JP2018088237A - Information processing device, imaging device, apparatus control system, movable body, information processing method, and information processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to accurately detect an object, such as a parallax image from information with which the position in the horizontal direction and the position in the depth direction of the object are associated.SOLUTION: A creation part creates two-dimensional distribution information indicating two-dimensional frequency distribution with which the positions in the horizontal direction and the positions in the depth direction of objects and a frequency value are associated, and a detection part detects detection objects on the basis of the two-dimensional distribution information. A separation position specification part specifies a separation position on the basis of the amount of feature created by integrating the frequency value according to the positions in the depth direction. An object separation part separates the detection objects at the specified separation position. The information processing device thus can separate and detect the respective objects connected by a parallax image and detected.SELECTED DRAWING: Figure 8

Description

  The present invention relates to an information processing device, an imaging device, a device control system, a moving body, an information processing method, and an information processing program.

  2. Description of the Related Art Today, automobile body structures and the like have been developed from the viewpoint of automobile safety, such as how to protect pedestrians and protect passengers in the event of a collision with a pedestrian or automobile. In recent years, with the development of information processing technology and image processing technology, technology for detecting humans, automobiles and the like at high speed is known. There is also known an automobile provided with a collision prevention system that uses these techniques to automatically brake before a collision to prevent the collision. In the case of this collision prevention system, a distance to a detection target such as a person or another vehicle is measured using a millimeter wave radar device, a laser radar device, a stereo camera device, or the like, and a brake is determined based on the result of the distance measurement. Take control. As a result, the brake can be automatically applied according to the distance to the detection target.

  Patent Document 1 (Japanese Patent Laid-Open No. 2014-096005) discloses an object detection apparatus that accurately detects grouped objects that are regarded as identical. This object detection apparatus groups the effective areas using the distance image, and then divides the area by focusing on the contour portion (edge) of the grayscale image with respect to the grouped objects.

  Patent Document 2 (International Publication No. 2012/017650) discloses an object detection device that can detect an object and a road surface with high accuracy. After detecting the road surface area, the object detection device detects data having a height equal to or higher than the road surface as an object candidate area, and determines the object and the road surface based on the shape feature.

  However, the distance measurement accuracy of the distance to the detection target decreases as the distance to the detection target increases. For this reason, there are cases where a plurality of objects to be detected located in the distance are coupled to each other and inconvenience is detected as one object.

  The present invention has been made in view of the above-described problems, and is an information processing device, an imaging device, a device control system, a moving body, an information processing method, and an information processing device capable of accurately separating and detecting a plurality of detection objects. An object is to provide an information processing program.

  In order to solve the above-described problems and achieve the object, the present invention generates two-dimensional distribution information indicating a two-dimensional frequency distribution in which the horizontal position and depth direction position of an object are associated with frequency values. A generation unit, a detection unit that detects a detection target based on two-dimensional distribution information, and a frequency value that is integrated according to a depth direction position to generate a feature amount, and a separation that specifies a separation position based on the feature amount A position identifying unit; and an object separating unit that separates the detection target at the separation position.

  According to the present invention, there is an effect that detection objects can be accurately separated and detected.

FIG. 1 is a diagram illustrating a position of a stereo camera provided in a vehicle that is a device control system according to the first embodiment. FIG. 2 is a diagram illustrating a stereo camera included in the vehicle and a configuration around the stereo camera. FIG. 3 is a diagram illustrating a configuration example of an imaging apparatus including a stereo camera. FIG. 4 is a functional block diagram of functions realized by the CPU of the device control system according to the first embodiment based on the vehicle control program. FIG. 5 is a diagram illustrating an example of the U map. FIG. 6 is a diagram showing a state in which one isolated region is set for two objects on the U map. FIG. 7 is a diagram illustrating the detection target object separated into two at the division position specified based on the feature amount. FIG. 8 is a flowchart illustrating a flow of separation processing according to the embodiment. FIG. 9 is a diagram for explaining how to generate a one-dimensional histogram. FIG. 10 is a diagram for explaining the class separation calculation operation. FIG. 11 is a diagram illustrating a state in which isolated regions are accurately set for two objects on the U map.

  Hereinafter, an apparatus control system according to an embodiment will be described with reference to the drawings.

(System configuration)
First, the device control system according to the embodiment includes a stereo camera 2 that is provided on the windshield of the vehicle 1 as shown in FIG. As will be described later with reference to FIG. 3, the stereo camera 2 includes two image sensors 22 and is an imaging unit that captures two images of the left visual field and the right visual field.

  FIG. 2 is a diagram illustrating a configuration example of the stereo camera 2 included in the vehicle 1 which is an example of the moving body and the vicinity thereof. The stereo camera 2 outputs, for example, two captured images to a vehicle ECU (Engine Control Unit) 3. The vehicle ECU 3 is installed in the vehicle 1 and controls the vehicle 1 such as engine control, braking control, travel lane keep assist, steering assist, and the like. Hereinafter, a vehicle will be described as an example of a moving body, but the device control system of the present embodiment can also be applied to a ship, an aircraft, a robot, and the like.

(Configuration of imaging device)
FIG. 3 is a diagram illustrating a configuration example of the imaging device 4 including the stereo camera 2. The imaging device 4 includes, for example, a stereo camera 2 and an image processing device 30. In the stereo camera 2, a camera unit 2 a for the left eye and a camera unit 2 b for the right eye are assembled in parallel (horizontal) to shoot a moving image (or still image) in the imaging target area.

  Each of the camera units 2a and 2b includes a lens 21, an image sensor 22, and a sensor controller 23. The image sensor 22 is, for example, a CCD (Charge Coupled Device) image sensor or a CMOS (Complementary Metal-Oxide Semiconductor) image sensor. The sensor controller 23 performs, for example, exposure control of the image sensor 22, image readout control, communication with an external circuit, image data transmission control, and the like.

  The image processing device 30 is provided, for example, in the vehicle ECU 3 shown in FIG. The image processing apparatus 30 includes, for example, a data bus line 300, a serial bus line 302, a CPU (Central Processing Unit) 304, an FPGA (Field-Programmable Gate Array) 306, a ROM (Read Only Memory) 308, and a RAM (Random Access Memory) 310. Serial IF (Interface) 312 and data IF (Interface) 314.

  The stereo camera 2 described above is connected to the image processing apparatus 30 via the data bus line 300 and the serial bus line 302. The CPU 304 controls the overall operation of the image processing apparatus 30 and executes image processing and image recognition processing. Luminance image data of captured images captured by the image sensors 22 of the camera units 2 a and 2 b are written into the RAM 310 of the image processing device 30 via the data bus line 300. Sensor exposure value change control data, image readout parameter change control data, various setting data, and the like from the CPU 304 or FPGA 306 are transmitted and received via the serial bus line 302.

  The FPGA 306 generates real-time parallax images by performing parallax calculation based on, for example, gamma correction, distortion correction (parallelization of left and right images), and block matching, which is a process that requires real-time processing for image data stored in the RAM 310. , Write again to the RAM 310. Note that the parallax image is information in which the vertical position, the horizontal position, and the depth direction position of the object are associated with each other.

  The CPU 304 performs control of each sensor controller 23 of the stereo camera 2 and overall control of the image processing apparatus 30. The CPU 304 acquires, for example, CAN (Controller Area Network) information of the host vehicle as parameters (vehicle speed, acceleration, steering angle, yaw rate, etc.) via the data IF 314.

  Vehicle control data, which is recognition data for objects such as preceding vehicles, humans, guardrails, and road surfaces, is supplied to the vehicle ECU 3 via the serial IF 312 and provided as a control function of the vehicle ECU 3, for example, an automatic brake system or a travel assist system. Etc. The automatic brake system performs brake control of the vehicle 1. The travel assist system performs travel lane keep assist, steering assist, and the like of the vehicle 1.

(Function of image processing device)
The CPU 304 of the image processing device 30 executes the vehicle control program stored in the storage unit such as the ROM 308 to realize each function illustrated in FIG. 4, and performs the preceding based on the parallax image captured by the imaging device 4. Detects objects such as vehicles, people, guardrails, and road surfaces. Then, vehicle control data is supplied to the vehicle ECU 3 according to the detected object. The vehicle ECU 3 performs brake control, travel lane keep assist, steering assist, and the like using the vehicle control data.

  In other words, the CPU 304 executes a vehicle control program stored in a storage unit such as the ROM 308, so that the voting processing unit 51, the isolated region detection unit 52, the depth separation unit 53, the voting correction unit, as shown in FIG. 54, the functions of the object detection unit 55 and the vehicle control unit 56 are realized.

  In this example, the voting processing unit 51 to the vehicle control unit 56 are realized by software. However, a part or all of the voting processing unit 51 to the vehicle control unit 56 is an IC (Integrated Circuit) or the like. It may be realized by hardware.

  Further, the vehicle control program may be provided by being recorded in a computer-readable recording medium such as a CD-ROM or a flexible disk (FD) in an installable or executable file. Further, the program may be provided by being recorded on a recording medium readable by a computer device such as a CD-R, a DVD (Digital Versatile Disk), a Blu-ray Disc (registered trademark), or a semiconductor memory. The vehicle control program may be provided by being installed via a network such as the Internet. The vehicle control program may be provided by being incorporated in advance in a ROM or the like in the device.

  The voting processing unit 51, which is an example of a generation unit, shows a frequency distribution in which the horizontal position and the depth direction position of an object are associated with each other based on a parallax image formed from a captured image of the imaging device 4. A U map (an example of two-dimensional distribution information) is generated. For convenience of explanation, the term “map” is used, but the map or image information is not actually formed, but an information group in which the horizontal position and the depth direction position are associated is formed. Should be understood.

  Specifically, the voting processing unit 51 calculates the frequency distribution of the parallax values with the horizontal axis representing the actual distance between the objects and the vertical axis representing the thinning parallax that varies the thinning rate according to the distance between the objects. A U map (bird's-eye view image, bird's-eye view image, bird's-eye view map), which is a two-dimensional histogram shown, is generated. Although it is an example, the voting processing unit 51 does not thin out parallax when the distance to the object is 50 m or more, but when the distance to the object is a medium distance of 20 m to less than 50 m, The parallax is thinned out to ½. The voting processing unit 51 thins out the parallax to 1/3 when the distance to the object is a short distance of 10 m to less than 20 m, and sets the parallax to 1 when the distance to the object is less than 10 m. / 8 is thinned out. Since the object is small in the distance, the parallax information is small, and the resolution of the distance is small, the parallax thinning process is not performed. On the other hand, in the case of a short distance, an object is photographed large, so that there is a large amount of parallax information and the resolution of the distance is large. In addition, since this process generates a bird's-eye view map (bird's-eye view image, bird's-eye view image) in order to make it easier to detect an object, the horizontal axis may be equivalent to an actual distance, not an actual distance. .

  FIG. 5 is an example of the U map generated using the thinning parallax according to the distance between the object and the object. The example of FIG. 5 shows a state in which the road between the left and right guard rails GL-GR has two lanes, and the vehicles C1 and C2 are running substantially side by side in each lane. In this example, the voting processing unit 51 generates a U map with the vertical axis representing the parallax value and the horizontal axis representing the actual distance, but the vertical axis represents the parallax value and the horizontal axis represents the value of the X coordinate system. The U map may be generated, and the processing described later may be executed using the U map. Even in this case, it is possible to obtain the same effect as in the case of using a U map in which the vertical axis represents the parallax value and the horizontal axis represents the actual distance. That is, it is only necessary to use an information group in which the horizontal position and the depth direction position of an object are associated with each other regardless of conversion between the vertical axis and the horizontal axis. In the following description, it is assumed that the depth direction is converted to a distance value instead of a parallax value. Note that information may be handled without changing the parallax value.

  Next, the isolated region detection unit 52 shown in FIG. 4 as an example of the detection unit detects each isolated region corresponding to an object existing on the U map. FIG. 6 shows an example of detecting an isolated area. In the case of the example of FIG. 6, the vehicle 1 provided with the device control system of the embodiment travels on the left lane of a two-lane straight road. This is an example where TR is running. Further, in the right lane, a taxi car TK, a light truck car KJ1, and a light truck car KJ2 are traveling in order from the front. Of the cars traveling in the right lane, the taxi car TK traveling in the foreground is in a state where the entire vehicle can be confirmed. Further, in the light truck car KJ1 traveling in front of the taxi car TK, the right tail lamp overlaps with the right end portion of the bonnet of the taxi car TK. The light truck automobile KJ2 is in a state where the left half is substantially overlapped with the light truck automobile KJ1 traveling behind and only the right half can be confirmed. The isolated area can be detected by using a well-known labeling process (a process for assigning IDs to adjacent pixels) and extracting those having the same ID and having a certain size or more. Not limited to this, various known methods can be applied.

(Separation process)
Here, it is assumed that the isolated region detection unit 52 erroneously detects the taxi car TK and the light truck car KJ1 as one vehicle because the objects having close distance values (parallax values) overlap each other. A frame (detection frame) surrounding the taxi car TK and the light truck car KJ1 shown in FIG. 6 indicates that such a false detection has occurred. Here, the frame indicates the position and size of the object, and is information in which the coordinates of the corners of a rectangle surrounding the object are associated with the height and width, for example.

  The depth separation unit 53, which is an example of the separation position specifying unit, is located in a distance range in which the isolated region detected by the isolated region detection unit 52 is likely to cause the coupling between objects such as the front and rear vehicles due to distance dispersion (parallax dispersion). If it is (if the depth separation start condition is satisfied), the depth separation process is executed. In other words, the distance separation (parallax dispersion) becomes large in the distance, and the objects are likely to be combined. Specifically, for example, “an isolated region having a closest distance of 30 m or more” is defined as the depth separation start condition. A numerical value such as “30 m” may be changed according to the performance of the stereo camera of the imaging device 4. In the case of the embodiment, the processing load on the CPU 304 can be reduced by not performing the depth separation processing described below for an isolated region located at a short distance.

  The depth separation unit 53 is a one-dimensional feature amount (product of frequency and distance value) having properties equivalent to the X coordinate of the image (the horizontal coordinate of the U map) for an isolated region that satisfies the above-described depth separation start condition. ) Is calculated. That is, the product of the frequency and the distance value corresponding to each X coordinate (or horizontal position) is calculated. Further, the depth separation unit 53 divides the calculated feature value at all the X coordinate positions (separation position candidates), calculates an evaluation value (feature value) for separating an object in the isolated region, and evaluates the evaluation value. The division position where becomes the maximum is specified. In this example, the division position where the evaluation value is the maximum is specified, but the division position where the evaluation value is the second largest, or the division position where the evaluation value is the second largest, etc. You may change the division | segmentation position to identify.

  The product of the frequency and the distance value described above is a feature amount corresponding to the degree of separation (ratio of interclass variance to intraclass variance) in a generally well-known discriminant analysis method. Therefore, by specifying the division position based on this feature amount, the target can be appropriately separated into two groups as shown in FIG. That is, the detection object can be accurately separated.

  The depth separating unit 53 divides the isolated region into two isolated regions when the difference in the average distance between the objects separated at the specified dividing position is longer than the length of one vehicle. The isolated region located in the depth direction is separated based on the statistical processing of the division position and the difference in the average distance, not the probability of the width. Thereby, it becomes difficult to be adversely affected by the distance dispersion (parallax dispersion) and the two-dimensional distribution of the U map, and for example, two vehicles traveling in front and rear can be correctly separated and detected.

  The flowchart of FIG. 8 shows the flow of the depth separation process. In step S1, the depth separation unit 53 calculates a feature amount of “distance value (Y coordinate value, depth) × frequency value” for each pixel (voting information) that forms an isolated region that satisfies the above-described depth separation start condition. . Then, the depth separation unit 53 performs integration processing on the feature amount calculated by each pixel for each column in the depth direction of the isolated region, and generates a one-dimensional histogram. Such a one-dimensional histogram has a dimension equivalent to the X axis of the parallax image. In this embodiment, for each X coordinate (lateral position) of the parallax image, the product of the frequency and the distance value corresponding to the X coordinate is calculated. However, the present invention is not limited to this, and the scale of the parallax image is reduced. For each X coordinate, a product of a frequency and a distance value corresponding to the X coordinate may be calculated.

  FIG. 9 is a diagram illustrating how a one-dimensional histogram is generated. FIG. 9 shows an example in which the taxi car TK shown in FIG. 6 and the light truck car KJ1 traveling in front of the taxi car TK are combined and detected as one isolated region. In FIG. 9, the lump on the depth side indicates the light truck car KJ1, and the front side indicates the taxi car TK. Further, the numerical value in the pixel in the light truck automobile KJ1 is an example of the feature amount calculated by the arithmetic expression of “distance value (Y coordinate value) × frequency value”.

  That is, in the example of FIG. 9, there is only one pixel in the first column of the light truck automobile KJ1, and the feature amount of “distance value (Y coordinate value) × frequency value” is “1”. Further, there are two pixels in the second column of the light truck automobile KJ1, and the feature values of “distance value (Y coordinate value) × frequency value” are “3” and “4”, respectively. The depth separation unit 53 performs integration processing of such feature amounts in the column direction (in FIG. 9, a symbol “+” indicates integration processing). As a result, a one-dimensional histogram of “1” for the first pixel and “7 (= 3 + 4)” for the second pixel is generated. By using such a one-dimensional histogram, it is possible to shorten the time for separation processing, which will be described later, and to use an image coordinate axis, so that accurate object separation processing can be performed.

  In the example of FIG. 9, one pixel of the light truck car KJ1 or taxi car TK does not necessarily correspond to one pixel of the one-dimensional histogram. As shown in the example of FIG. 9, a plurality of pixels of the light truck car KJ1 or taxi car TK may correspond to one pixel of the one-dimensional histogram.

  Next, in step S2, the depth separation unit 53 determines whether or not the width of the one-dimensional histogram generated in step S1 is larger than a predetermined width. Specifically, the depth separation unit 53 determines whether or not the width of the one-dimensional histogram is larger than “2 m”, which is the width of one normal vehicle, for example. Note that the width of the one-dimensional histogram can be converted into the width of the real world from the distance value within the detection frame of the isolated region.

  The fact that the width of the one-dimensional histogram is smaller than the width of one normal vehicle means that the isolated region is not an isolated region in which a plurality of objects are combined. For this reason, when the depth separation unit 53 determines that the width of the one-dimensional histogram is smaller than the width of one normal vehicle, the depth separation unit 53 stops the separation processing for the isolated region and performs the processing of the flowchart of FIG. finish. On the other hand, if the width of the one-dimensional histogram is larger than the width of one normal vehicle, the isolated area is likely to be an isolated area in which a plurality of objects are combined. For this reason, the depth separation unit 53 advances the process to step S3.

  In step S3, the depth separation unit 53 divides the one-dimensional histogram into two groups in order at each X coordinate within the histogram frequency range, and calculates the class separation degree evaluation value at each separation position. Perform the calculation process. FIG. 10 is a schematic diagram showing the processing content of the class separation degree calculation processing. As shown in FIG. 10, the depth separation unit 53 separates the one-dimensional histogram into two groups in the order of the X coordinate within the range where the histogram frequency is within a certain range. An arrow B shown in FIG. 10 indicates a separation position of the one-dimensional histogram, and currently indicates that the one-dimensional histogram is separated into two at approximately the center. For example, the depth separation unit 53 sequentially shifts the separation position from the left end to the right end of the one-dimensional histogram, and calculates the evaluation value of the class separation degree at each separation position.

  Note that the separation position of the one-dimensional histogram may be moved in any way, for example, may be sequentially moved from the right end to the left end of the one-dimensional histogram. Alternatively, first, the one-dimensional histogram is separated from the middle, the separation position is moved in order from the middle to the right end, and the computation is performed. After the computation corresponding to the right end is completed, the separation position of the one-dimensional histogram is returned to the middle, The calculation may be performed by moving the separation position in order from the left to the left end.

  As shown in FIG. 10, the depth separation unit 53 calculates the sum of the frequency values from the separation position to the left end of the one-dimensional histogram as “w1”, and the feature amount from the separation position to the left end (distance value × frequency value). The average value of “m1”, the sum of the frequency values from the separation position to the right end is “w2”, and the average value of the feature amount (distance value × frequency value) from the separation position to the right end is “m2” Then, the class separation degree is calculated by performing the following equation (1). The average values m1 and m2 correspond to the dimension of the distance value (Y coordinate value) of the U map. The average value m1 indicates an average value of feature amounts between the separation position of the one-dimensional histogram and the left end, and the average value m2 indicates an average value of feature amounts between the separation position of the one-dimensional histogram and the right end.

Class separation = w1 × w2 × (m1-m2) ^2.

  The depth separation unit 53 calculates the average values m1 and m2 by performing the following equation (2).

  Average value = (Σ (distance value × frequency value)) / (Σ frequency value) Equation 2

  Next, in step S4, the depth separation unit 53 identifies the separation position where the maximum class separation degree is calculated among the calculated class separation degrees. Further, the depth separation unit 53 calculates the difference (distance difference) between the average value m1 and the average value m2 of the two groups formed with the identified separation position as a boundary. In step S5, the depth separation unit 53 determines whether or not the calculated distance difference exceeds a predetermined value such as 10 m, which is the vehicle length of one large vehicle.

  The fact that the calculated distance difference does not exceed 10 m, which is the vehicle length of one large vehicle, means that each object corresponding to the two groups formed with the specified separation position as a boundary is, for example, a truck bed And an integrated object (isolated region) such as a driver's seat. For this reason, the depth separation unit 53 stops the separation process for the isolated region and ends the process of the flowchart of FIG.

  On the other hand, that the calculated distance difference exceeds 10 m, which is the vehicle length of one large vehicle, means that each object corresponding to the two groups formed with the specified separation position as a boundary is For example, it means physically different objects (isolated regions) such as two vehicles. For this reason, the depth separation unit 53 advances the processing to step S36, separates the isolated region combined into one at the specified separation position into two isolated regions, and ends the flowchart of FIG. . In such a separation process, it is difficult to be adversely affected by the parallax dispersion and the two-dimensional distribution of the U map, and for example, two vehicles traveling in the front-rear direction can be correctly separated.

  As described above, the feature amount (one-dimensional histogram) obtained by integrating the distance value and the frequency for each horizontal position is calculated from the information in which the horizontal position, the depth direction position (distance value), and the frequency of the object are associated with each other. Then, the object is separated based on the feature amount. As a result, the object can be accurately separated.

  The voting correction unit 54 shown in FIG. 4, which is an example of an object separation unit, sets an isolated region (separates the isolated region) on the U map based on the separation processing result of the depth separation unit 53. Fix isolated regions. FIG. 11 is a diagram illustrating an isolated region that has been reset. Before the correction, the taxi car TK traveling in the right oblique line and the light truck car KJ1 traveling in front of the taxi car TK have one isolated region set as one object as described with reference to FIG. It was. However, by performing the above-described separation processing, as shown in FIG. 11, it is possible to reset the isolated region with the taxi car TK and the light truck car KJ1 traveling in front of the taxi car TK as different objects. it can.

  The object detection unit 55 shown in FIG. 4 performs object detection processing for the corresponding vehicle, pedestrian, guardrail, sign, and the like based on each isolated region set accurately in this way. And the vehicle control part 56 supplies the vehicle control data used as an object detection result to vehicle ECU3 shown in FIG. The vehicle ECU 3 controls the vehicle 1 such as engine control and braking control of the vehicle 1 and travel lane keeping assist and steering assist.

  The above-described embodiments are presented as examples, and are not intended to limit the scope of the present invention. The novel embodiment can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. For example, a distance value (distance value) and a parallax value can be handled equivalently. For this reason, in the description of the above-described embodiment, a parallax image is used as an example of a distance image, but the present invention is not limited to this. For example, a distance image may be generated by integrating distance information generated using a detection device such as a millimeter wave radar or a laser radar with a parallax image generated using a stereo camera. Further, the detection accuracy may be further enhanced by using a stereo camera and a detection device such as a millimeter wave radar or a laser radar in combination with the detection result of the object by the stereo camera described above.

  Such embodiments and modifications of the embodiments are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

1 vehicle 2 stereo camera (imaging part)
2a Camera unit 2b Camera unit 3 Vehicle ECU
4 imaging device 30 image processing device 51 voting processing unit 52 isolated region detection unit 53 depth separation unit 54 voting correction unit 55 object detection unit 56 vehicle control unit

JP 2014-096005 A International Publication No. 2012/017650

Claims (10)

A generating unit that generates two-dimensional distribution information indicating a two-dimensional frequency distribution in which a horizontal position and a depth direction position of an object are associated with a frequency value;
A detection unit for detecting a detection object based on the two-dimensional distribution information;
The frequency value is integrated according to the depth direction position to generate a feature amount, and a separation position specifying unit that specifies a separation position based on the feature amount;
An information processing apparatus comprising: an object separation unit that separates the detection target at the separation position. The separation position specifying unit calculates a difference between one depth direction position and the other depth direction position of the detection object separated into two at the separation position, and the separation position where the difference exceeds a predetermined value The information processing apparatus according to claim 1, wherein: The information according to claim 1 or 2, wherein the separation position specifying unit specifies the separation position when a depth direction position of the detection object is farther than a predetermined position. Processing equipment. The said separation position specific | specification part produces | generates the said feature-value by integrating | accumulating the said frequency value corresponding to the horizontal direction position of the said detection target object along the said depth direction. The information processing apparatus according to claim 2. The separation position specifying unit specifies a plurality of separation position candidates, and specifies a separation position where the feature amount calculated based on each separation position candidate satisfies a predetermined condition. The information processing apparatus according to any one of claims 1 to 4. An imaging unit;
An imaging device comprising: the information processing device according to any one of claims 1 to 5. The information processing apparatus according to any one of claims 1 to 5,
A device control system comprising: a control unit that controls a predetermined device based on object information of the object separated by the object separation unit. It is controlled by the said control part of the apparatus control system of Claim 7, The moving body characterized by the above-mentioned. A generating step for generating two-dimensional distribution information indicating a two-dimensional frequency distribution in which the horizontal position and depth direction position of the object are associated with the frequency value;
A detecting step for detecting a detection object based on the two-dimensional distribution information;
A separation position specifying unit that generates the feature amount by integrating the frequency value according to the depth direction position, and specifies the separation position based on the feature amount; and
An object separation method in which an object separation unit separates the detection target at the separation position. Computer
A generating unit that generates two-dimensional distribution information indicating a two-dimensional frequency distribution in which a horizontal position and a depth direction position of an object are associated with a frequency value;
A detection unit for detecting a detection object based on the two-dimensional distribution information;
The frequency value is integrated according to the position in the depth direction to generate a feature amount, and a separation position specifying unit that specifies a separation position based on the feature amount;
An information processing program that functions as an object separation unit that separates the detection target at the separation position.

Images