JPH03282709A - Environment recognition device for mobile vehicle - Google Patents

Abstract

PURPOSE:To accelerate recognition whether or not a body in accordance with a mobile body candidate point is traveling by narrowing down the mobile body candidate point based on an edge extracted from prescribed two screens. CONSTITUTION:Travel path area extracting means 1, 2 which extract a travel path area by binarizing an input image, and an edge extracting means 3 which extracts the edge of the image in a horizontal direction advancing from the lower side to the upper side in an extracted travel path area are provided. Also, a mobile body candidate point extracting means 7 which narrows down the mobile body candidate point based on the size of an extracted edge, and mobile body recognizing means 5, 6, and 7 which recognize the fact that the mobile candidate point is a mobile body based on the amount of travel between the prescribed two pictures and that of its own vehicle are provided. In such a way, the mobile body candidate point can be narrowed down by performing edge extraction by applying image processing. Thereby, it is possible to recognize the mobile body before its own vehicle from the amount of travel between the prescribed two pictures and that of its own vehicle, which enables the recognition to be performed with few amount of calculation and in a short time.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an environment recognition device for a moving vehicle that recognizes a moving object through image processing.

(Prior Art) Conventionally, when recognizing a moving object using image processing, two consecutive screens are compared to confirm the same object, and it is determined whether or not it is a moving object. In this case, various characteristic amounts were extracted from the screen, calculated, and the results of the calculations were compared and verified to determine whether or not the objects were the same.

(Problem to be solved by the invention) However, in the conventional example described above, in order to determine whether or not the objects are the same, various feature quantities extracted from the screen are compared and verified by calculation, which requires a huge amount of calculation. The drawback is that it requires a large amount of time and time.

(Means for Solving the Problems) The present invention has been made for the purpose of solving the above-mentioned problems, and includes the following configuration as one means for solving the above-mentioned problems.

That is, a driving road area extracting means binarizes an input image to extract a driving road area, and a horizontal edge of the image is extracted from the bottom to the top within the driving road area extracted by the driving road area extracting means. an edge extraction means for performing extraction; a moving object candidate point extraction means for narrowing down moving object candidate points based on the size of the edge extracted by the edge extraction means;
The moving object recognition means recognizes that the moving object candidate point is a moving object based on the amount of movement of the moving object candidate point and the amount of movement of the own vehicle between two predetermined screens.

(Function) In the above configuration, edge extraction is performed by image processing to narrow down the moving object candidate points, and the movement in front of the own vehicle is determined based on the amount of movement of the moving object candidate points and the amount of movement of the own vehicle between two predetermined screens. Can recognize objects.

(Embodiment) Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram of an environment recognition device for a moving vehicle, which is an embodiment of the present invention.

In FIG. 1, an environment recognition device for a moving vehicle (hereinafter referred to as recognition device) inputs an image in front of the vehicle from an image input unit 1 consisting of a video camera, etc., and converts the input image into an image binarization unit 2. Binarize. Then, the edge extraction section 3 extracts edges from the binarized image by a method described later, and stores the image after edge extraction in the image storage section 4.

The sensor unit 5 includes a vehicle speed sensor 5a and a direction sensor 5b.The vehicle speed sensor 5a detects the moving speed of the own vehicle, and the direction sensor 5b detects the moving direction of the own vehicle. These detection results are then sent to the movement amount calculating section 6, where the movement amount of the own vehicle is calculated.

The main calculation unit 7 inputs the image after edge extraction stored in the image storage unit 4 and extracts moving object candidate points. The main calculation unit 7 extracts moving object candidate points for two consecutive screens, as will be described later, and calculates the amount of movement of the moving object candidate points between the two screens. Then, based on the calculation result of the movement amount of this moving object candidate point and the movement amount of the own troops from the movement amount calculating section 6, the moving object in front of the own vehicle is recognized, and the result is displayed on the display section 8 or displayed. Display audibly.

Hereinafter, a method for recognizing a moving object in the recognition apparatus of this embodiment will be described in detail.

FIGS. 2A and 2B are diagrams for explaining a method of extracting white lines on a road from an image taken in front of the vehicle. FIG. 2(A) is an image taken by the image input section 1 before image processing, and FIG. 2(B) is the image of FIG. 2(A) taken by the image binarization section 2 into a binary value. This is the image after it has been converted.

The threshold value T for binarizing the image in the image binarizing unit 2 is based on the area A at the lower center of the screen in FIG. 2(A) as a road area.
Average value t obtained by binarizing area A. By finding , it can be expressed by the following formula.

T=to +α... (1)
However, α=f(to).

When the image in Fig. 2 (A) is binarized using the threshold value T determined by the above formula (1), the white line on the road can be extracted, so the area indicated by diagonal lines in Fig. 2 (B) Find S.

Therefore, the edge extracting section 3 extracts edges in the horizontal direction of the image within the region S obtained by the above-described method using the mask shown in FIG. That is, as shown in FIG. 4(A), an object 21 that casts a shadow 21a within the region
If there is an object 20 that drops the object 20, when edge extraction is performed from the bottom of the screen upwards, the upper part is darker than the lower part, that is, the part that changes from bright to dark is an edge (see Figure 4 (described later)).
B) is extracted as 21b and 20b). The actual size of the object can be determined by measuring the width of the extracted portion, that is, the length of the edges, so those that exceed a certain value are narrowed down and saved as moving object candidate points, while other Remove things as noise. As a result, the image shown in FIG. 4(B) is obtained, which is stored in the image storage unit 4 as an image after edge extraction and moving object candidate point narrowing down.

Next, after a predetermined period of time (denoted as Δt) has elapsed after photographing the image shown in FIG. ), extract the edges in the horizontal direction of the image using the same method as described above,
This is stored in the image storage section 4 as an image after edge extraction. As a result, edges 22b and 23b shown in FIG. 4(D) can be extracted as moving object candidate points on that screen. Incidentally, here, for convenience, white lines on the road are written in the image after edge extraction.

Therefore, a method for determining whether the object corresponding to the moving object candidate point obtained by the above-described image processing is actually moving or standing still will be described below.

FIG. 5 is a diagram showing the positional relationship between two types of objects moving on a plane and an object stationary with respect to the plane. In the figure, a television camera 31 is fixed to the own vehicle 30 so as to be able to photograph a predetermined area in front of the traveling direction (arrow in the figure), including a moving object 21 and a stationary object 20 located in front of the own vehicle. There is.

Now, if the own troops 30 and the forward moving object 21 are at the positions shown by dotted lines on the plane 40 at time t1, and at the positions shown by solid lines at time t2 after the elapse of time Δ, the sensor unit 5 vehicle speed sensor 5a and direction sensor 5b
constantly monitors the speed and direction of your vehicle,
The movement amount calculation unit 6 calculates the movement amount Δ°C of the own vehicle from the movement speed, movement direction, and time Δt. On the other hand, the stationary object 20 is located at a fixed location on a plane regardless of the passage of time. Therefore, the image taken at time t1 is shown in Figure 4 (
A), and the image taken at time t2 is shown in FIG. 4(C), the images after horizontal edge extraction as described above are shown in FIG. 4(B) and FIG. 4(D), respectively.

Here, it is assumed that two images are taken from the vehicle moving at a certain speed with a time interval Δt long enough to recognize the movement of the photographed stationary object on the screen.

Since these images after edge extraction are stored in the image storage section 4, the main processing section 7 processes them at time 1. The images obtained in step 12 are compared to determine whether there is a moving candidate point common to both images. That is, the main calculation unit 7 determines 20b, 21b, 22b, and 23b from the images shown in FIG. 4(B) and FIG. 4(D) as movement candidate points common to both images, but among them, It can be seen that the positions of edges 20b and 23b have moved between the two screens, and the positions of edges 21b and 22b have not changed on the screen. Therefore, the main calculation unit 7 further calculates the distance between the movement amount of the own vehicle calculated by the movement amount calculation unit 6 and the edges 20b and 23.
Perform calculations to see if the amount of movement of b on the screen matches,
If the two match, it is determined that the edges 20b and 23b are edges extracted from the area S, that is, the shadow of the same stationary object on the road. Furthermore, the main calculation unit 7 determines that edges 21b and 22b, which have been determined not to be moving on the screen, are edges extracted from a moving object in front of the vehicle. Therefore, it can be seen that FIG. 4(A)(7) 20 and FIG. 4(C)(7) 23 are the same stationary object, and 21 and 22 are the same moving object.

Furthermore, by comparing FIG. 4(B) and FIG. 4(D), the amount of movement of the edges 20b and 23b on the screen and the amount of movement calculation unit 6 are calculated.
From the amount of movement of the vehicle calculated in Error 1, it is clear that edges 20b and 23b are the edges of the shadow of a stationary object on the road, and edges 21b and 22b are the edges of the shadow of a moving object in front of the vehicle. If the movement on the screen is such that it can be determined that
, 22b on the screen and the amount of movement of the own vehicle calculated by the movement amount calculating section 6, the speed of the moving object in front of the own vehicle can be calculated.

That is, when the own vehicle is moving at a constant speed, if the edge 22b in FIG. 4(D) is above 21b in FIG. It means that it is moving at a faster speed than the forward moving object 2.
1 means that the speed is slower than the own vehicle 30. Needless to say, the edge 21b in Figure 4 (B) and the edge 21b in Figure 4 (D)
22b are at the same position on the screen, it means that the forward moving object 21 is moving at the same speed as the own vehicle.

As explained above, according to this embodiment, points of change from light to dark in the images are extracted as edges for each of two consecutively photographed images, and the edges are narrowed down. By performing calculations based on the amount of movement of the edges and the amount of movement of your own troops, you can judge whether objects in front of your vehicle are moving or not, making it possible to recognize moving objects on the road quickly. It also has the effect of making it easier.

In the above embodiment, edge extraction is performed based on the brightness of the binarized image to search for candidate points of a moving object, but edge extraction may be performed using a color image.

In other words, the shadow directly below the vehicle clearly appears even without sunlight, and furthermore, the shadow does not have the color component of the road, and is RG
B By analyzing the color components of the shadow in the image, it is easy to distinguish between the shadow directly below the vehicle and the shadow with color components similar to those of the road below. This enables more accurate candidate point extraction of moving objects.

(Effects of the Invention) As explained above, according to the present invention, by narrowing down moving object candidate points based on edges extracted from two predetermined screens, the object corresponding to the candidate points is moved. This has the effect of speeding up the recognition of whether or not there is a person.

[Brief explanation of drawings]

Fig. 1 is a block diagram of an environment recognition device for a moving vehicle, which is an embodiment of the present invention; Fig. 2(A) is a diagram showing an original image captured by an image input unit; Fig. 2(B) Figure 3 shows the image after it has been binarized by the image binarization unit, and Figure 3 shows a mask for extracting edges in the horizontal direction of the image in the area S shown in Figure 2 (B). Figure 4 (A) is a diagram showing an image taken at a certain time t1, Figure 4 (B) is an image after performing edge extraction in the horizontal direction of the image in Figure 4 (A). Figure 4(C) is a diagram showing an image taken after Δ has elapsed from time tl. Figure 4(D) is a diagram showing the image taken after horizontal edge extraction of the image in Figure 4(C). The image shown in FIG. 5 is a diagram showing the positional relationship between two types of objects moving on a plane and an object stationary on the plane. In the figure, 1... image input section, 2... image binarization section,
5 3... Edge extraction section, 4... Image storage section, 5...
Sensor unit, 6... Movement amount calculation unit, 7... Main calculation unit,
8... Display unit, 20... Stationary object, 21... Moving object, 30... Self-vehicle, 31... Television camera.

Claims (1)

[Scope of Claims] An environment recognition device for a moving vehicle including an image input means for recognizing the outside world, comprising: a travel road region extracting means for extracting a travel road region by binarizing the input image; edge extraction means for extracting edges in the horizontal direction of an image from below to above within the travel road area extracted by the area extraction means; and a moving object based on the size of the edge extracted by the edge extraction means. A moving object candidate point extraction means that narrows down the candidate points, and a moving object candidate point that recognizes that the moving object candidate point is a moving object based on the amount of movement of the moving object candidate point between two predetermined screens and the amount of movement of the own vehicle. An environment recognition device for a moving vehicle, characterized in that it has a moving object recognition means.