JPH08221700A - Stop lamp recognition device - Google Patents

Abstract

PURPOSE: To recognize the illumination of stop lamps irrespective of the vehicle to be recognized and changes of image pickup environment. CONSTITUTION: This device is provided with an image pickup means 1 which picks up an image in the travel direction of this vehicle, a specific color part extracting means 2 which extracts a specific color part from image data from the image pickup means 1, a luminance data extracting means 3 which extracts luminance data from the image data, and a judging means 4 which judges that the stop lamps of a precedent vehicle is lighted when the luminance data of the specific color part exceeds a specific threshold value, and the judging means 4 is provided with a luminance value retrieval part 5 which retrieves stop lamp luminance being a luminance value larger the threshold value or tail lamp luminance being a luminance value less than the threshold value in the luminance data and a threshold value update part 6 which updates the threshold value by using a constant ratio to the stop lamp luminance or tail lamp luminance as a target value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a stop lamp recognizing device, and more particularly to a stop lamp recognizing device for recognizing a stop lamp of a preceding vehicle during traveling.

[0002]

2. Description of the Related Art In order to recognize a stop lamp of a preceding vehicle at night, it is necessary to distinguish the stop lamp from the tail lamp and the stop lamp because the lamp brightness has two levels. As an identification method, it is conceivable to set a certain threshold value determined from a statistical distribution of the lamp brightness of the tail lamp and the stop lamp, and recognize the lighting of the stop lamp with this threshold value as a boundary.

[0003]

However, in this method, since the brightness of the stop lamp is different for each vehicle,
There is an inconvenience that it is difficult to uniquely determine an appropriate threshold value. In particular, it is difficult to uniquely set a threshold value that takes into consideration the vehicle type and secular change.

Further, according to this method, the brightness of the lamp in the captured image is largely influenced by the aperture of the image pickup optical system, the shutter speed, the environment such as the weather, and thus it is difficult to appropriately adjust the brightness. is there.

[0005]

SUMMARY OF THE INVENTION An object of the present invention is to improve the problems of the prior art, and in particular, to provide a stop lamp recognition device capable of recognizing the lighting of a stop lamp regardless of changes in the vehicle to be recognized and the imaging environment. That is the purpose.

[0006]

Therefore, in the present invention, as a first means, an image pickup means for picking up an image of the traveling direction of the vehicle and a specific color for extracting a specific color portion in the image data from the image pickup means. Partial extraction means, brightness data extraction means for extracting brightness data in the image data, and determination that the stop lamp of the preceding vehicle is turned on when the brightness data of the specific color portion exceeds a predetermined threshold value And means. Of these, the determination means searches the brightness value in the brightness data of the specific color portion for a predetermined stop lamp brightness which is a brightness value higher than the threshold value or a predetermined tail lamp brightness which is a brightness value lower than the threshold value. And a threshold value updating unit that updates the threshold value with a fixed ratio to the stop lamp brightness or the tail lamp brightness as a target value.

As a second means, the predetermined stop lamp brightness of the first means is the lowest brightness among the brightness values larger than the threshold value in the brightness data of the specific color portion, and the predetermined tail lamp brightness is The maximum luminance of the luminance values smaller than the threshold value in the luminance data of the specific color portion is adopted.

As a third means, the brightness value search section of the first means has a function of searching the maximum value of brightness in the brightness data of the specific color portion, and the threshold value update section operates as follows. A function that updates the threshold value based on the stop lamp brightness when the maximum brightness value is higher than the threshold value, and a threshold value based on the tail lamp brightness when the maximum brightness value is lower than the threshold value. It is equipped with a function to update.

As a fourth means, the threshold updating unit of the first means calculates the reference value of the tail lamp brightness based on the latest tail lamp brightness and the tail lamp brightness up to the last time, and the latest stop lamp brightness. A function that calculates the reference value of the stop lamp brightness based on the previous stop lamp brightness and the reference value of the stop lamp brightness when the reference value of the stop lamp brightness is not within a certain ratio to the reference value of the tail lamp brightness. It has a function of updating to a value within a fixed ratio and a function of updating a threshold value as a fixed ratio to the updated reference value of the stop lamp brightness or the reference value of the tail lamp brightness. There is.

The present invention aims to achieve the above-mentioned object by these means.

[0011]

According to the first means, when the image pickup means images the traveling direction of the vehicle during the operation of the stop lamp recognition device, the specific color portion extraction means extracts the specific color portion from the image data from the image pickup means. Then, the brightness data extraction means extracts the brightness data in the image data. The determination means determines that the stop lamp of the preceding vehicle is turned on when the brightness data of the specific color portion exceeds a predetermined threshold value.

This judging means successively changes the threshold value using the brightness data. First, the brightness value search unit searches the stop lamp brightness that is a brightness value larger than the threshold value in the brightness data, and also searches the tail lamp brightness that is a brightness value smaller than the threshold value. At least one of the values is retrieved depending on the states of the stop lamp and tail lamp of the preceding vehicle. This stop lamp brightness is
It is a luminance value determined by the current threshold value that the stop lamp is lit. On the other hand, the tail lamp brightness is a brightness value that is not determined to be the lighting of the stop lamp based on the current threshold value.

Next, the threshold value updating unit updates the threshold value with a target value being a fixed ratio to the stop lamp brightness or the tail lamp brightness. For example, set a threshold value that exceeds a certain percentage of the stop lamp brightness,
In addition, the threshold value is set so as to be equal to or less than a certain value of the tail lamp brightness. This is because if the threshold value is too low for the stop lamp brightness, the tail lamp may be mistakenly recognized as a stop lamp, and if the threshold value is too high for the tail lamp, the stop lamp may be mistaken as a tail lamp. This is because there is a possibility of recognition.

For this reason, the threshold value updating unit changes the threshold value with a decrease in tail lamp brightness and an increase in stop lamp brightness. Then, since the brightness of the tail lamp of the preceding vehicle and the brightness of the stop lamp have a constant ratio, the threshold value is a value between the brightness of the tail lamp and the brightness of the stop lamp determined by this ratio. Therefore, for example, when the brightness of the tail lamp decreases when the preceding vehicle is replaced, the threshold value is updated to decrease accordingly, so even if the brightness value is lower than the previous time when the stop lamp is turned on, this is turned on. To determine.

Even if the tail lamp lighting is erroneously recognized as the stop lamp lighting immediately after the operation is started, the threshold value is updated by the ratio of the stop lamp brightness to the stop lamp lighting. At the threshold value, the lighting of the tail lamp is not recognized as the lighting of the stop lamp.

In the second means, the brightness value searching means searches the lowest value of the brightness values above the threshold value as the stop lamp brightness value, and the highest value of the brightness values below the threshold value. Search as the tail lamp brightness. Next, the threshold value updating unit updates the constant ratio of the luminance above the threshold value to the highest value among the target values, and also updates the constant ratio of the luminance below the threshold value to the lowest value among the target values. Have been updated as. In this way, since the stop lamp brightness and tail lamp brightness are the values closest to the threshold,
The threshold value is updated with good accuracy by a constant ratio from these luminances.

In the third means, threshold updating processing is performed as follows. The brightness value search unit searches for the highest brightness value in the brightness data. Next, the threshold value updating unit updates the threshold value based on the searched stop lamp brightness when the maximum value of the brightness is larger than the current threshold value. On the other hand, when the maximum brightness value is smaller than the threshold value, the threshold value is updated based on the tail lamp brightness.

Since the threshold value is updated based on the maximum value of the brightness in the brightness data in this manner, the threshold value is updated based on the stop lamp brightness while it is determined that the stop lamp is lighting. The setting process is performed, and while it is not determined that the stop lamp is lighting, the threshold updating process is performed based on the tail lamp brightness.

In the fourth means, reference values are calculated for each of the tail lamp brightness and the stop lamp brightness in order to perform continuous data processing. this is,
A tail lamp brightness reference value is calculated from the latest tail lamp brightness and the tail lamp brightness up to the previous time, and a stop lamp brightness reference value is calculated from the latest stop lamp brightness and the stop lamp brightness up to the previous time.

Further, since the ratio of the brightness of the tail lamp and the brightness of the stop lamp is within a substantially constant range,
When the reference value of the stop lamp brightness is not within the fixed ratio with respect to the reference value of the tail lamp brightness, the reference value of the stop lamp brightness is updated to be within the fixed ratio. This fixed ratio is, for example, a ratio that the reference value of the luminance of the stop lamp is 2 times or more and 3 times or less than the reference value of the luminance of the tail lamp. Next, the threshold value is updated as a constant ratio with respect to the updated reference value of the brightness of the stop lamp or the reference value of the brightness of the tail lamp. The ratio between each reference value and the threshold value is the first or the second.
It is similar to the means of.

By doing so, it is possible to predict the brightness of the stop lamp corresponding to the tail lamp in a state where only the tail lamp brightness is retrieved.
For example, when the preceding vehicle is replaced, the stop lamp brightness of this vehicle is predicted as a ratio by the tail lamp brightness of the vehicle after the replacement, and the threshold value is changed based on this prediction. Then, even when the preceding vehicle is replaced in this way, when the stop lamp is turned on, it is accurately determined.

In addition, since the reference value of the stop lamp or the like is calculated and the threshold value is updated based on the reference value, the determination process that is less likely to be affected by noise due to an instantaneous change in the imaging environment is performed. Is going.

[0023]

An embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a stop lamp recognition device according to the present invention. The stop lamp recognition device is an image pickup unit 1 that picks up an image of the traveling direction of the vehicle 10.
A specific color portion extracting means 2 for extracting the specific color portion 1e in the image data 1a from the image pickup means 1, a luminance data extracting means 3 for extracting the luminance data 1c in the image data 1a, and the specific color portion. The determination means 4 determines that the stop lamp 20a of the preceding vehicle 20 is turned on when the brightness data 1c of 1e exceeds a predetermined threshold value th.

Of these, the judging means 4 searches for a stop lamp brightness S which is a brightness value larger than the threshold value th in the brightness data 1c or a tail lamp brightness T which is a brightness value smaller than the threshold value th in the brightness data 1c. A search unit 5 and a threshold value updating unit 6 that updates the threshold value th with a target ratio being a fixed ratio to the stop lamp brightness S or the tail lamp brightness T are provided. In addition, in this embodiment, the lamp presence / absence detection unit 7 is provided to determine whether the tail lamp 20b or the stop lamp 20a is lit by labeling the specific color portion 1e and determining the presence or absence of the correlation between the labels. .

This will be described in detail.

The image pickup means 1 is a color image pickup means,
The color image data 1a is output to the specific color portion extraction means 2 and the luminance data extraction means 3 at regular intervals. Imaging means 1
Is only required to be able to obtain the specific color required by the specific color portion extracting means 2 and the image data 1a capable of recognizing a change in luminance. In this embodiment, a CCD camera is used and images are taken at intervals of 3 to 4 times per second.

The judging means 4 judges whether or not the stop lamp of the preceding vehicle 20 is turned on based on the threshold value th. Further, the threshold value th is updated for the next stop lamp lighting recognition. Further, the determination result of the stop lamp lighting is output to various devices. For example, it is a device that warns of approaching a preceding vehicle.

FIG. 2 is a block diagram showing the configuration of hardware resources. In this embodiment, an A / D converter 3A for converting the color image data 1a from the image pickup means 1 into a digital signal and a data conversion ROM 3B storing a map for data conversion are provided. These function as the specific color portion extraction unit 2 and the luminance data extraction unit 3. RO
The address of M3B corresponds to an arbitrary color represented by RGB from the A / D converter, and the output (map data) for this arbitrary color is stored in the ROM 3B, so the ROM 3B is input. It functions as a data converter that can obtain output results (specific color extraction, brightness conversion) converted for color signals. In this case, the image data 1a can be preprocessed at an extremely high speed because it operates without the CPU 4B.

The judging means 4 is a ROM for data conversion.
Image memory 4A for storing the digital signal output from the
And a CPU that processes various data based on a program
4B and.

FIG. 3 is a block diagram showing a preprocessing process of image data used for recognition of a stop lamp or the like. In the present embodiment, the specific color portion extracting means 2 extracts the specific color portion 1e of the 8-bit RGB digital signal (original image) 1b to obtain the 1-bit specific color extraction binary image 1d. Here, a typical tail lamp 20b and stop lamp 2 are used.
The color of 0a is the specific color. Further, the brightness data extracting means 3 converts the original image 1b into an 8-bit (256 gradations) monochrome brightness image 1c.

FIG. 4 is a diagram showing an example of these images, FIG. 4 (a) shows an example of the original image, and FIG. 4 (b) shows an example of the specific color extracted image 1d. To be precise, the original image is color multi-gradation data, but the simplified notation is used in the figure. Since only the specific color given in advance is cut out,
As shown in (b), the stop lamp portion of the preceding vehicle 10 is extracted well. The noise that is included to some extent is removed by labeling processing and determining whether there is a correlation between labels. The correlation between the labels is confirmed by whether or not the change in the position and size of the label between consecutive images is within a certain threshold value.

This embodiment has the following three characteristic points particularly in the recognition processing of lighting of the stop lamp 20a of the preceding vehicle 20.

(1). First, by utilizing the fact that the brightness of the tail lamp and the brightness of the stop lamp are in a constant ratio, the threshold value th is set with a constant ratio to the stop lamp brightness S or the tail lamp brightness T in the image data 1a as a target value. I am updating.

Here, a method is employed in which the ratio of the threshold value to the stop lamp brightness or the tail lamp brightness is obtained and the threshold value is updated according to this ratio. For example, when the brightness of the stop lamp to the brightness of the tail lamp is 3: 2, the threshold value th is ⅔ or more of the stop lamp brightness and 1.5 times or less of the tail lamp brightness. Therefore, when only the tail lamp brightness T is being imaged, the threshold value th corresponding to the tail lamp brightness T is induced to be 1.5 times or less of the tail lamp brightness T. Therefore, when only the lighting of the tail lamp 20b is being imaged, it is possible to update the brightness of the lighting of the stop lamp 20a by the tail lamp to an assumed threshold value.

The stop lamp brightness S is a threshold value t
The brightness value is higher than h, and the tail lamp brightness T is a brightness value lower than the threshold value th.
More specifically, the brightness value on the specific color portion 1e in the labeled image data is separated by the current threshold value th. In the present embodiment, since the brightness data is 8 bits, this brightness value is a density value of 256 gradations. One of the separated density values is used as the lamp brightness.

As the brightness of these lamps, it is desirable to use a value closest to the threshold value. Therefore, the stop lamp brightness S is the lowest value of the brightness values above the threshold value, and the tail lamp brightness T is the highest value of the brightness values below the threshold value. In practice, it is normal for there to be only stop lamp brightness or tail lamp brightness, so
It is the minimum value or the maximum value in the label of the specific color portion 1e.
Thus, by setting the values close to the threshold value to the lamp brightnesses S and T, the target value of the threshold value can be determined from these lamp brightnesses S and T at a constant ratio.

Further, the lamp brightness S, T may be an average of the respective brightness values with a threshold as a boundary, or may be a central value. The present embodiment is characterized in that the stop lamp brightness S and the tail lamp brightness T are separated with the threshold value as a boundary. What value is used as the stop lamp brightness S or the like is determined in consideration of the following points. ing.

First, the ratio of the threshold value to the target value is stable, and second, the number of processing steps such as retrieval is small. As mentioned above, the value closest to the current threshold was the best. However, the stop lamp brightness S may be averaged in consideration of the influence of noise, or may be the maximum brightness value.

(2). Secondly, the processing is branched depending on the relationship between the maximum brightness value in the specific color portion 1e and the threshold value th.

In order to effectively use the lamp brightness T and S, in this embodiment, the maximum value of the brightness in the image data and the threshold value th.
Which lamp brightness T or S is to be used is determined by the relationship with. Therefore, the following configuration is adopted.

The brightness value search unit 5 has a function of searching for the highest brightness value H in the brightness data 1c, and the threshold value updating unit 6 determines that the highest brightness value H is higher than the threshold value th. A function of updating the threshold value th based on the stop lamp brightness S when it is larger, and a threshold value th based on the tail lamp brightness T when the maximum value H of the brightness is smaller than the threshold value th. It has the function to do. Therefore, when only the tail lamp 20b is detected, the threshold th is calculated based on the tail lamp brightness T, while when only the stop lamp 20a is detected, the threshold value th is calculated based on the stop lamp brightness S. The value th is updated.

(3). In order to utilize the fact that the luminance of the tail lamp and the luminance of the stop lamp are at a fixed ratio, the luminance to be recognized as the stop lamp and the reference value of the luminance to be recognized as the tail lamp are calculated. This reference value is a value that takes into account the continuity of the stop lamp brightness S and the tail lamp brightness T and the ratio of each brightness. Specifically, it has the following configuration.

The threshold updating unit 6 calculates the reference value Tt of the tail lamp brightness T from the latest tail lamp brightness Tn and the previous tail lamp brightness Tm, and the latest stop lamp brightness Sn and the previous stop lamp brightness Tm. It has a function of calculating a reference value St of the stop lamp brightness based on the brightness Sm. This is done by updating the reference values Tt, St of each lamp brightness up to the previous time with the latest lamp brightness Tn, Sn.

Further, when the reference value St of the stop lamp brightness is not within the fixed ratio with respect to the reference value of the tail lamp brightness Tt, the threshold updating unit 6 sets the reference value of the stop lamp brightness St within the fixed ratio. It has a function to update. This is because when the brightness of the stop lamp 20a is set to be 2 to 3 times the brightness of the tail lamp 20b, when the reference value St of the stop lamp brightness is updated, this reference value St is 2 of the reference value of the tail lamp brightness.
Check if it is less than double. On the other hand, when the reference value Tt of the tail lamp brightness is updated, it is confirmed whether the reference value St of the stop lamp is three times or more the reference value Tt of the tail lamp.

When the reference value St of the stop lamp is twice or less or three times or more the reference value Tt of the tail lamp, the reference value St of the stop lamp is updated to be within the ratio. Further, the threshold updating unit 6 has a function of updating the threshold as a constant ratio with respect to the updated brightness of the stop lamp or the brightness of the tail lamp.

This (3). By the method using the reference value shown in (1), firstly, the ratio of the luminance of the stop lamp to the tail lamp can be directly used, and secondly, the influence of noise inevitably generated on the continuous luminance data 1c. Can be minimized.

Next, the operation will be described.

FIG. 5 is a flow chart showing an example of the stop lamp recognition processing. In order to recognize the stop lamp 20a of the preceding vehicle at night, it is necessary to detect the change in the brightness of the lamp and distinguish between the tail lamp 20b and the stop lamp 20a. First, the image data 1a in front of the vehicle 10 captured by the image capturing means 1 is captured (step S1). Next, the specific color portion extraction means 2 extracts the specific color portion 1e from the image data 1a and labels this stop lamp extracted image (specific color portion 1e) (step S).
2). Further, if the label is present (step S
3) Confirm the label correlation between the screens (step S4). This is the previously captured image data 1a
Then, it is checked whether or not the change in the label position and size between the latest image data 1a is within a predetermined threshold value. If there is no label (step S3), or if there is no correlation between screens (step s4), the process moves to the next screen.

When the correlation is confirmed, it is determined that the lamps 20a and 20b of the preceding vehicle are on (step S5). Next, the lamp brightness of the specific color portion 1e is calculated (step S6). This is a search process for the highest brightness value H. Further, when the maximum value H of the brightness exceeds the threshold value th (step S7), it is determined that the stop lamp 20a is turned on (step S8). On the other hand, when the maximum brightness value H does not exceed the threshold value (step S7), it is determined that the tail lamp 20b is turned on (step S9). Next, the threshold th is updated based on the currently processed image (step S10).

FIG. 6 shows that the threshold value th is the latest image data 1
It is a flow chart which shows an example of processing which updates based on a. By the processing shown in FIG. 6, the threshold value shown in FIG. 7 is updated. First, the terms on the flowchart shown in FIG. 6 are defined.

Maximum lamp brightness H ..... lmp_max Minimum value of lamp brightness above threshold value S ..... lmp_u_min Maximum value of lamp brightness below threshold value T ..... lmp_l_max threshold value th ..... brt_th

Reference value St of the minimum value lmp_u_min of the lamp brightness above the threshold value St ... ave_u Reference value Tt of the maximum value lmp_l_max of the lamp brightness below the threshold value Tt ..... ave_l

Here, ave_u is set to be 2 times or more and 3 times or less than ave_l.

First, the maximum value lmp_max in the brightness data is searched (step S21). Then, the lowest value lmp_u_min (stop lamp brightness S) in the lamp brightness equal to or higher than the threshold value is searched (step S22). This lmp_u_min
Is the minimum luminance value recognized as the lighting of the stop lamp 20a. Further, the maximum lamp brightness lmp_l_max (tail lamp brightness T) equal to or less than the threshold value is searched (step S23). This lmp_l_max is the maximum value of the brightness that is not determined as the lighting of the stop lamp 20a. The search for these three values is performed for each captured screen.

Further, it is determined whether or not the lamp is present in the screen based on the correlation between the labels on the specific color data (step S25). If the lamp does not exist in the screen, the process moves to the next screen.

If there is a lamp, the maximum value lmp_max in the brightness data is compared with the current threshold value brt_th (step S26). If the current threshold value brt_th is large, use lmp_l_max to set the reference value for tail lamp brightness, ave_l.
Is updated by the following equation (1) (step S27).

Ave_l _n = {ave_l _n-1 x (n-1) + lmp_l_max} / n ..... Expression (1)

Here, n is set to 10 to 30 when imaging is performed in a cycle of 3 to 4 times per second.

By this, the reference value of the tail lamp brightness
ave_l is the highest value lmp in the brightness data 1e of the current image
approaches _max. At this stage, the maximum value lm in the luminance data
p_max is treated as the tail lamp lighting.

Next, in step S26, the tail lamp brightness reference value ave_l and the stop lamp brightness reference value ave_l
Compare with That is, the maximum value of the luminance determined to turn on the tail lamp and the minimum value of the luminance determined to turn on the stop lamp are compared (step S28).

When the reference value ave_u of the stop lamp brightness is larger than three times the reference value ave_l of the tail lamp brightness, the brightness value of the stop lamp lighting is 3 times or more the brightness value of the tail lamp lighting. The stop lamp brightness reference value ave_u is set to the maximum brightness value currently being processed lmp_l_max and the corresponding tail lamp brightness reference value ave_l
Is updated to an appropriate value within 3 times the value of (2) using the following equation (2) (step S29).

Ave_u _n = {ave_u _n-1 x (n-1) + ave_l x 3} / n ... Equation (2)

According to this equation (2), ave_u, which was three times as large as ave_l or more, is brought close to three times. That is, the reference value av of the stop lamp brightness according to the current tail lamp brightness T
e_u is updated. Then, the process proceeds to step S30.

In step S28, the reference value ave_u of the stop lamp brightness is the reference value ave_l of the tail lamp brightness.
If it is within 3 times, the process proceeds to the next step S30.

In step S30, the threshold value brt_th is updated by the following equation (3).

Brt_th _n = {brt_th _n-1 x (n-1) + min (ave_ux0.9, ave_l x 1.5)} / n ... Equation (3)

In the equation (3), min (ave_u x 0.9, ave_l x
(1.5) indicates that the smaller one of ave_u x 0.9 and ave_l x 1.5 is selected here for calculation.

When the appropriate value of the threshold value is set to ave_u x 0.9 or less or ave_l x 1.5 or more by this equation (3), the threshold value brt_th is brought close to the range.

On the other hand, in step S26, the maximum value lmp_max in the brightness data is compared with the current threshold value. If the current threshold value is small, the stop lamp brightness lm
The reference value ave_u of the stop lamp brightness is updated by the following expression (4) using p_u_min.

Ave_u _n = {ave_u _n-1 x (n-1) + lmp_u_min} / n ... Equation (4)

By the equation (4), the reference value ave_u of the stop lamp brightness is brought close to the minimum brightness lmp_u_min which is equal to or more than the threshold value. Then, the reference value of this stop lamp brightness ave_u
And the tail lamp brightness reference value ave_l are compared (step S32).

When the reference value ave_u of the stop lamp brightness is smaller than twice the reference value ave_l of the tail lamp brightness, that is, the brightness value determined to be the stop lamp lighting is twice the brightness value determined to be the tail lamp. If not, the reference value ave_u of the stop lamp brightness is updated by the following equation (5).

Ave_u _n = {ave_u _n-1 x (n-1) + ave_l x 2} / n ..... Expression (5)

From this equation (5), the reference value ave_u of the stop lamp brightness is set to the tail lamp brightness lmp of the current screen.
It is set to an appropriate value in relation to _l_min. That is,
The minimum value of the brightness value judged as the stop lamp lighting is
The processing is performed so that the maximum value of the brightness detected as the tail lamp lighting is twice or more. Then, it proceeds to step S34.

In step S32, the reference value ave_u of the stop lamp brightness is the reference value ave_u of the tail lamp brightness.
If it is larger than twice, it is determined that the reference value ave_u of the stop lamp brightness is appropriate, and the process proceeds directly to step S34.

In step S34, the threshold value brt_th is updated by the following equation (6).

Brt_th _n = {brt_th _n-1 x (n-1) + (ave_u x 0.9)} / n ... Equation (6)

As a result, the threshold value brt_th is updated so as to approach 1.5 times the reference value ave_u of the stop lamp brightness.

FIG. 7 is an explanatory diagram showing an example of setting the brightness threshold value brt_th according to this algorithm and determining the stop lamp lighting. FIG. 7A shows the case where the lamp brightness is relatively stable. In such a case, the stop lamp and the tail lamp can be separately determined even when the brightness threshold value is set to a constant value. However, as shown in FIG. 7B, when the lamp brightness changes abruptly, the two cannot be separated by a fixed brightness threshold value.

FIG. 7B shows a vehicle 20A with high lamp brightness.
This is an example in which the vehicle 20B having low lamp brightness is recognized subsequent to. The brightness threshold value brt_th calculated based on the algorithm according to the present embodiment decreases as the lamp brightness decreases, and even in such a case, the stop lamp lighting can be recognized. FIG. 7C is an example in which the tail lamp brightness is higher than the brightness threshold value and is erroneously recognized (the inside of the chain double-dashed line, the tail lamp is erroneously recognized as a stop lamp). In this case, the stop lamp 20a of the preceding vehicle
When is turned on, the brightness threshold value brt_th rises, and the normal recognition state is restored.

As described above, according to the present embodiment, the threshold value th is changed according to the lamp brightness, so that the erroneous recognition due to the variation in the lamp brightness in the vehicle is reduced. Also, because the threshold itself is configured to change automatically,
It is resistant to environmental changes such as weather, the conditions for setting the imaging device are alleviated, and the burden on maintenance is reduced.

The stop lamp brightness threshold value setting method based on the algorithm of this embodiment is effective when combined with the inter-vehicle distance warning device 30 as shown in FIGS. In this case, the reliability of the alarm is high.

The inter-vehicle distance warning device 30 calculates a safe inter-vehicle distance from data such as the distance to the preceding vehicle 20 and gives an alarm depending on the ratio of the current inter-vehicle distance to the safe inter-vehicle distance. Is. The safe inter-vehicle distance used in this embodiment is given by the following equation (7).

[0085]

[Equation 1]

Here, each value is as follows. V: Own vehicle speed Vr: Relative speed td: Reaction delay time (constant) a: Deceleration (constant)

FIG. 7 is a table showing a method of evaluating the inter-vehicle distance by the inter-vehicle distance evaluating means 32. First, in this embodiment, the range of the inter-vehicle distance in which a warning is required is divided into the following four stages in relation to the safe inter-vehicle distance.

1. Inter-vehicle distance is less than 1/2 of safe inter-vehicle distance (<Rs / 2). The vehicle-to-vehicle distance is smaller than the safe vehicle-to-vehicle distance (<Rs). 3. Distance between vehicles is less than 1.5 times the distance between safe vehicles (<3Rs / 2) 4. Distance between vehicles is more than 1.5 times the distance between safe vehicles (≧ 3Rs
/ 2)

Further, as in the case of the conventional example, there are three stages of alarms as follows. 0. No alarm 1. Primary alarm (to call attention) 2. Secondary alarm (prompt operation)

In the example shown in FIG. 7, the own vehicle brake information indicating the brake state of the own vehicle and the brake information of the preceding vehicle 20 are used for the alarm level determination processing. As a result, it is possible to control the output of the alarm in consideration of the behavior of the preceding vehicle. Further, whether the vehicle is approaching or distant from the preceding vehicle is used as the traveling state information depending on the relative speed of the preceding vehicle. The brake information of the preceding vehicle is the determination result of the stop lamp recognition device according to the above-described embodiment.

According to the evaluation table shown in FIG. 7, when the vehicle 10 is not braking while the preceding vehicle 20 is braking and the vehicle is approaching the preceding vehicle 20, an alarm is issued regardless of the distance. Emit. When the brake of the host vehicle 10 is on and the brake of the preceding vehicle 20 is off, the secondary alarm is issued only when the distance between the inter-vehicle distances is 1/2 or less and the vehicle is approaching. In this way, based on the control of issuing an alarm according to the distance between the preceding vehicle 20 and
Since the relationship between the alarm and the inter-vehicle distance is set based on the own vehicle brake information or the preceding vehicle brake information, the alarm can be issued at an appropriate timing according to the traveling states of the preceding vehicle and the own vehicle.

FIG. 8 is a flow chart showing an example of the alarm output processing according to this embodiment. First, the inter-vehicle distance is measured (step S41). Next, the relative speed is calculated from the change in the inter-vehicle distance (step S42). Here, only the approaching or the distant is calculated. Further, the traveling state information of the own vehicle from the own vehicle brake sensor or the like is acquired (step S43). Next, the safe inter-vehicle distance is calculated (step S44).

Further, the stop lamp of the preceding vehicle is recognized by using the stop lamp recognition device (step S4).
5). Based on the information acquired from step S41 to step S45
a is set (step S46). Further, the warning level is judged based on the evaluation table shown in FIG. 7 (step S47), and the warning is issued (step S48).

In this inter-vehicle distance warning device, since the stop lamp lighting is accurately recognized even at night, and the warning level is determined in consideration of the braking state of the preceding vehicle, the accuracy of the warning output is high. Greatly improved.

[0095]

Since the present invention is constructed and functions as described above, according to this, the brightness value searching unit searches for the stop lamp brightness or the tail lamp brightness, and the threshold updating unit detects the stop lamp brightness or the tail lamp brightness. Since the threshold value is updated with a fixed ratio to the brightness as the target value, the threshold value can be changed as the tail lamp brightness decreases and the stop lamp brightness increases.

Further, the brightness of the tail lamp of the preceding vehicle and the brightness of the stop lamp have a constant ratio, and the threshold value is a value between the tail lamp brightness and the stop lamp brightness determined by this ratio. When the brightness of the tail lamp is lowered due to replacement, the threshold value can be updated to be lowered accordingly. Therefore, the determination means can determine that the stop lamp is lit even if the brightness value is lower than the previous value when the stop lamp is lit.

Even if the tail lamp lighting is erroneously recognized as the stop lamp lighting such as immediately after the operation is started, the threshold value is updated by the ratio to the stop lamp brightness when the stop lamp lighting is updated. With the threshold value, it is possible to avoid erroneously recognizing the lighting of the tail lamp as the lighting of the stop lamp. As described above, it is possible to provide an unprecedented excellent stop lamp recognition device capable of recognizing the lighting of the stop lamp regardless of the vehicle to be recognized or the change of the imaging environment.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of hardware resources of the present embodiment.

FIG. 3 is a block diagram showing a configuration in preprocessing of the present embodiment.

4 is an explanatory diagram showing a specific example of the pre-processing shown in FIG. 3, FIG. 4 (a) is a diagram showing an example of an original image, and FIG.
FIG. 6 is a diagram showing an example of a specific color extraction image.

5 is a flowchart showing an example of a stop lamp recognition process according to the configuration shown in FIG. 1. FIG.

FIG. 6 is a flowchart showing details of a threshold update process shown in FIG.

7 is a diagram showing a change in threshold value over time according to the flowcharts shown in FIGS. 5 and 6, FIG. 7A is an explanatory diagram showing a normal case, and FIG. FIG. 7C is an explanatory diagram showing a case where the vehicle has changed, and FIG. 7C is an explanatory diagram showing a case where the vehicle is initially misrecognized.

8 is a table showing an embodiment of an inter-vehicle distance warning device using the stop lamp recognition device shown in FIG.

9 is a flowchart showing an operation example of the inter-vehicle distance warning device according to the chart shown in FIG.

[Explanation of symbols]

 1 Imaging Device 2 Particular Color Part Extracting Means 3 Luminance Data Extracting Means 4 Judging Means 5 Luminance Value Retrieval Unit 6 Threshold Update Unit 10 Own Vehicle 20 Preceding Vehicle

Claims (4)

[Claims] 1. An image pickup means for picking up an image of a traveling direction of a vehicle,
Specific color portion extraction means for extracting a specific color portion in the image data from the image pickup means, luminance data extraction means for extracting the luminance data in the image data, and luminance data of the specific color portion have a predetermined threshold. And a determining means for determining that the stop lamp of the preceding vehicle is lit when the value exceeds the value, and the determining means is a predetermined stop having a brightness value larger than the threshold value in the brightness data of the specific color portion. Luminance value search unit that searches for lamp luminance or a predetermined tail lamp luminance that is a luminance value smaller than the threshold value, and a threshold value that updates the threshold value with a target ratio of the stop lamp luminance or the tail lamp luminance. A stop lamp recognition device comprising: an updating unit. 2. The predetermined stop lamp luminance is the lowest luminance value of luminance values larger than the threshold value in the luminance data of the specific color portion, and the predetermined tail lamp luminance is the specific color. 2. The stop lamp recognition device according to claim 1, wherein the brightness value is the highest brightness value among the brightness values smaller than the threshold value in the brightness data of the part. 3. The brightness value search unit has a function of searching for a maximum value of brightness in the brightness data of the specific color portion, and the threshold value updating unit has a maximum value of the brightness value from the threshold value. A function that updates the threshold value based on the stop lamp brightness when the value is large, and updates the threshold value based on the tail lamp brightness when the maximum value of the brightness is smaller than the threshold value. The stop lamp recognition device according to claim 1 or 2, further comprising: 4. The function for the threshold value updating unit to calculate a reference value of tail lamp brightness based on the latest tail lamp brightness and the previous tail lamp brightness, and the latest stop lamp brightness and the previous stop lamp brightness. A function of calculating a reference value of the stop lamp brightness, and when the reference value of the stop lamp brightness is not within a certain ratio with respect to the reference value of the tail lamp brightness, the reference value of the stop lamp brightness is a value within the certain ratio. 3. A function for updating the threshold value as described above, and a function for updating the threshold value as a constant ratio to the updated reference value of the stop lamp brightness or the reference value of the tail lamp brightness. Stop lamp recognition device.