JP5996421B2 - Obstacle detection device and vehicle using the same - Google Patents

Description

  The present invention relates to an obstacle detection device that detects an obstacle on a road using an image sensor and a vehicle including the obstacle detection device.

  2. Description of the Related Art Conventionally, some golf carts that run on a course in a golf course have an automatic running function. Automatic traveling is performed by detecting a magnetic field generated from an electromagnetic induction wire. In addition, there is an apparatus provided with an obstacle detection device that detects an obstacle in front of the golf cart by providing an ultrasonic sensor in front of the golf cart. When an obstacle is detected by the ultrasonic sensor, the golf cart is controlled to decelerate or stop.

(1) Technology of Patent Document 1 The golf cart described in Patent Document 1 is equipped with an automatic traveling device. Thereby, the traveling road can be automatically traveled according to the guide line provided on the traveling road. The distance to the obstacle is measured by the ultrasonic sensor, and the presence or absence of the obstacle is determined according to the distance.

JP 2003-5832 A

Conventional golf carts, as shown in FIG. 16, depending on whether an obstacle within the front of the vehicle distance L ₁ is present, and the control of the travel speed. The distance L ₁ is set to a range that does not contain the track out of the obstacle. However, if the running speed of the golf cart is increased, the braking distance becomes longer, and therefore obstacles ahead must be detected.

Therefore, when increasing the running speed, as shown in FIG. 17, based on whether the judgment result obstacle within the front of the vehicle distance L ₂ is present, the control of the travel speed. However, an obstacle outside the track may be detected by increasing the obstacle determination distance. The golf cart is decelerated or stopped by detection of an obstacle outside the running road, and there is a problem that the running efficiency of the golf cart is deteriorated even though the running speed is increased.

  This invention is made | formed in view of such a situation, Comprising: The obstacle detection apparatus which can determine whether an obstacle exists in a runway, and a vehicle using the same are provided. Objective.

In order to achieve the above object, the present invention has the following configuration.
That is, the obstacle detection device according to the present invention includes an image sensor that captures a front image when the characteristics of image information are different between the ground area of the road and the ground area outside the road , and the obstacle in the image. said the obstacle detecting unit for detecting an object, the obstacle ground area setting unit for setting a ground area of the obstacle in the image, the ground area and the ground area of the runway of the obstacle in the image A similarity calculation unit that calculates the similarity of the characteristics .

  According to the present invention, the image sensor captures a front image. The obstacle detection unit detects an obstacle in the captured image. The obstacle ground area setting unit sets the ground area of the detected obstacle in the captured image. The similarity determination unit calculates the similarity between the ground area of the obstacle and the ground area of the runway in the image. If the calculated similarity value is large, the obstacle ground area and the road ground area are similar, and therefore it can be determined that the obstacle exists on the road. If the calculated similarity value is small, the obstacle ground area and the road ground area are not similar, and therefore it can be determined that no obstacle exists on the road. In this way, it can be determined whether the detected obstacle is present in the runway or outside the runway.

  In addition, it is preferable that a distance information detection unit that detects distance information with a front object is provided, and the obstacle detection unit detects an obstacle in the image based on the distance information. Since the obstacle is detected using the distance information, the obstacle detection accuracy can be improved.

  Preferably, the image sensor includes a plurality of image sensors in a parallel stereo positional relationship, and further includes a parallax image creation unit that creates a parallax image based on an image of each image sensor. Since the image sensor includes a plurality of image sensors in a parallel stereo positional relationship and further includes a parallax image creation unit that creates a parallax image based on the image of each image sensor, the distance associated with the image information and the image information in front of the vehicle Information can be acquired with high accuracy.

  Moreover, it is preferable to provide a runway ground area setting unit that sets, in the image, a ground area of a runway having the same area as the area of the obstacle ground area in real space. Since the ground area of the obstacle and the ground area of the runway have the same area in real space, the similarity can be appropriately calculated based on the respective density values.

  In addition, the runway ground area setting unit calculates an area of the ground area of the obstacle in real space, and calculates the ground area of the runway having the same size in the real space as the calculated area. And a ground area setting unit that is set inside. According to this configuration, the ground area of the running road can be appropriately set so as to have the same area as the obstacle ground area in real space.

  In addition, it is preferable to include a normalization unit that normalizes the density value of the ground area of the obstacle and the density value of the ground area of the runway with respect to the histogram obtained by quantizing the gradation while reducing the gradation. A histogram is created by quantizing the density value of the obstacle ground area and the density value of the runway ground area while reducing gradation. Since normalization is performed on these histograms, it is possible to reduce the calculation time of similarity by reducing the amount of calculation of similarity, and further improve the accuracy of similarity.

  Moreover, it is preferable to provide a comparison unit that determines whether or not an obstacle exists on the runway by comparing the similarity with a predetermined threshold value. By comparing the similarity with a predetermined threshold value, it can be easily determined whether or not an obstacle exists on the road.

  Moreover, the vehicle which concerns on this invention is a vehicle provided with the said obstacle detection apparatus. Since the obstacle detection device is provided, it can be accurately determined whether or not the detected obstacle exists on the road. Even if an obstacle outside the track is detected, it can be properly recognized that the obstacle is outside the track, thus preventing the vehicle from decelerating or stopping due to an obstacle outside the track. Can do. Thereby, the running efficiency of the vehicle can be improved.

  Further, it is determined whether or not the vehicle deceleration control is to be performed by comparing the distance to the obstacle and the first distance, and the distance to the obstacle and the second distance longer than the first distance are determined. It is preferable to provide an obstacle distance determination unit that determines whether or not the similarity between the ground area of the obstacle and the ground area of the vehicle is calculated by comparing.

  The vehicle speed control of the vehicle is performed on the basis of the two-step distance reference with respect to the detected distance to the obstacle. It is determined whether or not the vehicle deceleration control is to be performed by comparing the distance to the obstacle with the first distance in the first stage. Thereby, when the distance between the obstacle and the vehicle is short, the deceleration control of the vehicle can be performed, and the collision between the vehicle and the obstacle can be avoided.

  Further, it is determined whether or not the similarity between the ground area of the obstacle and the ground area of the vehicle is calculated by comparing the distance to the obstacle with the second distance of the second stage longer than the first stage. . By this, when the distance between the obstacle and the vehicle is some distance away, after calculating the similarity between the obstacle ground area and the road ground area, and determining whether the obstacle is on the road, Vehicle speed control of the vehicle can be implemented.

  ADVANTAGE OF THE INVENTION According to this invention, the obstruction detection apparatus which can determine whether an obstruction exists in a runway, or exists out of a runway, and a vehicle using the same can be provided.

1 is a front view of a vehicle according to a first embodiment. It is a block diagram which shows the structure of the vehicle which concerns on an Example. 1 is a block diagram illustrating a configuration of an obstacle detection device according to Embodiment 1. FIG. It is explanatory drawing which shows an obstruction area | region setting. It is explanatory drawing which shows an obstruction ground area | region and a runway ground area | region. 6 is a flowchart illustrating a flow of obstacle determination according to the first embodiment. FIG. 6 is a front view of a vehicle according to a second embodiment. It is a block diagram which shows the structure of the obstruction detection apparatus which concerns on Example 2. FIG. It is explanatory drawing which shows the obstacle detection which concerns on Example 2. FIG. It is a block diagram which shows the partial structure of the obstruction detection apparatus which concerns on Example 2. FIG. It is explanatory drawing which shows an obstruction ground area | region and a runway ground area | region. It is explanatory drawing which shows the density | concentration histogram of an obstruction ground area | region. It is explanatory drawing which shows the density | concentration histogram of a runway ground area | region. It is explanatory drawing which shows the density | concentration histogram of an obstruction ground area | region. 10 is a flowchart illustrating a flow of obstacle detection according to the second embodiment. It is explanatory drawing which shows the obstruction detection area | region which concerns on a prior art example. It is explanatory drawing which shows the obstruction detection area | region which concerns on a prior art example.

  Embodiment 1 of the present invention will be described below with reference to the drawings. As an embodiment of a vehicle in the present invention, a golf cart that automatically runs is given. The vehicle is not limited to a four-wheeled vehicle, and may be a three-wheeled vehicle or a monorail type. Further, the vehicle is not limited to a golf cart, and includes an automatic guided vehicle that runs in a factory or an orchard. In the following description, front and rear and left and right are based on the direction in which the vehicle 1 moves forward.

1. Schematic configuration of vehicle Referring to FIGS. 1 and 2. FIG. 1 is a front view of a vehicle 1 according to the embodiment, and FIG. 2 is a functional block diagram showing a configuration of the vehicle 1. The vehicle 1 is a golf cart that travels automatically or manually in a golf course. The vehicle 1 can be automatically driven by being guided by electromagnetic waves emitted from a guide wire embedded in the runway. An image sensor 3 is provided in the center of the front surface of the vehicle 1. An example of the image sensor 3 is a visible light camera.

  Further, the vehicle 1 includes an obstacle detection device 5 that detects an obstacle on the road on which the vehicle 1 travels, an automatic travel control unit 7 that controls the vehicle 1 to automatically travel along the guide line, When the object detection device 5 detects an obstacle, the warning output unit 8 generates a warning to the driver and the surroundings, the vehicle speed control unit 9 that controls deceleration or stop by detecting the obstacle, and the wheel is driven to control the vehicle speed. A drive motor 11 whose rotational speed is controlled by the unit 9 is provided. In the present embodiment, the vehicle 1 is driven by a motor, but is not limited thereto, and may be driven by an engine.

2. Configuration of Obstacle Detection Device Next, the configuration of the obstacle detection device 5 provided in the vehicle 1 will be described with reference to FIG. FIG. 3 is a block diagram illustrating a configuration of the obstacle detection apparatus.

  The obstacle detection device 5 includes an image sensor 3 that captures an image in front of the vehicle 1, an obstacle detection unit 15 that detects an obstacle based on an image captured by the image sensor 3, and the detected obstacle. An obstacle ground area setting unit 17 that sets a ground area, and an obstacle determination unit 23 that determines whether an obstacle exists on the road from the similarity of the texture between the ground area of the obstacle and the ground area of the road. With. The obstacle detection unit 15, the obstacle ground area setting unit 17, and the obstacle determination unit 23 are configured by a microprocessor or an FPGA (Field Programmable Gate Array). Next, each component will be described in order.

  The obstacle detection unit 15 detects an obstacle from the image sent from the image sensor 3 by, for example, template matching. An example of a template is one in the form of a person and a vehicle. In addition to template matching, an object such as a person or a vehicle may be detected as an obstacle using a HOG (Histograms of Oriented Gradients) feature.

  The obstacle ground area setting unit 17 sets the ground area of the detected obstacle. The obstacle ground area setting unit 17 sets an area on the detected obstacle image on the obstacle area setting unit 19 and a ground area that sets a lower pixel area of the set obstacle area as an obstacle ground area. And a setting unit 21.

  This will be described with reference to FIG. FIG. 4 is an explanatory diagram showing the obstacle area setting. The obstacle area setting unit 19 sets the obstacle area Ba1 that surrounds the obstacle with a circumscribed rectangle by extracting the maximum value and the minimum value of the uv coordinates on the image DP1 of the detected obstacle Bs1.

  The ground area setting unit 21 sets the lower pixel area immediately below the set obstacle area Ba1 as the obstacle ground area Bg1. The width in the u-axis direction of the obstacle ground area Bg1 is set to the same size as the width of the obstacle area Ba1. The height of the obstacle ground area Bg1 in the v-axis direction is set to a predetermined size. Data of the density value of each pixel of the set obstacle ground area Bg1 is output to the obstacle determination unit 23.

  The obstacle determination unit 23 determines whether an obstacle exists on the road, based on the similarity S between the obstacle ground area Bg1 and the road ground area Cg1. The obstacle determination unit 23 calculates a similarity S between the obstacle ground region Bg1 and the road ground region Cg1 in the image, and a comparison unit 27 that compares the calculated similarity S with a threshold T. And have.

  The similarity calculation unit 19 calculates a similarity S between a predetermined density histogram of pixels in the road ground area Cg1 and a density histogram of pixels in the obstacle ground area Bg1. The runway ground region Cg1 is located in the center in the u-axis direction and in the lower part in the v-axis direction in the image Dp1. That is, a portion obtained by photographing the center front closest to the vehicle 1 (image sensor 3) is defined as a runway ground region Cg1. The runway ground area Cg1 is taken on the runway Rd. By calculating the similarity S between the obstacle ground area Bg1 and the running road ground area Cg1, the similarity between the obstacle ground area Bg1 and the running road Rd is calculated.

  The similarity S can be calculated, for example, by using a Bhattacharyya coefficient represented by the following equation. Usually, the density value can take a value of 256 steps from 0 to 255. In this case, the value of q takes a value from 0 to 255, and the value of N is 256.

  The comparison unit 27 compares the calculated similarity S with a predetermined threshold T. If the similarity S is greater than the threshold value T, it is determined that the obstacle Bs1 exists on the track Rd. If the similarity S is less than or equal to the threshold T, it is determined that the obstacle Bs1 does not exist on the track Rd. When it is determined that the obstacle Bs1 exists on the road Rd, a determination signal is sent to the warning output unit 8 and the vehicle speed control unit 9. When it is determined that the obstacle Bs1 does not exist on the track Rd, the determination signal is not output.

  Next, the operation of obstacle detection in the first embodiment will be described with reference to FIG. FIG. 6 is a flowchart showing a processing procedure for obstacle detection.

  An image is taken by the image sensor 3 provided on the front surface of the vehicle 1 (step S01). Next, the obstacle detection unit 15 detects the obstacle Bs1 in the photographed image Dp1 (step S02). The obstacle area setting unit 19 sets a circumscribed rectangular area surrounding the detected obstacle Bs1 as the obstacle area Ba1 (step S03). The ground area setting unit 21 sets an area having a predetermined size below the obstacle area Ba1 as the obstacle ground area Bg1 (step S04). A predetermined area at the lower center of the image DP1 is defined as a road ground area Cg1, and the histogram similarity S between the density value of the pixel of the road ground area Cg1 and the density value of the pixel of the obstacle ground area Ba1 is expressed as follows. Calculate (step S05). Based on the calculated similarity S, an obstacle determination is made as to whether or not the detected obstacle Bs1 is on the road of the vehicle 1 (step S06). If the similarity S is greater than the threshold T, it is determined that the obstacle Bs1 is on the road Rd, and if the similarity S is equal to or less than the threshold T, it is determined that the obstacle Bs1 is not on the road Rd.

  In response to the determination that the obstacle Bs1 exists on the road Rd, the warning output unit 8 including a speaker or the like issues a warning to the driver and the surroundings. Further, in response to the determination that the obstacle Bs1 exists on the travel path Rd, the vehicle speed control unit 9 performs the deceleration control and the brake braking, the drive motor 11 and the wheel rotation are braked, and the vehicle 1 is decelerated or stopped. To do.

  As described above, according to the first embodiment, the similarity S between the detected density value of the ground area Bg1 of the obstacle Bs1 and the density value of the runway ground area Cg1, that is, the runway Rd area is calculated. Since it is determined whether or not the obstacle Bs1 detected on the basis of the similarity S exists on the road Rd, it is possible to accurately determine the obstacle present on the road. Moreover, since the deceleration control of the vehicle 1 is performed only when it is determined that an obstacle exists on the road, the traveling efficiency of the vehicle 1 can be increased.

  Next, an obstacle detection apparatus according to the second embodiment will be described with reference to FIGS. 7 and 8. FIG. 7 is a front view of the vehicle in the second embodiment. FIG. 8 is a block diagram illustrating the configuration of the obstacle detection apparatus according to the second embodiment. In the second embodiment, the parts indicated by the same reference numerals as those in the first embodiment have the same configuration as in the first embodiment, and thus the description thereof is omitted here. The configurations of the vehicle and the obstacle detection device other than those described below are the same as those in the first embodiment.

  A feature of the second embodiment is that the accuracy of determining whether or not the detected obstacle exists on the road is further improved. In the first embodiment, whether or not an obstacle exists on the road is determined based on the image information. In the second embodiment, whether or not an obstacle exists on the road by adding distance information to the image information. Determine whether or not.

  The obstacle detection apparatus 6 according to the second embodiment includes a stereo camera 4, a memory 10, a parallax image creation unit 12, a distance image calculation unit 13, an obstacle detection unit 14, an obstacle distance determination unit 16, and an obstacle. The object ground area setting unit 18, the runway ground area setting unit 26, and the obstacle determination unit 24 are included. The parallax image creation unit 12, the distance image calculation unit 13, the obstacle detection unit 14, the obstacle distance determination unit 16, the obstacle ground region setting unit 18, the runway ground region setting unit 26, and the obstacle determination unit 24 It consists of a processor or FPGA.

  The stereo camera 4 is composed of two image sensors 4 a and 4 b provided in the center of the front surface of the vehicle 1. The stereo camera 4 includes a left image sensor 4a and a right image sensor 4b. Either the left image sensor 4a or the right image sensor 4b is set as a reference camera. The stereo camera 3 is not limited to two image sensors, and three or more image sensors may be used.

  The image sensors 4a and 4b of the stereo camera 4 are installed under a predetermined geometric condition. In this embodiment, the image sensors 4a and 4b are installed at a constant distance in the horizontal direction. That is, the image sensors 4a and 4b are parallel stereo. The image sensors 4a and 4b are general visible light sensors such as a CCD and a CMOS. Each of the image sensors 4a and 4b has been subjected to camera calibration in advance so that the positions of the rows of the captured images are matched, that is, the epipolar lines are matched. Note that an image photographed by the left image sensor 4a is a left image, and an image photographed by the right image sensor 4b is a right image.

  The memory 10 is an image buffer that temporarily stores images taken by the image sensors of the stereo camera 4. The stored images of the image sensors 4a and 4b are sequentially output to the parallax image creation unit 12.

  The parallax image creation unit 12 creates a parallax image by stereo matching from the input left image and right image. Examples of the stereo matching method include a method using SAD or a method using a Hamming distance.

  The distance image calculation unit 13 calculates a distance in real space to the object displayed by each pixel based on the parallax image. As a method for calculating the distance from the parallax image, a conventionally used parallel stereo method is used. That is, by using the distance between the two imaging sensors 4a and 4b and the focal length and the parallax image of the two imaging sensors 4a and 4b, distance information is attached to each pixel of the left image that is the reference image. A distance image can be calculated.

If a point is detected in a predetermined space in the real space, the obstacle detection unit 14 determines that some obstacle exists. This will be described with reference to FIG. FIG. 9 is an explanatory diagram showing obstacle detection. In the real space XYZ in the traveling direction of the vehicle 1, a point group PG that is at substantially the same distance from the vehicle 1 is detected within the space Sp of a predetermined detection range. By detecting these point groups PG as obstacles and extracting the maximum and minimum values of the uv coordinates of the corresponding obstacles in the reference image, an obstacle region Ba2 that surrounds the obstacles with a circumscribed rectangle is set. . Further, to detect the distance L ₀ to the point group PG from the distance image.

The obstacle distance determining unit 16 compares the predetermined threshold with the distance L ₀ to the detected obstacle, whether to implement the speed control of the vehicle 1 according to the distance L ₀ to the obstacle Determine whether. If the distance L ₀ to the obstacle is smaller than the first threshold L ₁ , the vehicle 1 is determined to be decelerated and the vehicle speed control unit 9 is instructed to decelerate. Threshold L _1, the distance the distance also trees and structures the vehicle 1 runway aside in the curve is secured at least is preferred. That is, if there is no obstacle on the road, the range from the vehicle 1 of the distance L _1, the distance of the obstacle is not detected.

Further, the vehicle speed when the automatic travel is set to have a sufficient margin braking distance relative to the distance L _1. In addition, when the vehicle 1 is equipped with GPS and can recognize the position of the own vehicle, the first threshold L ₁ may be prepared for each course on which the vehicle 1 travels.

In addition to the first threshold value L _1, the obstacle distance determining section 16 compares the distance L ₀ between ₂ both the second threshold value L to the detected obstacle. The second threshold value _{L 2} larger than the first threshold value _{L 1.} If the distance L ₀ to the obstacle is smaller than the second threshold value L ₂ at the first threshold value L ₁ or more, processing of determining the similarity of textures of the obstacle ground area and runway ground area is performed. When the distance L ₀ to the obstacle is greater than or equal to the second threshold value L ₂ , the vehicle 1 is not far away from the obstacle, so that the vehicle speed control of the vehicle 1 is not performed.

  The runway ground area setting unit 18 sets the ground area of the detected obstacle as in the first embodiment when an instruction to set the obstacle ground area is input from the obstacle distance determination unit 16. Please refer to FIG. FIG. 11 is an explanatory diagram showing an obstacle ground area and a runway ground area. The obstacle region setting unit 20 extracts the maximum value and the minimum value of the uv coordinates of the obstacle Bs2 on the image Dp2 corresponding to the detected point group PG, thereby surrounding the obstacle Bs2 with a circumscribed rectangle. Ba2 is set. Similarly, when a tree is detected in the space Sp of the detection range, the maximum value and the minimum value of the uv coordinates of the obstacle Bs3 on the image Dp2 corresponding to the detected point group are extracted, thereby An obstacle region Ba3 that surrounds the object Bs3 with a circumscribed rectangle is set.

  The ground area setting unit 22 sets the lower pixel area immediately below the set obstacle areas Ba2 and Ba3 as the obstacle ground areas Bg2 and Bg3. The width in the u-axis direction of the obstacle ground areas Bg2 and Bg3 is set to the same size as the width of the obstacle areas Ba2 and Ba3, respectively. The height in the v-axis direction of the obstacle ground areas Bg2 and Bg3 is set to a predetermined size. The predetermined size is, for example, a size corresponding to about 50 cm in real space. That is, the obstacle ground areas Bg2 and Bg3 are pixel areas obtained by projecting an area having a size of about 50 cm in real space below the obstacle areas Ba2 and Ba3 onto the image Dp2. In FIG. 9, corresponding regions in the real space of the obstacle ground region Bg2 and the running road ground region Cg2 are shown.

  The runway ground area setting unit 26 sets an area having the same area as the obstacle ground area in the real space as a runway ground area on the image. The runway ground area setting unit 26 includes an area calculation unit 28 and a ground area setting unit 29.

  The area calculation unit 28 calculates the area in the real space of the obstacle ground areas Bg2 and Bg3 set on the image Dp2 based on the distance image.

  The ground area setting unit 29 sets an area having the same size as the area calculated by the area calculating unit 28, the runway ground area Cg2 positioned in the center in the u-axis direction and in the lower part in the v-axis direction in the image Dp2. To do. That is, the runway ground area Cg2 is set on the photographed image Dp2 so that the area on the ground in the real space of the road ground area Cg2 is substantially the same as the area on the ground in the real space of the obstacle ground area Bg2. To do. In the captured image Dp2, an object far from the vehicle 1 appears smaller than an object in the vicinity of the vehicle 1, so the runway ground area Cg2 is larger than the obstacle ground area Bg2. Similarly, a road ground area Cg3 (not shown) in front of the vehicle having an area substantially the same as the area on the ground in the real space of the obstacle ground area Bg3 is set on the photographed image Dp2. Data of the density value of each pixel of each set obstacle ground area is output to the obstacle determination unit 23.

  The obstacle determination unit 24 includes a normalization unit 30, a similarity calculation unit 31 that calculates the degree of texture similarity between the runway ground region and the obstacle ground region, and a comparison unit 27.

  The normalization unit 30 normalizes the histograms of the density values of the obstacle ground areas Bg2, Bg3 and the runway ground areas Cg2, Cg3. This histogram is quantized. That is, the density values of the obstacle ground areas Bg2 and Bg3 and the density values of the runway ground areas Cg2 and Cg3 are normalized with respect to the quantized histogram by reducing gradation. Normally, the density value can take 256 levels from 0 to 255, and is quantized into, for example, 8 levels. Thereby, since the data amount of calculation of similarity can be suppressed, similarity can be determined at high speed. Further, although the number of pixels included in the runway ground areas Cg2 and Cg3 is different from the number of pixels included in the obstacle ground areas Bg2 and Bg3, the sum of the quantized histograms becomes 1 by normalization. By doing so, the two regions can be handled equally, and the accuracy of the similarity can be improved. FIG. 12 is an explanatory diagram showing a density histogram of the obstacle ground area Bg2, and FIG. 13 is an explanatory diagram showing a density histogram of the road ground area Cg2. Since both the vehicle 1 and the obstacle Bs2 are on, for example, a paved runway, both histograms are similar. FIG. 14 is an explanatory diagram showing a density histogram of the obstacle ground area Bg3. Since the obstacle Bs3 is on the grass, for example, outside the road, the density histogram of the obstacle ground area Bg3 tends to be different from the density histogram of the road ground area Cg2 on the road.

  The similarity calculation unit 31 calculates the similarity S ′ between the normalized histogram of the density value of the obstacle ground area Bg2 and the normalized histogram of the density value of the runway ground area Cg2, thereby obtaining the obstacle ground. The similarity S ′ between the region Bg2 and the runway Rd is calculated. For the calculation of the similarity S ′, a Bhattacharyya coefficient is used as in the first embodiment. In Example 2, the value of q in Equation 1 takes a value of 0 to 7, and the value of N is 8, which is the quantization number of the density value. Also, a similarity S ″ between the normalized histogram of the density value of the obstacle ground area Bg3 and the normalized histogram of the density value of the runway ground area Cg3 is calculated.

Comparing unit 27 compares the threshold value T ₂ which predetermined the calculated degree of similarity S '. If the similarity S 'is larger than the threshold value T _2, it determines the obstacle Bs2 is present on the runway Rd. If the similarity S 'is the threshold T ₂ or less, it is determined that the obstacle Bs2 is not on track Rd. When it is determined that the obstacle Bs2 exists on the road Rd, a determination signal is sent to the warning output unit 8 and the vehicle speed control unit 9. When it is determined that the obstacle Bs2 does not exist on the runway Rd, the determination signal is not output.

Is greater than the threshold value T ₂ the similarity S of the ground texture 'is predetermined, the detected obstacle Bs2 since it is determined to be on the track Rd, implementing the deceleration control. If the similarity S of the ground texture 'threshold T ₂ less than or equal to a predetermined detected obstacle Bs2 since it is determined that there is no on track Rd, deceleration control is not performed. For Figure 11, the similarity S of the obstacle ground region Bg2 and runway ground region Cg2 'is determined to be greater than the threshold value T _2, the obstacle Bs2 is determined to be present on the runway Rd. Further, the similarity S of the obstacle ground region Bg3 and runway ground area Cg3 '' is determined to be the threshold value T ₂ or less, the obstacle Bs3 is determined not to exist on the runway Rd.

  Next, an obstacle detection operation according to the second embodiment will be described with reference to FIG. FIG. 15 is a flowchart showing a processing procedure for obstacle detection.

A plurality of images ahead of the vehicle are taken by the stereo camera 4 provided on the front surface of the vehicle 1 (step S11). Next, stereo matching is performed based on the taken left image and right image to create a parallax image (step S12). Based on the created parallax image, a distance image obtained by adding distance information in real space to each pixel of the image Dp2 of the reference camera is calculated (step S13). A point group PG having a constant distance within the predetermined detection space Sp is detected as an obstacle (step S14). The distance L ₀ to the detected obstacle is compared with the first threshold value L ₁ (step S15).

If the distance L ₀ is less than the first threshold value L ₁ (Yes in step S15), the vehicle speed control unit 9 controls the deceleration because the distance between the vehicle 1 and the obstacle is close (step S16). . If the distance _{L 0} is the first threshold value _{L 1} or more (No in step S15), and further, compared with the distance _{L 0} to the detected obstacle and a second threshold value _{L 2} (step S21).

If the distance L ₀ to the obstacle is equal to or greater than the second threshold L ₂ (No in step S21), the vehicle 1 and the obstacle are sufficiently separated from each other, so that the vehicle speed control of the vehicle 1 is not performed and step S11 is performed again. Return to the process.

If the distance L ₀ to the obstacle is smaller than the second threshold value L ₂ (Yes in step S21), and proceeds to determining whether an obstacle is present on the track. A circumscribed rectangular area surrounding the obstacle Bs2 corresponding to the point group PG detected in the reference image Dp2 is set as the obstacle area Ba2 (step S22). An area having a predetermined size below the obstacle area Ba2 is set as the obstacle ground area Bg2 (step S23). The area of the set obstacle ground area Bg2 on the real space (XYZ space) is calculated (step S24). An area having the same area as the real area of the obstacle ground area Bg2 thus calculated is set as a runway ground area Cg2 at the center in the u-axis direction and the lower part in the v-axis direction on the image Dp2 (uv plane) ( Step S25).

Next, the histogram of the density values of the pixels in the obstacle ground area Bg2 and the histogram of the density values of the pixels in the road ground area Cg2 are normalized (step S26). The similarity S ′ of each normalized histogram is calculated (step S27). Calculated similarity S 'is compared with a threshold value T ₂ which is determined in advance (step S28). If the similarity S ′ is greater than the threshold T ₂ (Yes in step S28), the obstacle ground area Bg2 and the runway ground area Cg2 are similar in texture, so it is determined that an obstacle exists on the runway. Is done. Therefore, the vehicle speed control unit 9 performs deceleration control and brake braking (step S16). Thereby, rotation of drive motor 11 and a wheel is braked, and vehicle 1 decelerates or stops. Further, in response to the determination that the obstacle Bs2 exists on the runway Rd, a warning is issued to the driver and the surroundings from the warning output unit 8 constituted by a speaker or the like.

If the similarity S 'is the threshold T ₂ or less (No in step S28), since each of the texture of the obstacle ground region Bg2 and runway ground region Cg2 are not similar, it determines that the obstacle is not on track Is done. Therefore, the vehicle speed control of the vehicle 1 is not performed, and the process returns to step S11 again. Further, when a plurality of obstacles are detected, that is, when the obstacle Bs3 is detected in addition to the obstacle Bs2, each obstacle is detected on the road Rd according to the above steps. It is determined whether or not it exists.

  Thus, according to Example 2, since the quantized histogram of the density value of the ground area Bg2 of the detected obstacle Bs2 and the quantized histogram of the density value of the runway ground area Cg2 are normalized, The amount of data for calculating the similarity can be suppressed, and the calculation of the similarity S ′ can be performed at high speed. Furthermore, since two regions having different numbers of pixels can be handled equally, the accuracy of similarity can be improved. In addition, since the area in the real space of the obstacle ground area Bg2 and the runway ground area Cg2 is the same, the tendency of the histogram of the density value can be detected without bias, and the accuracy of the similarity S 'can be improved.

  Furthermore, since it is possible to determine whether or not the detected obstacle Bs2 exists on the road Rd based on the similarity S ′, it is possible to accurately determine the obstacle present on the road. In addition, since the deceleration control of the vehicle 1 is performed only when it is determined that the obstacle Bs2 exists on the travel path Rd, the traveling efficiency of the vehicle 1 can be increased.

  The present invention is not limited to the above embodiment, and can be modified as follows.

  (1) In the second embodiment, the stereo camera 4 is used to create a parallax image, and distance information is acquired by performing stereo matching. For example, the distance information may be acquired by a radar scanner. In this case, the image acquired by the monocular image sensor needs to be associated with the distance information acquired by the radar scanner in advance.

  (2) In the above embodiment, the obstacle detection device 5 or 6 is provided in the vehicle, but is not limited thereto. In addition, for example, it may be adopted in a vision system for a robot that travels autonomously or a support system for a visually impaired person.

DESCRIPTION OF SYMBOLS 1 ... Vehicle 3 ... Image sensor 4a ... Left image sensor 4b ... Right image sensor 5, 6 ... Obstacle detection apparatus 12 ... Parallax image creation part 13 ... Distance image calculating part 14, 15 ... Obstacle detection part 16 ... Obstacle distance Determination unit 17, 18 ... Obstacle ground area setting unit 25, 31 ... Similarity calculation unit 26 ... Runway ground area setting unit
27: Comparison unit 28 ... Area calculation unit 29 ... Ground area setting unit 30 ... Normalization unit

Claims (9)

When the characteristics of the image information are different between the ground area of the runway and the ground area outside the runway,
An image sensor for taking a forward image;
An obstacle detection unit for detecting obstacles in the image;
An obstacle ground area setting unit for setting a ground area of the obstacle in the image;
An obstacle detection apparatus comprising: a similarity calculation unit configured to calculate a similarity of the characteristic between the ground area of the obstacle and the ground area of the runway in the image. The obstacle detection device according to claim 1,
A distance information detection unit that detects distance information with a front object,
The obstacle detection device detects an obstacle in the image based on the distance information. In the obstacle detection device according to claim 1 or 2,
A plurality of the image sensors in a parallel stereo positional relationship,
An obstacle detection device including a parallax image creation unit that creates a parallax image based on an image of each image sensor. In the obstacle detection device according to claim 2 or 3,
An obstacle detection device comprising: a road ground area setting unit that sets, in the image, a ground area of a road having the same area as the area of the obstacle ground area in real space. The obstacle detection device according to claim 4, wherein the runway ground region setting unit includes:
An area calculation unit for calculating an area of the ground area of the obstacle in real space;
An obstacle detection apparatus comprising: a ground area setting unit configured to set a ground area of a running road having the same size in real space as the calculated area. In the obstacle detection device according to any one of claims 1 to 5,
An obstacle detection apparatus comprising: a normalization unit that normalizes a density value of an obstacle ground area and a density value of a ground area of a runway with respect to a histogram obtained by reducing gradation and quantizing. In the obstacle detection device according to any one of claims 1 to 6,
An obstacle detection device comprising: a comparison unit that determines whether an obstacle is present on the road by comparing the similarity with a predetermined threshold value.   A vehicle comprising the obstacle detection device according to any one of claims 1 to 7. The vehicle according to claim 8, wherein
It is determined whether or not vehicle deceleration control is to be performed by comparing the distance to the obstacle and the first distance, and the distance to the obstacle and the second distance longer than the first distance are compared. A vehicle including an obstacle distance determination unit that determines whether to calculate the similarity between the ground area of the obstacle and the ground area of the vehicle.

Images