JP4479532B2 - Fog detection device by night image - Google Patents

Description

  The present invention relates to a fog detection device that detects a fog generation state at night by recognizing and processing an optical axis of a light reflected in a camera image.

Conventionally, a front vehicle collision warning device and an ACC (Adaptive Cruise Control) device are known as a driving support device and an automatic driving device that support a driver. These driving support devices and automatic driving devices determine the control output based on information from a device that recognizes the environment outside the vehicle. Known devices for recognizing the environment outside the vehicle include devices such as a millimeter wave radar device and a laser radar device that measure the distance from an obstacle or a preceding vehicle. For example, in the distance measuring device disclosed in Japanese Patent Laid-Open No. 8-122437, the distance to the obstacle in the mist is measured using laser light.
JP-A-8-122437

  It is desirable that the above-described driving assistance device and automatic driving device determine the assistance contents reflecting the environment outside the vehicle. For example, if the forward vehicle collision warning device can output a warning at an earlier stage when fog is occurring than when fog is not generated, the driver can adjust the inter-vehicle distance more safely. .

  Therefore, the present invention processes the image including the optical axis of the light photographed by the camera to detect the fog generation state particularly during night driving, and reflects this in the control output of the driving support device and the automatic driving device. An object of the present invention is to provide a fog detection device that can perform this.

In order to achieve the above object, according to a first aspect of the present invention, there is provided a fog detecting device that detects the occurrence of fog when traveling at night, a first light that irradiates at least an imaging range of a camera, and at least the first light. A camera for photographing a surrounding environment including an optical axis, and an image processing / fog detecting means for processing an image including the surrounding environment including at least the optical axis of the first light photographed by the camera and outputting a fog detection signal The image processing / fog detection means includes an optical axis estimated position of the first light, wherein a maximum value of a luminance value of a horizontal scanning line of a captured image captured by the camera is calculated from a light state of the first light. When the luminance change amount between the region including the maximum value existing in the vicinity and the left and right optical axis estimation positions including the maximum value is small and the luminance change amount between the regions other than the region maintains a change amount within a predetermined range. fog Characterized in that considered as raw. As a result, it is possible to detect the generation of fog during night driving, and when there is a region on the photographed image where there is little change in luminance other than the optical axis of the first light used for fog detection, for example, a white line representing a driving lane In addition, fog can be detected without erroneous determination.

  According to a second aspect of the present invention, there is provided a light control unit that transmits a control command to the first light and outputs a light state of the first light acquired from the first light to an image processing / fog detection unit; The image processing / fog detection means outputs a fog detection signal based on the light state of the first light input from the control means and the image taken by the camera. Thereby, based on the light state of a 1st light, for example, a lighting / extinguishing state, or an optical axis direction, a fog detection apparatus can perform a fog detection process, and a detection precision improves.

  The invention according to claim 3 includes a second light other than the first light, and the light control means controls an optical axis direction of the first light and the second light to control the first light and the first light. While acquiring both control states of two lights, the light state of the first light and the light state of the second light input from the light control means and the image taken by the camera are processed to generate a fog detection signal. Image processing / fog detection means for outputting is provided. Thus, the fog detection device can detect the light state of the first light, for example, whether the first light is on or off, regardless of whether only the first light is on or a plurality of lights are on. Occurrence of fog to process images taken using information about the light axis direction of one light and the light state of the second light, for example, whether the second light is on or off, or information about the second light's optical axis direction Can be detected. Further, for example, by inputting the output of the fog detection device to the light control means, it is possible to control the optical axis directions of the first light and the second light according to the fog generation state.

  The invention according to claim 4 is characterized in that the light control means outputs at least one control command of a swivel angle and a tilt angle as a control command for the first light and the second light. Thereby, it is possible to control the optical axes of the first light and the second light having a mechanism capable of changing the optical axis direction based on the fog generation state.

  According to a fifth aspect of the present invention, the light control means controls at least one of a switching state between a high beam and a low beam, a swivel angle, and a tilt angle as a light state of both the first light and the second light. It is characterized by acquiring a state. Accordingly, the fog detecting device can determine the fog generation state based on the state of the first light and the second light having a mechanism capable of switching between the high beam and the low beam and a mechanism capable of changing the optical axis direction.

According to the sixth aspect of the present invention, the traveling lane recognition device that recognizes the traveling lane using the captured image captured by the camera detects the horizontal scanning line of the captured image used when detecting a coordinate value from the captured image. The setting algorithm is diverted as an algorithm for the image processing / fog detection means to detect the occurrence of fog. Thereby, since it is sufficient to process a part of the image elements on the photographed image, the processing time of the fog detection process can be shortened.

The invention described in claim 7 is characterized in that the camera is used in combination with a camera used by the travel lane recognition device. Thereby, since it is not necessary to provide the camera only for a fog detection apparatus, the production cost of an apparatus can be reduced.

The invention according to claim 8 is an algorithm for converting a coordinate value of a white belt-like line in the photographed image for detecting the occurrence of fog by the image processing / fog detecting means into a coordinate value of a road surface by geometric conversion. Is diverted from an algorithm for converting the coordinate value of the white line in the captured image used by the traveling lane recognition device into the coordinate value of the road surface by geometric conversion. Thereby, the production cost of the fog detection algorithm can be compressed, and high reliability can be obtained.

  Hereinafter, the best mode for carrying out the present invention will be described using Examples 1 to 5.

[Example 1]
In the present embodiment, an explanation will be given by taking as an example a fog detection device mounted on an automobile during night driving with reference to FIGS.

  FIG. 1 is a block diagram illustrating an example of a fog detection device. A camera 11 for photographing the front of the vehicle is attached to the rear side of the rearview mirror in the windshield of the automobile. The camera 11 is used as one function of cruise control and shares a white line detection camera for lane keeping that is installed in many commercial vehicles. Thereby, compared with the case where the camera 11 is newly installed, it is excellent in cost performance.

  A headlight 12 as a first light and a fog light 13 as a second light are provided at the front of the vehicle body. The headlight 12 is a headlight having an AFS (Adaptive Front-lighting System) function in addition to a high / low switching function, and outputs a high / low lighting state and a swivel angle of the headlight 12 to the light control unit 14. . The fog light 13 outputs a lighting / extinguishing state to the light control unit 14. The light control unit 14 is installed inside the vehicle. The headlight 12 calculates and outputs a target swivel angle from a steering angle, a vehicle speed, and a vehicle height of a steering (not shown). Output. Further, the lighting state of the headlight 12 and the fog light 13 and the swivel angle of the headlight 12 are output to the image processing / fog detection unit 17 as light state 16 data.

  The image processing / fog detection unit 17 is installed inside the vehicle, and performs image processing on the image 15 output from the camera 11 to determine the presence or absence of fog. FIG. 2 shows an image taken when fog is generated (hereinafter, fog generated image). In the fog generation image, the optical axis of the headlight 12 is shown as a white band-like line. The reason why the optical axis of the headlight 12 appears as a white belt-like line in the fog generation image of FIG. 2 is that light is scattered by fog (small particles of water). Further, in the fog generation image, the luminance gradually decreases as the distance from the white belt-like line increases. Under the generation of fog, the location where the white band-like line appears differs depending on whether the lighting state of the headlight 12 is a high beam or a low beam. However, the point that the optical axis clearly appears as a white belt-like line and the brightness gradually decreases as the distance from the white belt-like line is not changed in either lighting state.

  On the other hand, when no mist is generated, there are no small particles of water that scatter light. For this reason, a reflective object having a high reflectivity appears brightly on the road surface instead of clearly appearing as a white strip line due to scattering. For this reason, at the boundary between a white reflective object such as a center line or a guardrail and a black reflective object such as a road surface, the luminance on the black reflective object side is drastically lower than the luminance on the white reflective object side.

  As described above, there are two types of differences between the fog generated image and the image taken when no fog is generated (hereinafter referred to as the fogless image). The first difference is that the optical axis of the headlight 12 appears as a white band-like line in the fog generated image, and the white band-like line does not exist in the image without fog. The second difference is that the brightness distribution of the fog generation image gradually decreases as it moves away from the white belt-like line, and the brightness distribution of the fogless image shows that the brightness on the black reflective object side is the same as that on the white reflective object side. It is a point that it has fallen sharply.

  In the image processing / fog detection unit 17 in FIG. 1, since the fog is detected by using the above-described two kinds of differences, the detection processing includes two steps. In the first step, the image element on the image 15 from the light state 16 is estimated as to which the optical axis of the host vehicle appears as a white belt-like line, and the white belt-like center line clearly appears in the vicinity of the image device. This is a local maximum search process 51 described later with reference to FIG. The second-stage processing is luminance determination processing 52 described later with reference to FIG. 5 for determining whether the luminance gradually decreases as the distance from the white belt-shaped center line appears when a white belt-shaped line appears.

  The reason why the above-described two-stage process is used for fog detection will be described. Since there are headlights and street lamps of oncoming vehicles on the road, there may be white strip lines on the image 15 due to light other than the headlight 12 of the host vehicle. Therefore, the optical axis position of the host vehicle is estimated in advance from the light state 16, the relative position of the camera 11 and the headlight 12, and it is determined whether or not a white belt-like center line exists only in the vicinity of the estimated position. However, in this determination, when the optical axis estimated position and the white center line or guard rail drawn on the road overlap on the image 15, the optical axis is not required even if it is not a white strip line by the optical axis of the headlight 12. There is a possibility of misunderstanding that fog is generated because there is a white strip line at the estimated position. Therefore, when fog occurs, not only the optical axis of the headlight 12 appears as a white belt-like line, but also the feature that the luminance around the white belt-like line gradually decreases is used. As described above, the maximum value search processing 51 for searching whether the optical axis can be detected as a white band-shaped center line in the vicinity of the optical axis estimated position, and the luminance determination for determining whether the luminance gradually decreases as the distance from the white band-shaped line increases. Two-stage processing of processing 52 is required.

  The maximum value search processing 51 will be described with reference to FIGS. 3 (a) to 3 (d). FIG. 3A schematically shows a fog generation image. In order to improve the efficiency of image processing, the luminance on the horizontal scanning line is divided by dividing the image into units of n pixels in the horizontal direction as shown in the horizontal scanning line k, horizontal scanning line k + 1, and horizontal scanning line k + 2 shown in FIG. And detect the optical axis that appeared as a white strip line. If the image size is 640 pixels in the horizontal direction and 480 pixels in the vertical direction, the number of horizontal scanning lines used to detect the optical axis is t (where t = 480 / n). . The image processing method for processing the horizontal scanning lines every n pixels is already used in a white line recognition algorithm for a traveling lane recognition device already installed in many vehicles. If this is used as a fog detection algorithm, the cost is reduced. It is advantageous in terms of performance and reliability.

  3B is the horizontal scanning line k, FIG. 3C is the horizontal scanning line k + 1, and FIG. 3D is the relationship between the luminance of the horizontal scanning line k + 2 and the image element number in the horizontal direction, that is, for each horizontal scanning line. It represents the luminance distribution. As shown in FIG. 3B and FIG. 3C, when a white belt-like line is present in the image 15 due to the generation of fog, the central portion of the white belt-like line has high luminance, so that the left and right headlights 12 respectively There are two local maximums. However, since the brightness of the image is quantified such as 256 gradations, when the light reflection intensity is particularly strong, the brightness is uniformly saturated to the maximum gradation value (255 for 256 gradations) and the luminance distribution Becomes π-type. When the luminance is saturated as in the horizontal scanning line k + 2 in FIG. 3D, the element located in the middle of the saturated element set is set to the maximum value. In the following, the maximum values existing on the left side of the screen are referred to as left maximum values 31 and 35, and the right side is referred to as right maximum values 32 and 36.

  When extracting the coordinate value of the maximum value, the optical axis estimation position is estimated in advance from the mounting position of the camera 11 and the headlight 12 and the light state 16. A range in which the optical axis estimation position has a width is set as a maximum value search range, and it is searched whether or not a maximum value exists in the range. As shown in FIG. 3B to FIG. 3D, the range near the optical axis estimated position on the left side of the image 15 is set as the left search range 33 and the right side is set as the right search range 34 as with the maximum value. The local maximum search processing 51 performs a search on the horizontal scanning line for every n pixels, and has an image having a luminance value that is maximum in the left search range 33 and the right search range 34 on the scan line and has a luminance of a certain value or more. When the element (the left maximum value 31 and the right maximum value 32) exists, it is determined that the maximum value exists.

  Next, the brightness determination process 52 will be described with reference to FIGS. 3A and 3E to FIG. This process is executed only when it is determined in the above-described maximum value search process 51 that there is a maximum value on the horizontal scanning line. Similarly to the local maximum search processing 51, the luminance determination processing 52 also processes the luminance distribution on the horizontal scanning line every n pixels. As shown in FIGS. 3E to 3G, the area to be processed on the horizontal scanning line is divided into three areas of a left luminance determination area 37, a middle luminance determination area 38, and a right luminance determination area 39. The left luminance determination area 37 is an area from the leftmost image element to the left adjacent image element of the left maximum value 31. When the luminance is saturated as shown in FIG. 3G, the leftmost image element including the left maximum value 31 to the left of the leftmost image element among the saturated luminance elements is displayed. A region up to the adjacent image element is defined as a left luminance determination region 37. The medium luminance determination area 38 is an area from the image element on the right side of the left maximum value 31 to the image element on the left side of the right maximum value 32. Similar to the left luminance determination region 37, when the luminance is saturated as shown in FIG. 3G, the right of the image element located on the rightmost side among the elements including the left maximum value 31 in which the luminance is saturated. A region from the adjacent image element to the image element adjacent to the left of the image element located on the leftmost side among the elements saturated in luminance including the right maximum value 32 is defined as a medium luminance determination region 38. The right luminance determination area 39 is an area from the rightmost image element to the right adjacent image element of the right maximum value 32. Similar to the left luminance determination area 37, when the luminance is saturated as shown in FIG. 3G, the rightmost image element including the right maximum value 32 from the rightmost image element is rightmost. An area up to the image element on the right side of the image element located at is defined as a right luminance determination area 39.

  Within these left luminance determination region 37, middle luminance determination region 38, and right luminance determination region 39, the amount of change in luminance for each adjacent image element or predetermined image element is calculated, and the amount of change within a certain range is maintained. If so, it is determined that the brightness is gradually decreasing as the distance from the white belt-like line is increased, and it is determined that fog is generated.

  With reference to the flowchart of FIG. 4, the lighting / extinguishing process of the fog light 13 executed by the fog detecting device will be described. In the initial state of processing, the detection counter and the release counter are 0, and the mode is “idle”. The initial state is a state when the fog detecting device is started. After the start-up, every time an image 15 is taken by the camera 11, the processing from step S401 is performed. As described later, when the image 15 is captured by the camera 11 and the process is executed again with the fog light 13 on / off process temporarily stopped, the detection counter, the release counter, and the mode in the previous process are changed. Used. That is, the detection counter, the release counter, and the mode are declared as global variables, and when the image acquisition interrupt process by the camera 11 is performed, the process of FIG. 4 is performed. The mode takes one of three values: “idle”, “detecting”, and “lighting”.

  In step S401, the image 15 photographed by the camera 11 is acquired, and the process proceeds to step S402. In step S402, fog detection processing including maximum value search processing 51 and luminance determination processing 52 described later with reference to FIG. 5 is performed on the image 15 acquired in step S401, and the process proceeds to step S403. In step S403, branch determination is performed based on the conditions using the fog detection signal output by the fog detection process in step S402. If it is determined that the fog detection signal is fog generation (fog detection signal = 1), the process proceeds to step S404. If it is determined that there is no fog (mist detection signal = 0), the process proceeds to step S405. In step S404, branch determination is performed based on the mode value. If it is determined that the mode is “detecting”, the process proceeds to step S406. If it is determined that the mode is not “detecting”, the process proceeds to step S407. In step S406, 1 is added to the detection counter, and the process proceeds to step S408. In step S408, the detection counter is compared with the specified number of detections. If the detection counter exceeds the specified number of detections, the process proceeds to step S409, and if not, the process is temporarily terminated. However, whenever the process is temporarily terminated, the three values of the detection counter, the release counter, and the mode are held and used when the process is executed again. In this way, each of the three values of the detection counter, the release counter, and the mode is held, and the fog light 13 is turned on when the detection counter exceeds the specified number of detections, and the fog is turned on when the release counter exceeds the specified number of cancellations. The reason for turning off the light 13 is to prevent frequent turning on and off of the fog light 13.

  In step S409 following step S408, the mode is set to “lighting”, and in a subsequent step S410, the release counter is reset (= 0). In step S411 following step S410, the fog light 13 is turned on, and the process is temporarily terminated after the lighting. In step S 407, which proceeds when there is no fog (mist detection signal = 0) in step S 404, branch determination is performed on the condition of the mode value. If the mode is determined to be “idle”, the process proceeds to step S 412 and is not “idle”. If it is determined, the process proceeds to step S414. In step S412, 1 is added to the detection counter. In step S413 following step S412, the mode is set to “detecting”, and in the next step S414, the release counter is reset. Then, after executing step S414, the process is temporarily terminated. If the mode is not “idle” in step S407, the mode is “lighting”, so the release counter is reset in step S414, and the process is temporarily terminated.

  Next, a process when the fog detection signal becomes no fog in step S403 will be described. If the mode is “detecting” in step S405, 1 is added to the release counter in step S415, and if the mode is not “detecting”, the process proceeds to step S416. In step S417 following step S415, the release counter is compared with the specified release count, and if the release counter is equal to or greater than the specified release count, the process proceeds to step S418, and if it is less than the specified count, the process is temporarily terminated. In step S418, the mode is set to “idle”. In step S419 following step S418, the detection counter is reset, and then the process is temporarily terminated. In step S416, which is executed when the mode is not “detecting” in step S405, branch determination is performed based on the mode value. If the mode is “lighting”, the process proceeds to step S420. Is temporarily terminated. In step S420, 1 is added to the release counter. Further, in step S421 following step S420, the release counter is compared with the specified release count, and if the release counter is equal to or greater than the specified release count, the process proceeds to step S422, and if it is less than the specified release count, the process is temporarily terminated. In step S422, the mode is set to “idle”. In step S423 following step S422, the detection counter is reset. Further, the fog light 13 is turned off in step S424 following step S423, and the process is temporarily terminated.

  In the process of turning on / off the fog light 13 shown in FIG. 4, the reason why the fog light 13 is turned on when the detection counter satisfies the specified number of detections and the fog light 13 is turned off when the release counter satisfies the specified number of cancellations. When the white line such as the center line temporarily overlaps with the white belt-like line generated by the optical axis, there is a case where the fog detection condition does not hold and the lighting / extinguishing state of the fog light 13 is frequently switched. This is to avoid it.

  The processing executed by the image processing / fog detection unit 17 of FIG. 1 will be described using the flowchart of FIG. As described above, the fog detection process is performed in two stages. The left side of FIG. 5 is the local maximum search process 51, and the right side is the luminance determination process 52. These processes are step S402 in the flowchart of FIG.

  First, the maximum value search process 51 will be described. In step S501, a horizontal scanning line to be searched is determined. The initial search starts from the horizontal scanning line located at the top of the image. In step S502 following step S501, based on the relative position of the light state 16 and the camera 11 and the headlight 12, it is estimated in which range on the horizontal scanning line the optical axis of the vehicle headlight 12 will be searched. The ranges are defined as the left search range 33 and the right search range 34 described with reference to FIGS. 3 (b) to 3 (d). In step S503 subsequent to step S502, an image element having a maximum luminance value in the left search range 33 that is equal to or greater than a certain value in the left search range 33 on the set horizontal scanning line is searched. Hereinafter, an image element having a luminance value equal to or greater than a certain value and having the maximum luminance value within the search range is referred to as a bright point, and the image element number of the bright point is referred to as a bright point number. If a bright spot is detected during the search in step S503, the bright spot number is stored. In step S504 following step S503, it is determined whether a bright point has been detected in step S503. If a bright point is detected, the process proceeds to step S505. If not detected, the process proceeds to step S506. In step S505, when there is only one bright point in the left search range 33, the bright point is set as the left maximum value 31. When there are a plurality of bright points in the left search range 33, that is, when the luminance value is saturated with the maximum value of the gradation, the image element positioned at the center of all the bright points is described with reference to FIG. The left maximum value is 31.

  In step S507 following step S505, a bright spot is searched for in the right search range 34. If a bright spot is detected during the search in step S507, the bright spot number is stored. In step S508 following step S507, it is determined whether a bright point has been detected in step S507. If a bright point is detected, the process proceeds to step S509. If not detected, the process proceeds to step S506. In step S509, if there is only one bright point in the right search range 34, the bright point is set to the right maximum value 32. When there are a plurality of bright points in the right search range 34, that is, when the luminance value is saturated with the maximum value of the gradation, the image element located at the center of all the bright points is described with reference to FIG. And the right maximum value 32. Executing step S509 means that there are local maximum values in both the left search range 33 and the right search range 34. In step S510 subsequent to step S509, branch determination is performed on the condition that the search for all the horizontal scanning lines is completed. If it is determined that the search for all the horizontal scanning lines has been completed, the process returns to step S512 of the luminance determination process 52. If it is determined that the search has not been completed, the process returns to step S501 to start the search using the next horizontal scanning line. . If a bright point cannot be detected in the left search range 33 or the right search range 34, the process moves from step S504 or step S508 to step S506, and it is determined that there is no white strip line.

  Next, the luminance determination process 52 will be described. If it is determined in the preceding maximum value search processing 51 that a maximum value exists, in step S512, a horizontal scanning line for performing luminance determination is determined. Next, in step S513 following step S512, the fixed ranges in the region other than the bright point calculated in the preprocessing are set as the left luminance determination region 37, the middle luminance determination region 38, and the right luminance determination region 39. In step S514 subsequent to step S513, the amount of change in luminance between adjacent image elements in the luminance determination regions 37, 38, and 39 is calculated. In step S515 following step S514, it is determined whether or not the luminance change amount calculated in step S514 is within a certain range. If the luminance change amount is out of a certain range, it is determined that the maximum value detected in the preprocessing, that is, the white belt-like line is a guardrail or a center line, and in step S517, the fog detection signal is not fogged (fog detection signal = 0). If the luminance change amount is within a certain range, the process proceeds to step S516. In step S516, it is identified whether the luminance change amount of all the luminance determination areas, that is, the left luminance determination area 37, the middle luminance determination area 38, and the right luminance determination area 39 has been determined. If the entire area is determined, the process proceeds to step S518. If the entire area is not determined, the process returns to step S514. In step S518 following step S516, if determination of all horizontal scanning lines is not completed, the process returns to step S512, and if determination of all horizontal scanning lines is completed, the process proceeds to step S519. In step S519, the fog detection signal is output as fog generation (fog detection signal = 1).

  In this embodiment, fog was detected by processing the camera image 15 including the optical axis of the headlight 12. This fog detection algorithm generates fog when the optical axis of the headlight 12 appears in the image 15 as a white belt-like line and the brightness gradually decreases as the distance from the center of the white belt increases. It is an algorithm that is considered to be. Since this apparatus automatically turns on the fog light 13 when fog occurs, it is possible to provide driving assistance effective for the driver. In addition, since the white line recognition camera of the traveling lane recognition device installed in many vehicles is shared, the cost performance is excellent, and since the fog detection algorithm based on the white line recognition algorithm is used, the reliability is high.

[Example 2]
Example 2 will be described with reference to FIG. The difference in configuration between the second embodiment and the first embodiment is that an ACC control means 61 is added in this embodiment. In addition, about the structure equivalent to Example 1, the code | symbol similar to Example 1 is attached | subjected, and description in this Example 2 is abbreviate | omitted.

  ACC is cruise control that keeps the distance between the vehicle and the preceding vehicle constant. When the fog detection signal output from the image processing / fog detection unit 17 is fog generation, the ACC control unit 61 opens the inter-vehicle distance longer than the normal inter-vehicle distance, limits the target speed that can be set, or ACC Preventing accidents due to poor visibility by disabling the function itself.

Example 3
Example 3 will be described with reference to FIG. The difference in configuration of the third embodiment from the first embodiment described above is that a brake assist means 71 is added in this embodiment. In addition, about the structure equivalent to the above-mentioned Example, the code | symbol similar to each Example is attached | subjected, and description in this Example 3 is abbreviate | omitted.

  As shown in FIG. 7, in the third embodiment, a brake assist means 71 is added. The brake assist means 71 is set so that the ratio of increasing the driver's pedaling force to the brake pedal in an emergency is higher than that during normal driving, so that even a driver with less pedaling force can perform sufficient braking and braking. It is a means that makes it possible to shorten the distance. The fog detection signal is used as a setting condition of the ratio for increasing the pedaling force. If fog is generated, the ratio of increasing the driver's force on the brake pedal can be increased so that the driver can reliably improve the brake performance even if an obstacle in the fog suddenly appears. It can be pulled out.

Example 4
Example 4 will be described with reference to FIG. The difference in configuration between the fourth embodiment and the first embodiment is that a PCS control means 81 and an obstacle detection means 82 are added in this embodiment. In addition, about the structure equivalent to the above-mentioned Example, the code | symbol similar to each Example is attached | subjected, and description in this Example 4 is abbreviate | omitted.

  PCS (Pre-Crash Safety System) uses obstacle detection means 82 such as millimeter wave radar to detect obstacles ahead, rolls up the seat belt according to the degree of danger, and alerts the occupant. This is means for restraining the occupant to the seat with a belt in order to reduce the impact when the inevitability cannot be avoided. By inputting the fog detection signal to the PCS control means 81, when the obstacle detection means 82 detects an obstacle in the fog, the seat belt can be wound up earlier than usual to alert the occupant.

Example 5
The difference between the fifth embodiment and the previous embodiment is that in the present embodiment, the content of the fog detection signal is four levels of large, medium, small, and none according to the fog density, whereas the above-described implementation is performed. In the example, the fog detection signal is a point that has two stages of fog generation or no fog.

  The stage detection of the fog detection signal in the present embodiment is based on the maximum luminance value described in the first embodiment. Specifically, when the maximum luminance value is very large, the fog density is high, when the luminance value is high, the fog density is high, and when the luminance value is slightly high, the fog density is low. By expressing the fog concentration in detail and combining it with driving support devices such as ACC, brake assist, and PCS, driving support with higher safety is possible. For example, when combined with ACC in the second embodiment, the inter-vehicle distance and the target speed can be adjusted according to the concentration level, so that finer ACC control is possible. Since humans tend to find obstacles later when the fog is darker, in the brake assist of the third embodiment, the ratio of increasing the driver's pedal force on the brake pedal is set according to the fog density. The occupant can perform more effective braking. Due to the same tendency, also in the fourth embodiment, more effective PCS control can be performed by optimizing the timing to call attention according to the degree of density.

[Other Examples]
In the above-described embodiment, the white light recognition camera used for the headlight 12 and the traveling lane recognition device is used.

  The image processing / fog detection unit 17 used in the above-described embodiment may be incorporated in another processing apparatus. For example, the processing of the image processing / fog detection unit 17 may be performed using the ECU of the traveling lane recognition device, or the fog detection algorithm may be incorporated into the traveling lane determination algorithm.

  As for the light, any light other than the headlight 12 may be used as long as the optical axis appears when fog occurs. In the embodiment, the headlight 12 having the AFS function is used. However, the headlight 12, the fog light 13, and the driving light having no AFS function may be used. In addition, the AFS function may be a format in which only the swivel angle is adjusted to widen the irradiation range in the horizontal direction, and a vertical tilt angle can be set.

  In the fog detection algorithm, it is described that the horizontal scanning line for every n pixels on the image is processed. However, the vertical scanning line may be processed for every m pixels, or all image elements on the image may be processed. .

  The fog detection signal has been described using an example of two stages or four stages. However, the fog detection signal may be divided into a plurality of stages of three stages or more, and may be represented by a degree expression by a fuzzy set instead of a step expression. .

It is a block diagram which shows schematic structure of the fog detection apparatus used in Example 1. FIG. It is a camera image in case the fog used in Example 1 has generate | occur | produced. FIG. 6 is a diagram schematically illustrating a method of dividing a horizontal scanning line of a camera image, a luminance distribution of a horizontal scanning line, a maximum value search processing range, and a luminance determination processing region used in the first embodiment. It is a flowchart of the lighting / extinguishing algorithm of the fog light used in Example 1. 3 is a flowchart illustrating processing of an image processing / fog detection unit used in the first embodiment. It is a block diagram which shows schematic structure of the fog detection apparatus provided with ACC used in Example 2. FIG. It is a block diagram which shows schematic structure of the fog detection apparatus provided with the brake assistance means in Example 3. FIG. It is a block diagram which shows schematic structure of the fog detection apparatus provided with PCS used in Example 4. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Camera 12 Headlight 13 Fog light 14 Light control part 15 Image 16 Light state 17 Image processing and fog detection part 31 Left maximum value 32 Right maximum value 33 Left search range 34 Right search range 35 Left maximum value (multiple bright points exist) If you want to)
36 Right maximum (when there are multiple bright spots)
37 left luminance determination area 38 middle luminance determination area 39 right luminance determination area 51 maximum value search processing 52 luminance determination processing 61 ACC control means 71 brake assist means 81 PCS control means 82 obstacle detection means

Claims (8)

In a fog detection device that detects the occurrence of fog when traveling at night,
A first light that illuminates at least the imaging range of the camera;
A camera for photographing the surrounding environment including at least the optical axis of the first light;
Image processing / fog detection means for processing an image including the surrounding environment including at least the optical axis of the first light photographed by the camera and outputting a fog detection signal ;
The image processing / fog detection means has a local maximum value of a luminance value of a horizontal scanning line of a photographed image taken by the camera in the vicinity of an optical axis estimated position of the first light calculated from a light state of the first light. And a region where the luminance change including the maximum value present at the left and right optical axis estimation positions is small, and generation of fog when the luminance change amount between the regions other than the region maintains a change amount within a predetermined range; A fog detection device characterized by being regarded . A light control means for transmitting a control command to the first light and outputting the light state of the first light acquired from the first light to the image processing / fog detection means;
2. The image processing / fog detection means for outputting a fog detection signal based on a light state of the first light input from the light control means and an image taken by the camera. Fog detector. A second light other than the first light is provided,
The light control means acquires the control state of both the first light and the second light by controlling the optical axis direction of the first light and the second light,
Image processing / fog detection means for processing the light state of the first light and the light state of the second light input from the light control means and the image taken by the camera and outputting a fog detection signal. The fog detection device according to claim 2, wherein 4. The fog detection device according to claim 3, wherein the light control unit outputs at least one control command of a swivel angle and a tilt angle as a control command for the first light and the second light. The light control means acquires at least one control state of a switching state between a high beam and a low beam, a swivel angle, and a tilt angle as a light state of both the first light and the second light. The fog detection device according to claim 3. A setting algorithm for the horizontal scanning line of the captured image used when the traveling lane recognition device that recognizes the traveling lane using the captured image captured by the camera detects a coordinate value from the captured image is the image processing / fog The fog detecting apparatus according to claim 1, wherein the detecting means is diverted as an algorithm for detecting the generation of fog. The fog detection apparatus according to claim 6 , wherein the camera is used in combination with a camera used by the travel lane recognition apparatus. The algorithm for converting the coordinate value of the white band-like line in the captured image for detecting the occurrence of fog by the image processing / fog detection means into the coordinate value of the road surface by geometric conversion is used by the traveling lane recognition device. The fog detecting device according to claim 1 , wherein an algorithm for converting a coordinate value of a white line in the photographed image to a coordinate value of a road surface by geometric conversion is diverted .

Images