JPH01161507A - Monitoring device on moving body - Google Patents

Abstract

PURPOSE:To prevent a target, such as a mark, guide plate, etc., from being overlooked and make reconfirmation of the target possible by discriminating the target by watching the feature in shape of the target. CONSTITUTION:Picture signals from the video camera 21 of an original picture information generating means 2 are converted into the picture information of the black and white level. Since a target of a guide plate, etc., is provided in midair, an MPU 1 can discriminate the target from the original picture information. When the presence of the target is set by an MPU 1, the original picture information is stored in a storing means 3. The stored content of the means 3 is displayed on a CRT 46 in accordance with the instruction of an output means 4. Therefore, the target can be reconfirmed easily. Moreover, when the presence of the target is set, it is informed by means of a buzzer and, at the same time, the moving distance is displayed on the CRT 46. Accordingly, the target can be prevented from being overlooked. Since monochrome picture processes are sufficient, the device used for discriminating the target can be simplified.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention relates to a mobile monitoring device for monitoring targets such as signs and guidance displays related to the operation of automobiles, railway vehicles, ships, etc. .

(Prior Art) For example, in a car, a driver must select necessary information from a large amount of visual information and make an appropriate decision. One of these is road signs and guide displays, but depending on the driving conditions, there may be no time to look at these displays.

Furthermore, if the points of interest are different, these displays may be overlooked.

Therefore, with the aim of further improving safety and drivability,
1985 research presentation materials of the Chubu Branch of Automotive Technology Food (
On June 12, 1982), the following technology was introduced under the title "Development of a Road Sign Recognition System."

- Input an image outside the vehicle using a video camera, - Extract road signs by paying attention to the color information in the input image, and - Recognize the content of the extracted image and notify the driver.

Road signs currently in use are standardized and
Since the colors used for road signs are limited, this technology can be said to be highly reliable.

(Problems to be Solved by the Invention) However, the above technique requires color image processing. Color image processing is R (red), G (green)
In order to perform three systems of processing, ie, B (blue), the amount of processing required is simply three times that of monochrome image processing. Therefore, it is inevitable that the equipment will become larger, the processing will become more complicated, and the cost will increase.

The present invention provides a mobile monitoring device that monitors targets such as signs and guide displays with a simple configuration, and makes it possible to prevent overlooking and reconfirm these targets, thereby further improving safety and drivability. The purpose is to contribute to improvement.

[Structure of the invention]

(Means for Solving the Problems) In order to achieve the above object, in the present invention, an original image information generating means for generating original image information corresponding to at least a part of a scene in a set monitoring direction; A threshold is set based on the correlation between the optical feature component of the original image information included in the set threshold setting area and its occupied area, and using the threshold value, the information contained in at least one monitoring area set in advance in the original image information area is set. The original image information is binarized for optical feature components to generate a binary image information group consisting of a set of first level image information and second level image information, and the first level image included in the binary image information group is When the spatial occupancy rate of the information group is higher than a predetermined value, the spatial feature of the first level image information group included in the binary image information group is extracted, and the spatial feature is replaced with the pre-stored feature. If the determination means sets that there is a target, the determination means sets that there is a target; the storage means; the output means; the output instruction means; control means for instructing output of information read from the storage means;

(Operation) According to this, since the determination is made by paying attention to the morphological features of the target, monochrome image processing is sufficient, and the apparatus becomes simple.

For example, think of road information boards installed on expressways. These road guide boards are standardized and are generally installed at positions where they can be seen through the gap line to facilitate recognition at night. In other words, a road guide board with a uniform shape is floating in the air. From this, it can be easily seen that the road guide board can be extracted by a very simple process that only involves matching the shapes.

According to the present invention, when the determination means sets that there is a target such as a sign or guide display, predetermined information is written into the storage means, so that the target can be reconfirmed by giving an output instruction. For example, if the original image information at the time of setting a target is stored and displayed in response to an output instruction, the target can be easily reconfirmed from the displayed image.

Therefore, in a preferred embodiment of the present invention, when a target is set, a notification is made to that effect, image information at that time is stored, and distance measurement is started. Thereafter, when an output instruction is given, a display is performed based on the stored image information, and the distance traveled since the information was written is output. This effectively prevents the target from being overlooked and increases the significance of the information when reconfirming.

Other objects and features of the present invention will become apparent from the following description of embodiments with reference to the drawings.

(Embodiment) FIG. 1 shows the system configuration of a road guide board monitoring device for automobiles that embodies the present invention as an example. This device consists of a microcomputer (hereinafter referred to as MPU) 1, an image input unit 2. Image memory 3. Out crab knit 4. Crab knit 5. It is composed of RAM6, ROM7, etc.

There are two power sources: the first is supplied from the subunit 81 of the constant voltage unit 8 and is used for heavy loads such as displays, and the second is supplied from the subunit 82 of the constant voltage unit 8 and used for MPUI and Used for logic loads such as various memories. An accessory mode switch Acc and a main switch MSW are interposed between the on-board battery BT and the subunit 81. In other words, the logic load power is always supplied, but the heavy load power is not supplied unless the switches Acc and MSW are turned on.

The image input crab unit 2 has a dedicated microprocessor (hereinafter referred to as CPU) 20, and a video camera 21. A/
It is mainly composed of a D converter 22 and a buffer memory 23.

The video camera 21 is installed near the upper center of the front window indoors, and images the scene in front of the room and outputs an analog image signal of 1 frame of 512×480 pixels. This output is given to the A/D converter 22 and the CRT driver 42 of the display unit 4.

The A/D converter 22 converts the analog image signal provided from the video camera 21 into digital data (gradation 0 is the black level, gradation 255 is the white level) of 256 gradations for each pixel (
gradation data) and provided to the CPU 20 and buffer memory 23.

The CPU 20 monitors the gradation data sequentially provided by the A/D converter 22, determines the suitability of the image captured by the video camera 21, controls the aperture of the video camera 21, and controls the input/output of the A/D converter 22. It also performs input/output control of the buffer memory 23, etc.

The image memory 3 includes a memory area MO (hereinafter referred to as MO memory; others have the same meaning) that stores gradation data for one frame (512 x 480 pixels) outputted from the one-image Kananit 2, and a memory area MO that has undergone predetermined processing. /4 frames (
It has five memory areas M1 to M5 that store gradation data for 256×240 pixels.

The display unit 4 has a dedicated CPU 40, and a character generator 41. Buffer memory 42, latch 43
．． D/A converter 44. CRT driver 45. CRT
46. It is mainly composed of a buzzer driver 47 and a buzzer Bz.

The buffer memory 42 stores digital image data obtained by combining the gradation data for one quarter frame read from the Ml-M5 memory and the character data generated by the character generator 41 according to instructions from the CPU 40. holds one line of digital image data output from the buffer memo 1J42. The CPU 40 controls the input/output of the buffer memory 42 and the latch 43 to enlarge the data for one quarter frame four times (2×2 times).

The D/A converter 44 converts the digital image data provided from the latch 43 into analog data and converts it into analog data.
give to

The CRT driver 45 receives an analog image signal of the front scene given from the video camera 21 of the one-image input unit 2 or the D/A converter 4 under the control of the CPU 40.
4, select the analog image signal given from CRT43
to drive.

The buzzer driver 46 energizes/de-energizes the buzzer Bz in response to instructions from the CPU 40.

The input unit 5 includes an input/output interface 50, operation switches 51, 52, and a counter unit 53. It includes a distance detection switch 54, a rotating magnet 55, and the like.

Although not shown, the operation switches 51 and 52 are located at the center of the steering wheel.

The rotating magnet 55 is a four-pole permanent magnet that rotates in conjunction with the output shaft of the transmission, and the distance detection switch 54 is turned on/off in response to the magnetism of the rotating magnet 55.
Repeat off. In other words, the distance detection switch 54 outputs four pulses for each rotation of the rotating magnet 55, and the number of pulses corresponds to the distance traveled by the vehicle.

The output pulse of the distance detection switch 54 is given to the counter unit 53. The counter unit 53 includes six independent counters C8 to C5, each of which counts up by one each time the pulse applied from the distance detection switch 54 falls. The MPUI resets each counter individually, so each counter individually counts the number of pulses that distance detection switch 54 has output since it was reset.

The ROM 7 stores MPUI control programs, control data, etc., and the RAM 6 stores data being processed.

Next, the general operation of the device of this embodiment will be explained.

Figure 4 shows one image taken in the traveling direction by the video camera 21.
It is the original image corresponding to the scene, and is the road surface.

The white lines separating driving lanes, road information boards, the sky, and the surrounding scenery are shown.

In this, Wl is a vertically elongated window set on the left side of the original image area (hereinafter referred to as the original image area), and usually includes the sky and a part of the scenery, that is, the ground.

Since the difference in gradation between all corresponding pixels and pixels corresponding to the ground is large, the gradation and number of pixels for the pixels included in this window Wl (corresponds to the occupied area since the pixel size is constant)
As shown in FIG. 6, when the relationship between
``A mountain H2 consisting of a group of pixels corresponding to the ground with a low gradation appears. In other words, the gradation Lt of the valley between these peaks
h is a threshold value for extracting the sky from the original image.

Figure 7 shows the original image binarized using the gradations Lt and h as thresholds.
This is a value image in which the road surface, road guide board, and surrounding scenery are separated from the sky and white lines (hatching indicates a group of black pixels with a low gradation).

By the way, since the installation position of the road guide board is almost fixed, the locus of the road guide board that appears on the original image can be limited to a certain range based on the perspective point. Therefore,
First, a part of the binary image is extracted using a first extraction pattern PL (see FIG. 5) assuming that the road guide board is located far away, and its black pixel ratio is examined. At this time, when the black pixel rate is low, the black pixel rate in the second cutout pattern P2 (see FIG. 5), which is assumed to be closer than the first cutout pattern PI, is checked. Furthermore, when the black pixel ratio in the first cutout pattern Pl is high, the same processing as described above is performed on an image captured after a predetermined time to check the black pixel ratio in the second cutout pattern P2.

In either case, when the black pixel rate in the second cutout pattern P2 is low, it is determined that there is no target road guide board, but when it is high, after performing appropriate smoothing processing, the black pixel group Extract the morphological features of.

When this feature matches the shape of the standardized road guide board shown in Fig. 11 or 12, the upper left corner of the original image
The gradation data of the area No. 4 is stored in association with a number, the counter corresponding to that number is reset, and the image indicated by the gradation data is enlarged four times and displayed on the CRT 46.
to be displayed.

Furthermore, in response to the operation of the switch 52, the stored image of the road guide board and the distance traveled by the car since the image was stored are displayed on the CRT 46.

The operation outline of MPUI explained above is shown in Figures 2 and 3.
This will be explained in more detail with reference to the flowchart shown in the figure.

When the onboard battery BT is installed and a predetermined voltage is supplied to each part, the MPU 1 initializes input/output ports, various memories, flags, registers, etc. at St (corresponding to the numbers attached to the flowchart; hereinafter the same meaning). become Note that in the following, it is assumed that switches Ace and MSW are turned on.

In S2, the operation of the switch 51 is monitored, and when the rising edge is detected, the process proceeds to 83 and below, and if the flag FG1 is being reset, it is set in S4, and if it is being set, it is reset in S5.

In S6, the value of the internal timer T1 is checked. This timer TI
This will be explained later, but it should be understood that it is clear here.

Proceeding from S6 to 820, marker detection processing is executed. The marker detection process will be explained below with reference to FIG.

In the label detection process, first, in 5201, the flag FL is set.
Examine G. This flag FLG is a flag for selecting a cutting pattern, and the explanation will be continued assuming that it is currently being reset.

Proceeding from 5201 to 5203, the CPU 20 of the image input crab unit 2 is instructed to output image data.

In the image input crab knit 2, the video camera 21
The image of the front scene is captured by the A/D converter 22, and the gradation data is stored in the buffer memory 23. When the 1MPUI instruction is given, the latest gradation data is sent to the image memory 3. Output. M
The PUI controls the write address of the image memory 3 and writes this gradation data to the MO memory. When the writing of the gradation data to the MO memory is completed, in 5205, the pixels in the window Wl shown in FIG. Create a histogram showing the relationship between each gradation and the number of pixels.

The origin of this window W1 is the upper left corner of the original image area of 512 x 480 pixels, and the coordinates are expressed in pixel units with the horizontal direction being the X axis and the vertical direction being the Y axis.

2 specified diagonally by (10,10), (30,400)
This is a rectangular area of 0x390 pixels.

When creating a histogram, the gradation of each pixel is read while scanning this area, and the corresponding register is counted up.

As mentioned above, two large peaks appear in this histogram as shown in FIG.
Set h.

In step 5207, all pixels in the MO memory are binarized using the threshold value Lth to create binary data and written into the RAMG.As mentioned above, the image (binary image) represented by this binary data is: As shown in FIG. 7, it consists of a black pixel group corresponding to the road surface, road guide board, and surrounding scenery, and a white pixel group corresponding to the sky and white lines. Note that in this embodiment, the black pixel is II O11, and the white pixel is It 1 #l.
Let it be expressed as

Since we are currently discussing the case where the flag FLG is reset, the process proceeds from 5208 to 5209, where the first cutting pattern Pl is added to the binary image and its black pixel rate is checked.

The first cutting pattern P1 is based on the coordinates (80, 10
15 specified diagonally by 0) and (230,190)
This is a rectangular area of 0x90 pixels (see FIG. 5), and the number of black pixels is counted while scanning the binary data corresponding to the pixels in this area.

After counting the number of all black pixels in the first cutout pattern PI, the value is stored in the accumulator A and compared with a predetermined value 4100 in 5210. This value corresponds to approximately 30% of the total number of pixels included in the first cutout pattern P1.

When the total number of black pixels (value of accumulator A) is 4100 or more, that is, when the black pixel rate is about 30% or more, the process proceeds to 5211, but when it is less than 4100, that is, the black pixel rate is about 30%. 8 if less than %
Proceed to 213 and below.

The processing from 8213 onwards is the same as the following explanation except that the original image is not updated, so the explanation here will be omitted.

At 5211, the flag FLG is set and the counter Co
is reset and returns to the main routine (Fig. 2).

From this point on, the counter Co is set to the distance detection switch 54.
Starts counting the number of pulses output by.

Next time this sign detection process is executed, the flag FL
Since G is set, the process advances from 5201 to 5202. At 5202, the value of the counter Co is checked, and if the value is less than the value N (30) corresponding to a distance of 30 m, the process returns to the main routine as is, but the value N (30
), then in step 5203, the latest gradation data at that time is written to the MO memory in the same way as above. That is, the gradation data at this time is created from the front scene imaged approximately 30 meters ahead of the time when the flag FLG was set.

In steps 8205 to 5207, the threshold value h is set based on the current gradation date and night, and binary data is created in the same manner as described above.

This time, the flag FLG is set, so 5208
The process proceeds to step 5212, where the flag FLG is reset, and then, in step 5213, the black pixel ratio in the second cutout pattern P2 is checked.

The second cutout pattern P2 has coordinates (10, 30
) and (190, 130) diagonally specified by 180
This is a rectangular area of x100 pixels (see FIG. 5), and the number of black pixels is counted while scanning the binary data corresponding to the pixels in this area as described above.

After counting the number of all black pixels in the second cutout pattern P2, the value is stored in the accumulator A and compared with a predetermined value 5400 in 5214. This value corresponds to approximately 30% of the total number of pixels included in the second cutout pattern P2.

When the total number of black pixels (value of accumulator A) in the second cutout pattern P2 is less than 5400, that is, when the black pixel ratio is less than about 30%, the process returns to the main routine, but otherwise, it returns to 8215.
Proceed below.

In 5215, the black pixels in the second cutout pattern P2 are reduced. This process is performed using the second cutting pattern P2.
While scanning the inside of the pixel, focus on each pixel sequentially, and even if the pixel you are paying attention to (the pixel of interest) is a black pixel, if any of the 8 surrounding pixels (8 neighboring pixels) is a white pixel, it is forced. This is the process of replacing the pixel with a white pixel. In other words, the 8th
Using a 3 x 3 pixel matrix as shown in the figure, extract the pixel of interest eO and 8 neighboring pixels e1 to e8, and if all the pixels in the matrix are black, leave it as is, and if there is even one white pixel, focus Replace pixel eo with a white pixel.

When this reduction processing is completed, enlargement processing is performed in step 5215. This processing corresponds to the exact opposite processing of the reduction processing described above. In other words, if all the pixels in the 3x3 pixel matrix input are white pixels, even one If there is a black pixel, the pixel of interest eQ is replaced with a black pixel.

By the above reduction and enlargement processing, the second cutting pattern P2
Black pixel groups of 3×3 pixels or less and protrusions included in the image are removed. That is, smoothing processing is performed in 5215 and 5216.

In step 5217, a corner detection process is executed to detect a corner of a black pixel group from the binary data in the second cutout pattern P2 after the smoothing process. Since the road guide board appears as a rectangular block, upper left corner detection, upper right corner detection, lower left corner detection, and lower right corner detection are performed here.

In detecting the upper left corner, the template shown in FIG. 9a is used. This template weights each binary data in a 3x3 pixel matrix, and e2s
From the sum of e3ses and e6, eot et e e
7 shows an operation to reduce the sum of 7 and e8. In this case, if the pixel of interest is the upper left corner of a rectangular black pixel block, the calculated value at this time will be the maximum value 4, as shown in FIG. 10a. That is, when the maximum value 4 is detected, its coordinates are written in the list as the upper left corner.

In the upper right corner detection, 3 as shown in Fig. 9b is used.
Pixels e1pe3+e4 and e in the X3 pixel matrix
Pixel e from the sum of 8 binary data. .

A template is used to reduce the sum of binary data of e5ee6 and e7. In calculations using this template, if the pixel of interest is the upper right corner of a rectangular black pixel block as shown in FIG. 10b, a maximum value of 4 is obtained, so its coordinates are written in the list.

In the lower left corner detection, 3 as shown in Fig. 9c is used.
Pixels e4wes+e7 and e in the X3 pixel matrix
Pixel e from the sum of 8 binary data. .

A template is used to reduce the sum of binary data of 8i, e2, and e3. In calculations using this template, if the pixel of interest is the lower left corner of a rectangular black pixel block as shown in FIG. 10c, a maximum value of 4 is obtained, so its coordinates are written in the list.

In the lower right corner detection, 3 as shown in Figure 9d.
Pixels elee2ee6 and e in ×3 pixel matrix
pixel e from the sum of binary data of 7. .

A template is used to reduce the sum of binary data of e3e84 and e5. In calculations using this template, if the pixel of interest is the lower right corner of a rectangular black pixel block as shown in FIG. 10d, a maximum value of 4 is obtained, so its coordinates are written in the list.

If there is at least one upper left corner and at least one lower right corner in the list created in the corner detection process, 52
In step 19, the slope of the diagonal line specified by them is determined. In this case, when there are a plurality of upper left corners and/or lower right corners, the slope of the diagonal line specified by each corner is determined.

Figures 11 and 12 show the standards for road guide boards actually used on expressways. If there is one that matches the slope of the diagonal line connecting the upper left corner and the lower right corner of the board, it is determined that the black pixel block matches the road guide board, and the process proceeds to step 8224.

If the slopes of the diagonals do not match, or if there is no pair of upper left corner and lower right corner that can identify the diagonal in the list created in the corner detection process, proceed to 8221 and below and check the list created in the corner detection process again. . At this time, if there is at least one upper right corner and at least one lower left corner in the list, 5
Calculate the slope of the diagonal at 222 and 5223,
The compatibility between this inclination and the inclination of the diagonal line connecting the upper right corner and the lower left corner of the road information board is examined. As a result, if it is determined that there is compatibility, the process proceeds to step 8224 and below, and if it is determined that there is no compatibility, the process returns to the main routine from 5223. If there is no pair of upper right corner and lower left corner for which a diagonal can be specified in the list created in the course detection process, the process returns to the main routine from step 5221.

At 5224, the value of the numeric register i is incremented by 1 (+1). However, as a result, when the value of register i exceeds 5, 1 is stored at 8226.

5227, the counter Ct 1 specified by the value of the numeric register i. For example, if the value of i is 3, the counter C
3. Reset. In other words, the counter Ci counts the number of pulses output by the distance detection switch 54 after this point.

8228, the tone data of the upper left 1/4 area of the original image area, that is, the coordinates (0, 0) and the coordinates (256, 24
0) performs contour enhancement processing on the gradation data within the area specified diagonally. This process is a process of subtracting data obtained by spatially second-order differential of the gradation data from the gradation data of the original image.

The characteristics of this processing will be explained using the example shown in FIG.

Figure 13 (,) is a graph showing the relationship between the spatial spread and gradation of certain gradation data distributed two-dimensionally, and when this is first differentiated with respect to the spatial spread, the figure (b) is shown.
A graph as shown in FIG. 2 is obtained, and by further differentiating it, a graph as shown in FIG. 2(c) is obtained. Using this graph to find the difference between the graph in (a) and the graph in (a), we can see that the difference in gradation in areas where the gradation changes spatially, as shown in (d) in the same figure, is emphasized, i.e. in the contour area. A graph is obtained.

In this embodiment, second-order differential data is obtained using a Brassian matrix as shown in FIG.

In other words, from 4 times the pixel of interest eQ, the pixel elte3*es
The value obtained by subtracting the sum of , and e7 is taken as the second-order differential data of the pixel of interest QQ.

This contour enhancement processing improves the resolution of characters written on the road guide board in the original image.

5229 stores the gradation data of the upper left 1/4 area of the original image area that has been subjected to one edge enhancement process in the Mi memory specified by the value of the register i, for example, if the value of i is 3, it is stored in M3.
Store in memory.

At 5230, the flag FGI is checked, and if it is set to reset 1, the process returns directly, but when it is set, the internal timer T2 is cleared and started at 5231, and at 5232, the CP of the display unit 4 is reset.
It instructs U40 to energize the buzzer Bz, and instructs it to output and display the image data in the Mi memory, that is, the image data stored in 5229.

As a result, the CPU 40 activates the buzzer Bz via the buzzer driver 47, stores the given image data in the buffer memory 42, enlarges it four times, and sends it to the D/A.
Convert the corresponding image to the CRT via the CRT driver 46.
46.

After this, in 5233, the internal timer T2 waits for the value tl corresponding to about 0.5 seconds, and then in 5234, the CPU 40 is instructed to deactivate the buzzer Bz, and then in 5234, the CPU 40 is instructed to deactivate the buzzer Bz.
At 35, the internal timer T2 takes a value t2 corresponding to approximately 3 seconds.
After waiting for the CPU 40 to
A reset instruction is issued to switch the display on the RT 46 to the normal front scene.

That is, when the flag FGI is set, the buzzer Bz is energized for about 0.5 seconds each time a road guide board is detected, and the road guide board is enlarged and displayed on the CRT 46 for about 3 seconds.

Please refer to FIG. 2 again. The case where the switch 52 is operated will be explained below. In S9, the switch 5
When the start of operation No. 2 is detected, the flag FG is set in S10.
Check 2. At this time, if flag FG2 has been reset, it is set in Sll and the value of register i is stored in register j.

Proceeding from Sll to 315, the internal timer TI is cleared and started.

In S16, a value obtained by subtracting the value of the register j from the value of the register i is stored in the accumulator A. Since the value of register i has just been stored in register j in Sll, these values are equal and the value of accumulator A is O. Therefore, proceed to 319, where Mj specified by the value of register j
The image data of the memory, the count data of the counter Cj, and the data to the accumulator are displayed on the C of the display unit 4.
It is output to the PU 40 to instruct 1 display.

The count data of the counter Cj is input to the distance detection switch 54 after the MPUI writes the image data to the Mj memory.
is the number of pulses output, and indicates the distance traveled from that point. Further, the value of the accumulator A indicates how many years ago the image data in the Mj memory corresponds to the latest image data. Therefore, the CPU 40 sets the counter Cj
Convert the count data of
The message specified by the value of is selected, and the character generator 41 is instructed to generate character data corresponding to each message, and the character generator 41 is combined with the image data of the 1Mj memory and stored in the buffer memory 42. Thereafter, as described above, the image data including character data is enlarged four times and displayed on the CRT 46.

Next, when the switch 52 is operated, the flag FC is
2 is set, so at 312 the register j
Decrement the value by 1. This allows the digit register j
When the value of becomes less than O, the value is changed to 5 in Si2.
Correct to.

As described above, the internal timer T1 is cleared and started in S15, and the value obtained by subtracting the value of the digit register j from the value of the digit register i is stored in the accumulator A in S16. Since the value of the register j is decremented each time the switch 52 is operated, the value of the accumulator A at this time increases each time, but since the value of the register j is corrected in S14, this value will not be negative. It may become. In that case, in step 818, 5 is added for correction.

In S19, M specified by the value of the register j
The image data of the j memory, the count data of the counter cj, and the data of the accumulator A are outputted to the CPU 40 of the display unit 4 to instruct display.

In this case, if the value of accumulator A is a, the image containing the road guide board detected a times ago, the distance traveled from there, and the message indicating that it was a time ago are displayed on the CRT.
46 and is enlarged and displayed.

When the switch 52 is no longer operated, the process advances to S6 and the value of the internal timer TI is checked. This value corresponds to approximately 5 seconds t
If the value exceeds o, the process proceeds to S8, where the flag FG2 is reset and a reset instruction is issued to the CPU 40 to switch the display on the CRT 46 to the normal front scene.

In other words, when the switch 52 is operated, an image including the latest road information board and the distance traveled from there are displayed.

and a message are displayed on the CRT 46 for about 5 seconds. If the switch 52 is operated again while this display is displayed, the 1 display is sequentially switched to the old image, the distance traveled from the detection point, and a message, each of which is displayed for about 5 seconds. However, after the display corresponding to the oldest image, the display returns to the display corresponding to the latest image.

Incidentally, in the above embodiment, the threshold values Lt and h are set using the vertically long window W1 set to the left side of the original image area, but there is no need to be limited to this. For example, the window W2 shown in FIG. 4 is a horizontally long rectangular area set at the upper right of the original image area. This region contains only images corresponding to the sky in the front scene. therefore,
When we create a histogram of gradation versus number of pixels for the pixels in this area, there is only one cursed mountain consisting of a group of pixels corresponding to the sky, which is exactly the mountain from Figure 6! (This is the mode excluding 1. Using the gradation at the foot of this mountain or the gradation at the top of the mountain, you can obtain a threshold that separates the sky from the original image. Using the threshold obtained in this way, You can also get the same result as above.

Further, in the above embodiment, two cutout patterns are used, but more patterns may be used.

In this case, the effect remains unchanged even if only the inside of the cutout pattern is binarized.

Further, the cutout pattern is not limited to a rectangle, but may be set based on a triangle whose apex is a perspective point, for example.

Furthermore, in the above embodiment, the image quality is improved by contour enhancement processing when displaying the detected image, but image quality improvement by histogram correction processing is also effective.

FIG. 15 is an example of a histogram of gradation versus number of pixels shown for pixels within the display area. According to this, the lowest gradation S in this display area is higher than the gradation O, and the highest gradation E
is lower than gradation 255.

Therefore, by using the linear function shown by graph f2 in FIG. 16 to convert the input that starts at tone S and ends at tone E from tone 0 to tone 255, the pixels in the display area will be The histogram of gradation versus number of pixels is expanded as shown in FIG. This improves the contrast of the displayed image and increases the resolution of characters and the like.

Incidentally, the graph f1 in FIG. 16 shows a linear function with equal input and output.

〔Effect of the invention〕

As explained above, according to the present invention, target determination processing is performed by focusing on the shape of the target, so that simple monochrome image processing can be used to perform sufficient determination. Moreover, since predetermined information is stored when it is determined that there is a target based on the determination, reconfirmation is possible by reading out the stored information later.

In particular, as described in the explanation of the embodiment, when a target is detected, it is notified, its image is stored, distance measurement is started, and then the image and its image are stored in response to an output instruction. Displaying the distance traveled from a point in time will have a great effect on preventing the target from being overlooked and confirming the target after it has passed.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a system 11 of an automobile road guide board monitoring device. FIGS. 2 and 3 are flowcharts showing the general operation of the microcomputer 1 shown in FIG. FIG. 4 is a plan view showing an example of a sampled front scene. FIG. 5 shows the cutout pattern PL set within the original image area,
It is a top view which shows P2. FIG. 6 is a histogram of tone versus number of pixels shown for pixels included in the window Wl of the original image shown in FIG. FIG. 7 is a plan view showing a binary image obtained by converting the original image shown in FIG. 4 into a binary image. FIG. 8 is a plan view showing a 3×3 pixel matrix. FIGS. 9a, 9b, 9c, and 9d are plan views showing templates for corner detection. Figures 10a, 10b, 10c and 10d are plan views showing corners appearing in a binary image. FIG. 11 and FIG. 12 are plan views showing an example of a road guide board. FIG. 13 is an explanatory diagram for explaining the contour enhancement process. FIG. 14 is a plan view showing a Laplacian matrix used for contour enhancement processing. FIG. 15, FIG. 16, and FIG. 17 are explanatory diagrams for explaining the histogram correction process. Microcomputer (judgment means, control means) 2: Image input crab unit (original image information generation means) 20: Microprocessor 21: Video camera 22: A/D converter 2
3: Buffer memory 3: Image memory (storage means) 30: Input/output interface 4: Display unit (output means) 40: Microprocessor 41: Character generator 42: Buffer memory 43: Latch 44: D/A converter 45: CRT driver 46: CRT (image display means) 47: Buzzer driver (energizing means) 46: Buzzer (notification means) 5: Input/output interface 51.52: Operation switch 52: (output instruction means) 5
3: Counter unit 54: Distance detection switch 55: Rotating magnet 53, 54, 55: (Movement information detection means) 6: RA
M7: ROM8: Constant voltage unit 81.82: Subunit BT: On-board battery Ace: Accessory mode switch MS: Main switch wl, W2: Window (threshold setting area) PI, P2:
Cutout pattern (monitoring area) patent applicant Aisin Seiki Co., Ltd.

Claims (10)

[Claims] (1) Original image information generation means for generating original image information corresponding to at least a part of the scene in the setting monitoring direction; optical characteristic components of the original image information included in a threshold setting area set in advance in the original image information area; A threshold value is set based on the correlation with the occupied area, and using the threshold value, the original image information included in at least one monitoring area set in advance in the original image information area is binarized for optical feature components, and the first level image is A binary image information group consisting of a set of information and second level image information is generated, and when the spatial occupancy rate of the first level image information group included in the binary image information group is higher than a predetermined value, the second level image information A determining means for extracting a spatial feature of a first level image information group included in a value image information group, and setting a target when the spatial feature matches a pre-stored feature; a storage means; an output means; Output instruction means; and when the determination means sets that there is a target, predetermined information is written in the storage means, and when there is an instruction from the output instruction means, the output means is instructed to output the information read from the storage means; A mobile monitoring device comprising: a control means; (2) The determination means extracts at least two corners of the first level image information group, and sets a target when a rectangle designated by the corners matches pre-stored rectangle information, A mobile monitoring device according to claim (1). (3) The determination means binarizes the original image information included in a plurality of monitoring areas set in advance in the original image information area, and determines the first level image included in the binary image information group corresponding to at least one monitoring area. Claim 1 or 2, wherein when the spatial occupancy rate of the information group is higher than a predetermined value, the spatial feature of the first level image information group included in the corresponding binary information group is extracted. The mobile monitoring device according to paragraph (2). (4) The output means includes an image display means for displaying image information, and the control means writes at least a part of the original image information into the storage means when the determination means sets a target. The mobile monitoring device according to paragraph (1). (5) The output means includes a notification means and an urging means for energizing the notification means, and the control means instructs the urging means to activate when the determination means sets a target. A mobile monitoring device according to scope (1). (6) Original image information generating means for generating original image information corresponding to at least a part of the scene in the setting monitoring direction; the optical characteristic components of the original image information included in the threshold setting area set in advance in the original image information area; A threshold value is set based on the correlation with the occupied area, and using the threshold value, the original image information included in at least one monitoring area set in advance in the original image information area is binarized for optical feature components, and the first level image is A binary image information group consisting of a set of information and second level image information is generated, and when the spatial occupancy rate of the first level image information group included in the binary image information group is higher than a predetermined value, the second level image information Judgment means for extracting spatial features of the first level image information group included in the value image information group, and setting target presence when the spatial features match pre-stored features; Detecting movement information detection means; storage means; output means; output instruction means;
When the determination means sets that there is a target, it writes predetermined information into the storage means and instructs the movement information detection means to start detection, and when there is an instruction from the output instruction means, the output means reads out the information from the storage means. A mobile body monitoring device comprising: a control means for instructing the output of the information detected by the movement information detection means and for instructing the output of the movement information detected by the movement information detection means. (7) The determination means extracts at least two corners of the first level image information group, and sets a target when a rectangle designated by the corners matches pre-stored rectangle information, A mobile monitoring device according to claim (6). (8) The determination means binarizes the original image information included in a plurality of monitoring areas set in advance in the original image information area, and determines the first level image included in the binary image information group corresponding to at least one monitoring area. Claim 6 or 7, wherein when the spatial occupancy rate of the information group is higher than a predetermined value, the spatial feature of the first level image information group included in the corresponding binary image information group is extracted. The mobile monitoring device according to paragraph (7). (9) The output means includes an image display means for displaying image information, and the control means writes at least a part of the original image information into the storage means when the determination means sets a target. The mobile monitoring device according to paragraph (6). (10) The output means includes a notification means and an urging means for energizing the notification means, and the control means instructs the urging means to activate when the determination means sets a target. A mobile monitoring device according to item (6).