if a positive determination is made in step S13, the process proceeds to step S14.

In step S14, a size reference value for an object in an image is set. In this embodiment, the size reference value is set to the height and width value of the object corresponding to the type of object recognized by the type recognition process.

In step S15, the second distance D2 is calculated. In this embodiment, the second distance D2 is calculated based on the number of pixels corresponding to the distance from the top to bottom of a rectangular area surrounding an object in the image and a reference value for the size of the object.

In step S16, it is determined whether the object in the image is a person of short stature. In this embodiment, if the object in the image is a person and its size reference value is below the short stature threshold, the object in the image is determined to be a person of short stature. For example, if 1.7 m is set as the size reference value when the object type is an adult and 1.2 m is set as the size reference value when the object type is a child, the short stature threshold may be set to a value between 1.2 m and 1.7 m.

If the determination in step S16 is negative, the process proceeds to step S20. On the other hand, if the determination in step S16 is positive, the process proceeds to step S17.

In step S17, it is determined whether the gradient of the road ahead of the vehicle on which the vehicle is traveling is either an upward gradient or a downward gradient. The gradient of the road ahead of the vehicle may be determined based on at least information on the gradient of the road on which the vehicle is traveling, which is included in the driving situation information. If a positive determination is made in step S17, the process proceeds to step S21. On the other hand, if a negative determination is made in step S17, the process proceeds to step S18.

In step S18, it is determined whether pitching of the host vehicle has occurred. Pitching of the host vehicle includes pitching that occurs due to acceleration or deceleration of the host vehicle, and pitching that occurs momentarily due to the road environment on which the host vehicle is traveling. The determination of whether pitching of the host vehicle has occurred may be made based on at least information related to the pitch angle of the host vehicle among the driving condition information. If a positive determination is made in step S18, the process proceeds to step S21. If a negative determination is made in step S18, the process proceeds to step S19.

In step S19, the reliability of the actual distance is determined. In this embodiment, the reliability of the actual distance is determined by determining whether the distance difference ΔD, which is the absolute value of the difference between the first distance D1 and the second distance D2, is equal to or less than the distance difference threshold Dth. Here, the distance difference threshold Dth is a fixed value. If the determination is positive in step S19, the process proceeds to step S20, where it is determined that the reliability of the actual distance is high. On the other hand, if the determination is negative in step S19, the process proceeds to step S21, where it is determined that the reliability of the actual distance is low.

In this embodiment, if it is determined in step S20 that the reliability of the actual distance is high, the process proceeds to step S22, where the actual distance, which is the average value of the first distance D1 and the second distance D2, is adopted as the processing result of the image recognition process. In this case, for example, in the collision prevention operation, the actual distance is used to calculate the collision prediction time TTC. On the other hand, if it is determined in step S21 that the reliability of the actual distance is low, the process proceeds to step S23, where the actual distance is not adopted as the processing result of the image recognition process. In this case, for example, in the collision prevention operation, the distance and relative speed to the object detected by the radar device 21 are used to calculate the collision prediction time TTC.

The present embodiment described above provides the following advantages:

The reliability of the actual distance is determined based on the distance difference ΔD between the first distance D1 and the second distance D2. In this case, if the distance difference ΔD between the first distance D1 and the second distance D2 is large, the accuracy of either of the distances D1 and D2 may be reduced, and by understanding the distance difference ΔD, the reliability of the actual distance can be properly determined.

It is possible that the reference value for the size of an object is set smaller than the actual size of the object. For example, it is possible that the type of person as an object is erroneously determined to be a child when in fact the person is an adult. In this case, there is a concern that the reference value for the size of an object is set smaller than the actual size of the object, and the second distance D2 is calculated to be shorter than the actual distance.

Taking this into consideration, the system is configured to determine whether or not to perform a reliability determination of the actual distance based on the size reference value of the object. In more detail, if the object in the image is a person and its size reference value is below the short height threshold, the object in the image is determined to be a short person. If the object in the image is determined to be a short person, the reliability of the actual distance is determined based on the gradient of the road on which the vehicle is traveling, the occurrence of pitching of the vehicle, and the distance difference ΔD. This makes it possible to determine the reliability of the actual distance in a situation where there is concern that an incorrectly set size reference value for the object will cause the second distance D2 to be calculated as shorter than the actual distance, resulting in unnecessary collision prevention operations being performed.

When the vehicle is pitching or the road ahead of the vehicle has a slope, there is a high possibility that an error will occur in the first distance D1 calculated based on the bottom end position of the object in the image. In this case, there is a concern that the reliability of the actual distance cannot be appropriately determined.

Therefore, in this embodiment, it is determined whether or not to perform a reliability determination of the actual distance based on the driving situation information. This makes it possible to prevent a reliability determination of the actual distance from being performed in a situation in which the reliability determination of the actual distance cannot be performed appropriately.

Second Embodiment
The second embodiment will be described below, focusing on the differences from the first embodiment. In the first embodiment, whether or not to adopt the actual distance, which is the average value of the first distance D1 and the second distance D2, as the processing result of the image recognition process was determined based on the determination result by the reliability determination unit 36. In contrast, in this embodiment, the value of the actual distance adopted as the processing result of the image recognition process is changed based on the determination result by the reliability determination unit 36.

Figure 4 shows the procedure for the image recognition process performed by the ECU 30. This process is repeated at a predetermined interval. Note that in Figure 4, the same step numbers are used for processes that overlap with Figure 3.

If the result of step S13 is negative, it is determined that the accuracy of the second distance D2 is reduced, and therefore it is determined that the processing of step S19 cannot be performed appropriately, and the process proceeds to step S30. In step S30, it is determined that the reliability of the actual distance is low. In this case, in step S31, the actual distance calculated based on the first distance D1 and the second distance D2 is not adopted as the processing result of the image recognition process.

If the determination in step S17 is positive, it is determined that the accuracy of the first distance D1 has decreased, and the process proceeds to step S30. If the determination in step S18 is positive, it is also determined that the accuracy of the first distance D1 has decreased, and the process proceeds to step S30.

If the answer is positive in step S13 and negative in steps S17 to S19, the process proceeds to step S32. In step S32, it is determined that the reliability of the second distance D2 is low. In this case, in step S33, of the first distance D1 and the second distance D2, the first distance D1 is adopted as the actual distance, which is the processing result of the image recognition process. In this embodiment, steps S13 to S15, S17 to S20, S22, and S30 to S33 correspond to the "change unit."

In this embodiment, in the process of step S19, a different value is used as the distance difference threshold Dth depending on the type of object in the image. In detail, the value of the distance difference threshold Dth used to determine the reliability of the actual distance when the type recognized by the type recognition process is a human type is set to be smaller than the value of the distance difference threshold Dth used to determine the reliability of the actual distance when the type recognized by the type recognition process is a vehicle type. Here, the human type is a type recognized as a person, such as an adult and a child. Also, the vehicle type is a type recognized as a vehicle, such as a passenger car, a truck, a bus, etc.

The present embodiment described above provides the following advantages:

According to this embodiment, the cause of the decrease in the reliability of the actual distance is determined based on the distance difference ΔD, the type of object, and the driving situation information. If it is determined that the cause of the decrease in the reliability of the actual distance is a decrease in the reliability of the second distance D2, the first distance D1 is adopted as the actual distance, which is the processing result of the image recognition process. As a result, even in a situation where the reliability of the actual distance decreases, the actual distance can be calculated based on the first distance D1 and the second distance D2 while suppressing a decrease in the accuracy of the actual distance.

In the type recognition process in step S12, it is considered that at least a plurality of different sizes of people types and a plurality of different sizes of vehicle types are recognized by distinguishing between them. In this case, it is considered that the race type of the person as the target object may be erroneously recognized, and the vehicle type of the vehicle as the target object may be erroneously recognized. For example, it is considered that the type of the person as the target object may be erroneously recognized as a child when it is actually an adult, and the type of the vehicle as the target object may be erroneously recognized as a passenger car when it is actually a bus. When the race type or vehicle type is erroneously recognized, the size reference value of the target object is set incorrectly, the accuracy of the second distance D2 decreases, and the distance difference ΔD increases.

However, in this case, the difference in size when distinguishing by race is smaller than the difference in size when distinguishing by vehicle type. Therefore, if a fixed value is used as the distance difference threshold Dth, there is a concern that when the type of target object is included in the race type, the reliability determination may be erroneously determined to be high reliability.

Therefore, in this embodiment, the value of the distance difference threshold Dth used to determine the reliability of the actual distance when the type recognized by the type recognition process is a race type is made smaller than the value of the distance difference threshold Dth used to determine the reliability of the actual distance when the type recognized by the type recognition process is a vehicle type. This makes it possible to appropriately determine the decrease in reliability of the actual distance caused by erroneous recognition of race when the type of the target object is a race type.

<Other embodiments>
The above embodiment may be modified, for example, as follows.

- In the above embodiment, the setting unit 32 sets a size reference value for an object in an image based on the type of object recognized by the type recognition unit 31, but this may be changed.

The setting unit 32 may estimate the actual size of the object based on the first distance calculated by the first calculation unit 33 and the size of the object in the image, and may set the object size reference value based on the estimated value. In this case, if the size of the object in the image is the same, when the first distance based on the bottom end position of the object in the image is long, the setting unit 32 may estimate the estimated value of the object size to be larger than when the first distance is short. Also, if the first distance is the same, when the size of the object in the image is large, the setting unit 32 may estimate the estimated value of the object size to be larger than when the size of the object in the image is small. The setting unit 32 may select, as the object size reference value, a category value closest to the estimated value of the object size, from among category values that are values assumed to be the size of the object.

In this embodiment, in the image recognition process shown in FIG. 3, steps S12 and S13 do not need to be performed. In other words, in the image recognition process, the actual distance can be calculated and the reliability of the actual distance can be determined without performing type recognition processing.

- In the above embodiment, the collision prevention operation is configured to brake the vehicle using the brake device 43, but this may be changed. For example, the collision prevention operation may be configured to either issue a warning using the warning device 41 or steer the vehicle using the steering device 44.

- In the above embodiment, the driving assistance control is configured to perform a collision prevention operation, but this may be changed. For example, the driving assistance control may be configured to perform ACC (Adaptive Cruise Control). ACC control is a control in which another vehicle traveling ahead of the host vehicle in the direction of travel is selected as a preceding vehicle, and the host vehicle is controlled using the accelerator device 42, the brake device 43, and the steering device 44 to follow the preceding vehicle. In addition, for example, the driving assistance control may be to display on a display or the like the distance from the host vehicle to an object recognized by object recognition processing.

The vehicle control device and the method thereof described in the present disclosure may be realized by a dedicated computer provided by configuring a processor and memory programmed to execute one or more functions embodied in a computer program. Alternatively, the vehicle control device and the method thereof described in the present disclosure may be realized by a dedicated computer provided by configuring a processor with one or more dedicated hardware logic circuits. Alternatively, the vehicle control device and the method thereof described in the present disclosure may be realized by one or more dedicated computers configured by combining a processor and memory programmed to execute one or more functions with a processor configured with one or more hardware logic circuits. In addition, the computer program may be stored in a computer-readable non-transient tangible recording medium as instructions executed by the computer.

22...imaging device, 30...ECU, 33...first calculation unit, 34...second calculation unit, 35...actual distance calculation unit, 36...reliability determination unit, 50...person.

Claims (6)

An object recognition device (30) that acquires an image captured by an imaging device (22) and calculates a distance from a vehicle to an object (50) present in the image,
a first calculation unit (33) that calculates a distance from the host vehicle to the object as a first distance based on a bottom end position that is a position of a bottom end portion of the object in the image;
a second calculation unit (34) that calculates a distance from the host vehicle to the object as a second distance based on a size of the object in the image;
an actual distance calculation unit (35) that calculates an actual distance from the host vehicle to the object based on the first distance and the second distance;
a reliability determination unit (36) that determines the reliability of the actual distance based on a distance difference between the first distance and the second distance;
An object recognition device comprising: A setting unit (32) for setting a size reference value which is a reference value of the actual size of the object,
the second calculation unit calculates the second distance based on a size of the object in the image and the size reference value of the object set by the setting unit;
The object recognition device according to claim 1 , wherein the reliability determining unit determines whether or not to determine the reliability of the actual distance to the object based on the size reference value of the object. an acquisition unit that acquires driving status information of the host vehicle, the driving status information being at least one of information regarding a pitch angle of the host vehicle and information regarding a gradient of a road on which the host vehicle is traveling;
The object recognition device according to claim 1 , wherein the reliability determination unit determines whether or not to perform a reliability determination of the actual distance based on the traveling condition information. The system includes a type recognition unit (31) that includes people and vehicles as target objects and recognizes, as object type recognition, at least a plurality of human types having different sizes and a plurality of vehicle types having different sizes by distinguishing between them,
The second calculation unit calculates the second distance based on a size of the object in the image and the type recognized by the type recognition unit;
The reliability determination unit is
If the distance difference is greater than a threshold, it is determined that the reliability of the actual distance is low, and if the distance difference is smaller than the threshold, it is determined that the reliability of the actual distance is high.
2. The object recognition device according to claim 1, wherein the threshold value used to determine the reliability of the actual distance calculated by the actual distance calculation unit when the type recognized by the type recognition unit is a human is set smaller than the threshold value used to determine the reliability of the actual distance calculated by the actual distance calculation unit when the type recognized by the type recognition unit is a vehicle. A type recognition unit (31) that recognizes the type of the object;
an acquisition unit that acquires driving status information of the host vehicle, the driving status information being at least one of a pitch angle of the host vehicle and a gradient of a road on which the host vehicle is traveling;
The reliability determination unit determines a cause of a decrease in reliability of the actual distance based on the distance difference, the type of the object, and the traveling situation information,
The object recognition device according to claim 1 , further comprising a change unit that changes the value of the actual distance calculated by the actual distance calculation unit in accordance with the cause of the decrease. A program executed by a control device (30), acquiring an image captured by an imaging device (22), and calculating a distance from a vehicle to an object (50) present in the image,
a first calculation step of calculating a distance from the host vehicle to the object as a first distance based on a bottom position that is a position of a bottom end portion of the object in the image;
a second calculation step of calculating a distance from the host vehicle to the object as a second distance based on a size of the object in the image;
an actual distance calculation step of calculating an actual distance from the host vehicle to the object based on the first distance and the second distance;
a reliability determination step of determining a reliability of the actual distance based on a distance difference between the first distance and the second distance;
Programs including.

Images