JPH0351746A - Detecting apparatus for cloudiness of surface of window glass - Google Patents

Abstract

PURPOSE:To improve detecting accuracy by comparing the lightness distribution on the surface of glass with reference distribution. CONSTITUTION:The lightness distribution of luminous flux 2 from window glass 1 is transduced into an electric signal in an image input means 4 and inputted. In a reference-lightness-distribution forming means 4, lightness distribution A when the entire surface of the glass 1 is bright and lightness distribution B when the entire surface is cloudy are obtained based on the signals from the means 3. Reference lightness distribution T is obtained. Lightness distribution Ct at an arbitrary time point (t) which is obtained with the means 3 is compared with the distribution T. The clouding region on the surface of the glass 1 is detected based on the relation of the magnitudes between both distributions. When the results of judgments of the cloudinesses which are detected at the time points (t) and (t + DELTAt) that are very close are both judged as cloudy, the region is judged as the cloudy region. In this way, the evaluation can be performed automatically and quickly, and the accurate detection can be performed.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a device for detecting a fogged area on a windshield surface of an automobile, etc., and is used, for example, to evaluate the performance of a defrosting device such as a defroster. It is related to the device.

[Prior art]

Conventionally, the detection and evaluation of the fogging state of the windshield surface of a vehicle (1), for example, the detection and evaluation of the defroster's defrost removal performance, has generally been visually evaluated or inspected by an inspector or the like.

In addition, as a device for detecting the freezing state or dew condensation state of a part of the window glass, for example,
There is one described in Publication No. 243.

This is 1M! , the light incident on the reflective plate attached to the glass exhibits different reflection characteristics depending on how frozen the window glass is (the amount of reflected light decreases when the glass surface is frozen).
This is used to detect the freezing state of the glass surface.

[Problem to be solved by the invention]

However, visual evaluation by one person has to be a subjective evaluation by the evaluator, so the criteria are vague and the evaluation values vary widely.Furthermore, the evaluator is highly fatigued and the inspection time is short. There are problems such as length.

In addition, in the conventional icing detection device as described above, a sensing part with a special structure is attached to a part of the windshield surface to detect the icing state of that part.
It is not suitable for detecting clear and cloudy areas on the entire windshield surface, or determining the area ratio of clear and cloudy areas, and therefore is not suitable for evaluating the performance of defrosters and other defrosters. The problem was that it could not be applied to experiments.

In order to solve the above-mentioned problems, the applicant has developed a non-contact method by capturing an image of the brightness distribution on the windshield surface of a vehicle, converting it into an electrical signal, and processing the image.

An application has already been filed for a device that automatically and quickly detects fogging on the windshield surface (Japanese Patent Application No. 63-283906).
issue).

However, with the above device, when detecting the reflection from the windshield surface by emitting inspection light from outside the vehicle, if there is a highly bright reflector in the vehicle interior, such as the passenger's face, There was a risk that the area would be mistakenly judged as a cloudy area. Therefore, in defroster fog removal performance tests, etc.
In order to approximate practical conditions, there is a risk that errors may occur in the cloudy area detection results when testing is carried out with an inspector in the vehicle.

The present invention has been made in order to solve the above problems, and is capable of automatically and quickly detecting a fogged area on a windshield installed in a vehicle, and when there is an occupant inside the vehicle. but.

To provide a window glass surface fogging detection device capable of accurately detecting a true cloudy region by distinguishing between reflection caused by the reflection and a true cloudy region.

[Means to solve the problem]

In order to achieve the above object, the present invention has the following features.

The invention is constructed as described in the claims.

Hereinafter, the present invention will be explained in detail based on the functional block diagram shown in FIG.

In FIG. 1, the brightness distribution of a luminous flux 2 reflected or transmitted from a windshield 1 mounted on a vehicle is converted into an electrical signal by an image input means 3 and input.

Note that the illumination at this time is, for example, irradiated from outside the vehicle toward the window glass. The above-described image input means 3 corresponds to, for example, the video camera 102 and analog-to-digital converter 103 shown in FIG. 2, which will be described later.

Next, the reference brightness distribution generation means 4 inputs the signal from the image input means 3, and calculates the brightness distribution A (X + y) when the window glass surface is completely sunny and the brightness distribution A (X + y) when the window glass surface is completely cloudy. Brightness distribution B(x.

y) and from these two brightness distributions, a standard brightness distribution T (x + y) that is used as a criterion for determining whether it is sunny or cloudy.
= K [A (x + y) + B (x r
y ) ) (where 0 x K x 1) is generated. Note that the brightness distribution B(x, y) when the entire window glass surface is clouded is measured with the window glass surface artificially clouded.

As can be seen from the above equation, the reference brightness distribution T(x.

y) is the brightness distribution A (X * y
) and brightness distribution B (x * y
).

The above reference brightness distribution generation means 4 corresponds to, for example, the threshold image generation section 107 and the alignment calculation section 106 shown in FIG. 2, which will be described later.

Next, the cloudy area detection means 5 detects the brightness distribution C of the window glass surface obtained at an arbitrary time t by the image input means 3.
i (x ry ) and the above reference brightness distribution 'r(X,
y) to detect a cloudy area on the window glass surface based on the size relationship between the two, and the fogging determination results detected at two nearby times t and t+ΔL are both determined to be cloudy at the same time. The area is determined to be a cloudy area at the time point.

Usually, the inspector riding the vehicle under test is moving to perform various tasks, so the areas that are determined to be cloudy at two close points as described above are due to reflections from the occupant. Instead, it can be determined that the area is a true cloudy area. Furthermore, since the speed at which fog disappears due to the defroster, etc. is considerably slower than the operating speed of the occupant as described above, by setting the above time interval Δt to an appropriate value (for example, 2 to 3 second intervals), the true It is possible to distinguish between changes in the cloudy area and reflections from the occupant.

This cloudy area detecting means 5 corresponds to, for example, a cloudy area extracting section 110 and an erroneously extracted area removing section 111 shown in FIG. 2, which will be described later.

As mentioned above, in the present invention, the brightness distribution A (K e y ) when the window glass surface is completely sunny and the brightness distribution B (x
r y ) and the reference brightness distribution T (X T y )
The reference brightness distribution T (x y y) and the brightness distribution Ct (
x + y) to detect a cloudy area on the windshield surface, and only areas where detection results at two nearby points in time are both determined to be cloudy are determined to be cloudy areas. ing.

Therefore, in the present invention, evaluations that conventionally relied on visual inspection can be performed automatically and quickly, and since the reference brightness distribution is determined under the current inspection conditions, accurate detection is always possible regardless of lighting conditions, etc. can be done.

Further, even in a test in which the driver rides with equipment in accordance with practical conditions, it is possible to eliminate errors caused by reflections from the occupant and to accurately detect foggy areas.

〔Example〕

FIG. 2 is a diagram showing one embodiment of the present invention, in which the present invention is used as a windshield fogging detection device for evaluating the defroster performance of an automobile.

In FIG. 2, a video camera 102 is installed in front of a window glass 101 at a position that can cover the entire window glass surface.

Further, the analog-to-digital converter 103 converts the signal from the video camera 102 into a digital signal, and the image memories AlO4, BiO2, and D109 are memories that store the signals.

In addition, the alignment calculation unit 106 and the threshold image generation unit 107
, cloudy region extractor 110 and erroneously extracted region remover 111
1 is composed of, for example, a microcomputer. Also,
The result output unit 112 is an image memory group that stores the cloudy extraction results obtained by the cloudy area extracting unit 110 and the erroneously extracted area removing unit 111.

Alternatively, it is a display device such as a CRT.

Next, the effect will be explained.

First, when the entire window glass is clear,
An image (brightness distribution of the windshield surface) taken by the video camera 101 is A/D converted and stored in the image memory AlO4. This is defined as image A (Key). It is assumed that no occupant is in the vehicle in this state. In addition, during this inspection, lighting is normally provided to illuminate the windshield surface from the outside of the vehicle, but the video camera 101 is used with the passenger in the vehicle and with the entire windshield artificially fogged. The photographed image is A/D converted and stored in one image memory B105. This is referred to as image B(x,y).

As mentioned above, image A (Key) is an image obtained without a passenger, and image B (x+y) is an image obtained with a passenger on board, so there may be a positional shift between the two due to the sinking of the vehicle body, etc. It is occurring. Therefore, 1 alignment calculation unit 106
The two are aligned at .

The above positioning process will be described below with reference to FIGS. 3 and 4.

As shown in the processing flow of FIG. 4, first, image A(xt
y) Set area X inside. Specifically, the operator may set it using a digitizer or the like. Hereinafter, according to the flow shown in FIG. 4, a region having a brightness pattern most similar to region X is searched from image B(x+y). This is image A(x,y
) and image B (x.

This is done by calculating a shift amount M that minimizes the inter-region distance of y). As a result, the image shifted by M is defined as A'(xty).

Note that FIG. 3(a) is an example of the above region X in image A (xty), FIG. 3(b) is an example of image B(X, y), and FIG. image A' (x r
y).

It should be noted that if a passenger is on board from the beginning, the above positioning calculation is not necessary.

Next, referring back to FIG. 2 again, the threshold image generation unit 107 will be explained.

If we focus on one point on the window glass when the window glass surface is cloudy or sunny as described above, the brightness of that point is as shown in FIG. 5. Then, when the brightness at point P is approximately halfway between the brightness (2) in a cloudy state and the brightness I2 in a sunny state, it can be determined that the observation point is sunny. Note that Fig. 5 shows a case where the vehicle is illuminated from the outside and the interior color of the cabin is dark, so the brightness is higher in cloudy conditions due to reflection from the windshield surface. When the interior color is bright, I x < I z, and the brightness may be higher on sunny days due to reflection from the interior.

The threshold image generation unit 107 generates an image corresponding to the brightness of point P in FIG.

If this image, that is, the reference brightness distribution is 'r (xty), then T (x, y) = K [A'' (x + y) 1B
(X, y)] (where O<K<1). In the above equation, K should normally be set to 0.5. T (x t y
) is stored in the image memory ClO3. FIG. 6 is a flowchart showing the calculation procedure for obtaining the above reference brightness distribution (threshold image).

Also, at this time, A (x ry ) and B (x, y)
The point where the brightness does not change whether the windshield surface is cloudy or sunny can be considered to be a part of the image other than the windshield surface, such as the body 100, so it is necessary to extract only the windshield surface. can be done at the same time. for example.

As shown in Figure 6, A (x + y) and B (X
A point where the absolute value of the difference from +y) is smaller than a predetermined threshold 'rh is determined to be a part other than the window glass surface, and is set as T(xry) = O so that no further processing is performed.

Next, the entire surface of the windshield is artificially fogged as described above, and then the defroster is activated.

Images are input using the video camera 102 at regular intervals, and these images are stored in the image memory D1o9. This is assumed to be image Ct(xty).

Also, from the generation of c c (x + y), Δin sac
Afterwards, input the image again. Ct□t(xp
y). Note that Δt is set to, for example, 2 to 3 seconds.

Next, the cloudy region extraction unit 110 extracts a cloudy region by comparing the reference brightness distribution T (x + y) with the above Ct (x # y) and Ct + mt (X t 'j), respectively. . Namely.

The point is u 1 n, and the point judged to be sunny is “O”.
By storing the images D t
(x s y ) and image Dt+at(x , 'J
) Generate wo.

Next, the erroneously extracted area removal unit 111 extracts a portion that is erroneously determined to be a cloudy area. That is, if a passenger is riding in the vehicle being measured.

There is a risk that parts with high brightness values, such as the passenger's face and hands, may be mistakenly extracted as cloudy areas. Such erroneous extraction points can be detected by comparing one image Dt(xty) with the image □, (x, y).

In other words, since the operating speed of the occupants inside the vehicle is considerably higher than the speed at which fog disappears by defrosters, etc., two images obtained at an appropriately short time interval Δt (2 to 3 seconds) are compared and one point in time is determined. If a point is determined to be a cloudy area at time t and is not determined to be a cloudy area at +ΔL, it can be determined that the point is a part where the occupant was present at time t and is not a cloudy area. . In other words, in the above judgment, it is judged to be cloudy,
Dt(x, y = 1 and Dt+at(X + 3')
” The points marked O can be judged as incorrectly extracted points.

FIG. 7 is a flowchart showing the calculation procedure for detecting the erroneously extracted portion, and Et(x+y) indicates an image from which the erroneously extracted portion has been removed.

The result output unit 11, such as a CRT or a disk, outputs the image E t (x ly ) showing the true cloudy area from which the erroneously extracted areas have been removed as described above as the cloudy extraction result at time t.
Output to 2.

As described above, cloudy regions can be automatically extracted. Also, by summing the points "1" and "0" in the above image Et(X+y),
It is possible to determine the area of the cloudy region and the area of the clear region, and it is also easy to determine the area ratio of the two.

As mentioned above, each image memory of the result output unit 112 has
Since the fogging state of the windshield surface is sequentially stored at certain time intervals after the defroster is operated from a completely fogged state, the performance of the defroster can be evaluated based on the cloudy state.

FIG. 8 is a flowchart showing the entire calculation process in the embodiment shown in FIG.
2, the state of the window glass is illustrated on the right side of P6 and pH.

Further, FIG. 4 is a flowchart showing details of the portions P3 to P5 (alignment calculation) in FIG. 8, and FIG. 6 is a flowchart showing details of the portion P6 (threshold generation calculation) in FIG. 8. FIG. 7 is a flowchart showing the PLO portion (erroneous extraction portion removal calculation) of FIG. 8.

Although a windshield is illustrated as the window glass 101 in FIG. 2, the invention is not limited thereto, and the present invention can be similarly applied to side windows, rear windows, and the like.

〔Effect of the invention〕

As explained above, in the present invention, the brightness distribution A (x, y) when the window glass surface is completely cloudy and the brightness distribution B (X
t y ) and the reference brightness distribution T(x.

y), and calculate the standard brightness distribution T (x + y) and the brightness distribution c of the window glass surface obtained at an arbitrary time t.
A cloudy area on the windshield surface is detected by comparing With this configuration, evaluations that conventionally relied on visual inspection can be performed automatically and quickly, and since the standard brightness distribution is determined under the current inspection conditions, accurate detection is always possible regardless of lighting conditions. It can be done. In addition, even in a test with a passenger on board, which corresponds to practical conditions, it is possible to eliminate errors caused by reflections from passengers and accurately detect foggy areas.

Excellent effects such as these can be obtained.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the functions of the present invention, Fig. 2 is a block diagram showing the configuration of an embodiment of the invention, Fig. 3 is a diagram showing alignment processing, and Fig. 4 is alignment calculation. An example of a flowchart showing the procedure, FIG. 5 is a characteristic diagram of the reference brightness distribution, FIG. 6 is an example of a flowchart showing the calculation procedure for obtaining a threshold image (reference brightness distribution), and FIG. 7 is an example of an error diagram. FIG. 8 is an example of a flowchart showing the entire calculation procedure in the embodiment of FIG. 2. FIG. <Explanation of symbols> 1... Windshield 2... Luminous flux 3... Image input means 4... Reference brightness distribution generation means 5... Cloudy area detection means

Claims (1)

[Claims] Image input means for converting the brightness distribution of the windshield surface of a vehicle into an electrical signal and inputting the image; and inputting the signal of the image input means to determine the brightness when the windshield surface is completely sunny. The distribution A (x, y) and the brightness distribution B (x, y) when the window glass surface is completely cloudy are determined respectively, and from these two brightness distributions, a reference brightness distribution T is used as a criterion for determining whether it is clear or cloudy. (x, y) = K [A (x, y) + B (x, y)] (where 0<K<1);
The brightness distribution C_t (x, y) of the windshield surface obtained at an arbitrary time t by the image input means is compared with the above reference brightness distribution T (x, y), and based on the magnitude relationship between the two, the windshield surface Cloudy area detection means that detects the cloudy area above and determines the area where the cloudy determination results detected at two adjacent times t and t+Δt are both determined to be cloudy at both times to be the cloudy area at the time. A window glass surface fogging detection device comprising: