JPH03118615A - Environment recognizor for mobile vehicle - Google Patents

Abstract

PURPOSE:To detect in real time the end of a road by inputting the traveling direction of a mobile vehicle as the substantial main scanning direction of a raster scan system and the front image of the vehicle as the serial picture signals respectively. CONSTITUTION:A camera 1 scans data by a raster scan system and takes out a data train. Thus the main scanning direction is coincident with the traveling direction of a mobile vehicle. A bright and stable density signal is obtained within a road area from the multi-level picture signal undergone the main scan in the traveling direction of the vehicle. At the same time, a density signal having high density and a large fluctuation is obtained in an area out of the road. A comparator 3 compares the density value d0 of the head picture element latched by an initial value memory 2 with the density value dx of the subsequent picture signal. Then din showing an area within the road is outputted to a frame memory 6 when ¦d0-dx¦<epsilon is satisfied and a counter energizing signal (= 1) is outputted to a counter 4. Meanwhile dout showing an area out of the road is outputted together with a counter energizing signal (= 0) when ¦d0-dx¦>=epsilon is satisfied respectively. The output of the counter 4 shows the road end address of each line of a picture element.

Description

[Detailed Description of the Invention] (Industrial Application Field) The present invention relates to an environment recognition device for a mobile vehicle that autonomously controls its driving. Regarding speeding up.

(Prior Art) In recent years, various mobile vehicles have been proposed that autonomously avoid obstacles while traveling.

In such autonomous driving control, for example,
As in No. 4-26913, in order to recognize road knots from the environment, an image of the outside world is input using an image input means such as a camera, and image processing is performed on this image to recognize the outside world environment. Common.

FIG. 3 is a diagram illustrating a method of processing for tangled end recognition according to this conventional example. The conventional method shown in FIG. 3A is to read the outside world with a camera, temporarily store the outside world image in a frame memory, and perform arithmetic processing for road edge detection on this outside world image. This detection process utilizes the homogeneity of the data within the road, and detects a boundary between an area where the data is homogeneous and an area where the data is non-uniform.

(Problem to be Solved by the Invention) FIGS. 3B and 3C explain why a frame memory is necessary in the conventional method shown in FIG. 3A. FIG. 3B schematically depicts a state in which an image including a road is stored in the frame memory. It is stored in frame memory (the direction is raster scan from left to right on the page).

The algorithm for detecting tangled ends is as shown in Figure 3C.
It is based on the empirical fact that the image data of the on-road area in the input image is more homogeneous than that of the off-road area. This is because there are basically no obstacles to driving on a road, and the edges are usually straight or curved smoothly.

Therefore, in order to increase the speed of road edge detection, road edge detection scanning is performed from the bottom of the image to the top, and areas where data homogeneity is lost are recognized as tied edges.

For a moving vehicle, recognizing an object outside the road does not have much meaning, so if a road edge is detected, there is no need to perform road edge detection processing for the area outside the road. This is the reason why road edge detection scanning is performed from the bottom of the image to the top.

If the road edge detection scanning direction is made to match the image input direction, that is, if the road edge detection process is performed from left to right in the paper, the road edge detection process will be started from the area outside the road, and as described above, Since the image data in this area usually changes in a complicated manner, this detection process involves a lot of unnecessary processing due to the complicated changes in the image data. In this case, the tangled edge cannot be reached without passing through the out-of-road area images, which require many complex calculations due to the complex changes.

For this reason, road edge detection scanning is performed from the bottom of the image to the top.

However, in order to make this process possible, as is clear from FIG. 3C, it is a prerequisite that an image is stored in the frame memory before starting the tied end search process. This is why a frame memory is required in the prior art, as shown in Figure 3A.

Therefore, in the prior art, all processing, particularly road edge detection, is not started until the image has been stored in the frame memory, and this has been one of the causes of low efficiency in conventional environment recognition.

The present invention has been made in view of the problems of the prior art, and its purpose is to provide an environment recognition device for a moving vehicle that enables high-speed environment recognition in order to be able to perform road edge detection at high speed. This is what we provide.

(Means and operations for achieving the object) The configuration of the present invention for achieving the above object is to provide an environment recognition device for a moving vehicle that inputs an image of the external environment and recognizes the environment from this image. An image input means that uses a raster scan power type in the substantial main scanning direction and inputs an image in front of the vehicle as a serial image signal, and a road edge that performs road edge detection by density change detection with respect to the serial image signal. The present invention is characterized by comprising a detection means.

That is, since the scanning direction in which the image is read and the direction in which the processing for detecting the tangled ends match, the image reading and the tangled end recognition processing can be performed in parallel.

(Embodiment) Hereinafter, a preferred embodiment of an environment recognition device to which the present invention is applied will be described with reference to the accompanying drawings.

FIG. 1A is a block diagram showing the hardware configuration of this embodiment. The feature of this configuration is that the road edge detection device 5 detects a tangled edge in real time with respect to the human input data string of the image signal being read by the CCD camera 1. When a connected end is detected, its address is output. In addition, the detection device 5 stores data dlR in the frame memory 6 indicating that pixels that are determined to be located within the road area, and data dlR that indicates that pixels that are determined to be located outside the road area. Write data d out.

FIG. 1B shows how the camera 1 is installed. That is, the camera is rotated 90 degrees in the direction of the half-hour hand around the optical axis and fixed while facing the direction of travel of the vehicle. Since the camera 1 uses a raster scan method to scan the data in the built-in image sensor and extract a data string, the scanning in the main scanning direction coincides with the direction in which the vehicle is traveling. That is, as shown in FIG. 1B, main scanning is performed from the bottom to the top of the page, and sub-scanning is performed from the left to the right of the page.

FIG. 2 explains the operation of the road edge detection device 5. In Figure 2 (
In a), the data string input to the detection device 5 is:
The output signal from camera 1 is shown.

As shown in (a), the input data string (n o , n r
.

nz, ns) are multivalued image signals scanned in the main direction in the traveling direction. In the area inside the road, a relatively bright (stable density signal) is obtained.On the other hand, in the area outside the road, a density signal with relatively high density but with large fluctuations is obtained.Such input data Column (no, n+,
no + ns) is input to the initial value memory 2 and the comparator 3 in synchronization with the horizontal synchronization signal (HSYNC).

The initial value memory latches the density value at the beginning of the input image signal in synchronization with HSYNC, that is, for each scan. This stored value is retained until the next HSYNC. The comparator 3 compares the density value of the first pixel latched in the memory 2 with the density value of the image signal after that pixel. The initial value latched in memory 2 is d. When the pixel density of each data string is d8, the comparator 3 outputs ld, -d, and when l<ε, the frame memory stores d, which indicates that the area is within the road.
in, and a count energizing signal=1 is output to the counter 4. In addition, when ldo −d, I≧ε, d indicates an area outside the road. ut and a count energizing signal=O. Since the counter 4 is reset at the same time as the content of the counter 4 is output at HSYNC, the counter 4 counts the number of pixels for which IdO-d,+<ε between HSYNC. Therefore, the counter output ao 1 al
+al +a3 indicates a linking edge address for each line of pixels.

In this way, the tangled ends are detected from the side where the camera 1 outputs the image signal, so the tangled ends are recognized in real time.

The present invention can be modified and improved in various ways without departing from the spirit thereof.

For example, in the above embodiment, the road edge detection results are stored in the frame memory 6, but the frame memory is not essential to the present invention, and the present invention is not limited to the set position of the camera. Furthermore, the present invention is not limited to the type of image input means.

(Effects of the Invention) As explained above, according to the environment recognition device for a moving vehicle of the present invention, it is possible to detect road tangles in real time.

[Brief explanation of drawings]

Fig. 1A is a diagram showing the configuration of an embodiment to which the present invention is applied. Fig. 1B is a diagram explaining the installation mode of the camera of this embodiment. Fig. 2 is a diagram explaining the principle of road edge detection in the embodiment. FIG. 3A is a diagram showing the configuration of an example of an environment recognition device according to a conventional example, and FIGS. 3B and 3C are diagrams illustrating drawbacks in the conventional example. In the figure, 1... CCD camera, 2... Initial value memory, 3...
- Comparator, 4... Counter, 5... Roadside detection device,
6... Frame memory.

Claims (1)

[Claims] (1) In an environment recognition device for a moving vehicle that inputs an image of the external environment and recognizes the environment from this image, the vehicle traveling direction is the substantial main scanning direction of the raster scan method, and the image in front of the vehicle is converted into a serial image signal. 1. An environment recognition device for a moving vehicle, comprising: an image input means for inputting the serial image signal; and a road edge detection means for detecting a road edge by detecting a change in density with respect to the serial image signal.