JP2001052185A - Method and device for detecting vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a vehicle from image data even when these image data are deteriorated by dirt stuck on the lens surface of an image input device or filter surface protecting a lens. SOLUTION: A fixed reference value t0 is provided for capturing the difference of a luminance value at initial time. At the initial time without dirt, a luminance value a1 at a certain pixel position in the state of picking up the image of the vehicle is stored for each pixel of an entire picture. A value obtained by subtracting the difference between the luminance value a1 at said each pixel position and a luminance value a2 at a relevant pixel position at the time of operation from the fixed reference value t0 is provided for the entire picture, this mean value is calculated and provided as a new reference value t1 and when the difference of a luminance value greater than the new reference value t1 is detected at a certain pixel position, the presence of a change in the relevant pixel is discriminated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for detecting the presence of a vehicle by image processing.

[0002]

2. Description of the Related Art Conventionally, when detecting the presence of a vehicle by image processing, if image data obtained from an image input device is deteriorated by dirt attached to a lens surface of the image input device or a filter surface protecting the lens. In some cases, the amount of information required for vehicle detection cannot be obtained, and the vehicle cannot be detected as a vehicle. In addition, in order to prevent a vehicle detection error when a large amount of dirt adheres, it is necessary to set the dirt cleaning interval relatively frequently and periodically.

[0003]

SUMMARY OF THE INVENTION As described above, according to the present invention, even if the image data is deteriorated by dirt attached to the lens surface of the image input device or the filter surface for protecting the lens, the vehicle can be reconstructed from the image data. And a device therefor.

[0004]

In order to solve the above-mentioned problems, a vehicle detection method according to the present invention obtains image data by an image input device and sets a luminance value larger than a reference value between the image data and the reference image data. In performing the vehicle detection process when there is a difference, the reference value is changed according to the luminance value of the image data during operation.

[0005] At this time, the difference in the luminance value is detected at the initial stage.
Fixed reference value t for₀The lens surface of the image input device
Or at the beginning when the filter surface protecting the lens is clean.
The brightness value at a certain pixel position in a state where the vehicle is imaged
a₁Is stored for each pixel of the entire screen, and the
Brightness value a₁And the luminance value a at the pixel position during operation
₂With the fixed reference value t₀Value obtained by subtracting
For all screens, calculate the average value,
Reference value t₁At a certain pixel position.
Reference value t₁When a larger difference in luminance value is detected,
It is determined that the pixel has changed.

A vehicle detection device according to the present invention acquires image data by an image input device and performs a vehicle detection process when there is a difference in luminance value between the image data and the reference image data that is larger than a reference value. the vehicle detection apparatus of the image processing type, a fixed reference value t ₀ for capturing the difference between the luminance values at the initial time, the initial time in a clean the filter surface to protect the lens surface or lens of the image input device and image the vehicle a reference pixel luminance value storage unit that stores a luminance value a ₁ for each pixel of the entire screen at a certain pixel position of a state, the luminance value a ₂ of the pixel position at the time of operation and the luminance value a ₁ at each pixel position the difference of the obtained value obtained by subtracting from a fixed reference value t ₀ for full-screen, further a new reference value calculating section obtained by a new reference value t ₁ calculates the average value, at a certain pixel position Vehicle detected based serial and new reference value t ₁ what was the pixel is changed when more detects the large difference in luminance value between determining the pixel change detecting unit, the pixel change detection judgment of the pixel change detecting unit And an image processing unit for performing the above image processing.

[0007]

The operation of the vehicle detection method is as follows. The first step, is previously provided with a fixed reference value t ₀ for capturing the difference between the luminance values at the initial time. The initial time to a clean filter surface to protect the lens surface or lens of the image input device, and stores the luminance value a ₁ in a certain pixel position of a state of the captured vehicle for each pixel of the entire screen. Then, the difference between the luminance value a ₂ of the pixel position at the time of operation and the luminance value a ₁ in each pixel position to obtain the values for the entire screen is subtracted from the fixed reference value t _0. Then, an average value is calculated for all the screens to obtain a new reference value t _1.
Get as. Next, a new reference value t _{1 is} set for each pixel.
When a larger difference in luminance value is detected, it is determined that the pixel has changed.

The operation of the vehicle detection device is as follows. Fixed reference value t ₀ is intended to capture the difference between the luminance values at the initial time. Reference pixel luminance value storage unit, the initial time in a clean the filter surface to protect the lens surface or lens of the image input apparatus, a luminance value a ₁ in a certain pixel position of a state of the captured vehicle for each pixel of the entire screen Remember. New reference value calculation unit, to obtain the difference between the luminance value a ₂ of the pixel position at the time of operation and the luminance value a ₁ at each pixel position, a value obtained by subtracting from the fixed reference value t ₀ for full-screen , obtained by a new reference value t ₁ and further calculates the average value. Pixel change detecting unit determines that a change occurs to the pixel when it detects a large difference in luminance value than the new reference value t ₁ at a certain pixel position. The image processing unit performs image processing for vehicle detection based on the pixel change detection determination of the pixel change detection unit.

[0009]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. FIG. 1 is a flowchart illustrating one embodiment of a vehicle detection method according to the present invention. FIG. 2 is a functional block diagram illustrating an embodiment of the vehicle detection device according to the present invention. In FIG. 2, reference numeral 21 denotes an image input device constituted by a TV camera, for example. 22 is a fixed reference value storing unit for holding data of a fixed reference value t _0. 23 is a reference luminance value storage unit,
Initial time to a clean the filter surface to protect the lens surface or lens of the image input apparatus 21, and stores the full screen luminance value a ₀ at the pixel position with a state the vehicle is not.
Reference numeral 24 denotes a reference pixel luminance value storage unit.
The initial time without stains on the filter surface to protect the lens surface or lens, and stores the full screen luminance value a ₁ in a certain pixel position of a state of the captured vehicle. 25 is a new reference value calculating unit, the difference between the luminance value a ₂ of the pixel position at the time of operation and the luminance value a ₁ in each pixel position, from a fixed reference value t ₀ of the fixed reference value storing section 22 It subtracted obtained for the entire screen values obtained, obtained by a new reference value t ₁ by calculating the average value. 26 is a pixel change detecting unit determines that the to the pixel has been changed when it detects a large difference in luminance values than the new reference value t ₁ for each pixel. An image processing unit 27 performs image processing for vehicle detection based on the pixel change detection determination of the pixel change detection unit 26. A control unit 28 controls the entire apparatus.

FIG. 3 is a diagram for explaining an example of an installation state of the vehicle detection device according to the present invention. In FIG.
Reference numeral 21 denotes an image input device. Reference numeral 31 denotes a vehicle detection processing unit which includes a fixed reference value storage unit 22, a new reference value calculation unit 25, a pixel change detection unit 26, an image processing unit 27, and a control unit 28 in FIG. 32 is a vehicle running from left to right in the figure, 3
Reference numeral 3 denotes an imaging range of the image input device 21, and reference numeral 34 denotes a road.

FIG. 4 shows an image from the image input device 21. When no dirt is attached to an image pickup surface such as a lens surface of the image input device 21 or a filter surface for protecting the lens, the vehicle 32 on the road 34 is moved. An image. 5 is an image in the case where light dirt (indicated by a group of diagonal lines falling to the right) is attached to the lens surface or the entire filter surface protecting the lens in FIG. 4, and FIG. 6 is an image in the case where no vehicle is present in FIG. It is.

In FIG. 3, a vehicle 32 running on a road is shown.
Are picked up by the image input device 21 and processed by the vehicle detection processing unit 31. The image is composed of luminance values for each pixel.

Normally, in an apparatus for detecting the entry of a vehicle by image processing, the difference between the luminance value of each pixel of the image when the vehicle is not present and the pixel when the vehicle is present is used as information for vehicle detection. Now, FIG. 6, that is, the luminance value of an arbitrary pixel p ₀ in the initial absence vehicle is clean the filter surface to protect the lens surface or lens of the image input device 21 _(x, y) to a _0. The pixel p ₁ in FIG. 4 when no dirt is attached and the vehicle 32 is present.
The luminance value at the same position as (x, y) is a ₁ , and the pixel p ₂ (x,
If the luminance value of y) was a _2, the relationship between the luminance value due to the influence of dirt,

[0014]

A ₁ > a ₂ (1)

## EQU1 ## Therefore, if we decide to capture the difference in brightness over a certain level as information necessary for vehicle detection,
When the threshold value of the difference between t _0,

[0016]

| A ₁ −a ₀ |> t ₀ (2)

Or

[0018]

| A ₂ −a ₀ |> t ₀ (3)

It is necessary that the following condition is satisfied. That is, defining the threshold t ₀ for the luminance value a ₀ for any pixel p _{0 (x,} y) in the initial absence vehicle is clean the filter surface to protect the lens surface or lens of the image input device 21. When the difference between the luminance value exceeds the threshold value t _0, as the difference in meaning to the pixel is generated, the process of vehicle detection. However, due to the lapse of time during operation, the magnitude of the difference between a ₂ and a ₀ is reduced by the threshold value t _{0 due} to contamination on the lens surface of the image input device 21 or the filter surface protecting the lens.
, And information necessary for vehicle detection may not be obtained.

Therefore, the following is performed to determine a new threshold value. The image input device 21 starts the road 3 in the initial stage when the lens surface or the filter surface protecting the lens is clean.
4 is imaged. At this time, no vehicle exists on the road 34 as shown in FIG. Fixed reference value storing unit 22 and holds the threshold value _{t 0} (Figure 1: P1). Further, the image input device 21 captures an image in a state where the vehicle 32 exists in advance as shown in FIG. 4 at an initial stage when there is no dirt on the lens surface or the filter surface for protecting the lens. Reference pixel luminance value storage unit 24, in a state in which the vehicle 32 is present, the luminance value a ₁ of each pixel position of image data stored for the whole screen (Figure 1: P2).

Next, the image input device 21 constantly captures an image of the imaging range 33 during operation. Here, in consideration of the contamination due to the lapse of time accompanying the operation of the image input device 21,
If it decides to capture the difference above a certain luminance as information required for detection of the vehicle, when the threshold value of the difference between t _1,

[0022]

| A ₂ −a ₀ |> t ₁ (4)

However,

[0024]

T ₁ = t ₀ − (a ₁ −a ₂ ) (5)

It is desirable that Therefore, FIG.
For the same vehicle at the beginning, which is clean,₂= A₁Is
From t₁= T₀And the fixed reference value t₀Is clean
The reference value (threshold) required to obtain vehicle detection information at the initial stage
Become. Therefore, the new reference value calculation unit 25 calculates the reference pixel luminance value
Brightness value a at each pixel position in storage unit 24₁And image input device 2
Luminance value a at the pixel position of the image data of No. 1₂With
The difference is obtained, and this value is further fixed to a fixed reference value t.₀Deducted from
At all pixel positions on the screen, and then calculate the average
And calculate this as the new reference value t₁(Threshold t₁)
(FIG. 1: P3). Brightness value a₂The image to get
Although the vehicle 32 is shown, the vehicle is always in the imaging range.
There is no guarantee that there will be a
It is more soiled. This stain is a time lapse
The brightness value a ₂Is the degree of dirt
(E.g., become smaller) and the new reference value t
₁Is the luminance value a of the image data₂Changes according to
If the luminance value a₂Is smaller as is smaller)
Becomes The pixel change detection unit 26 receives a signal from the image input device 21
Brightness value a at each pixel position of currently input image data₂And base
The luminance value a at the same pixel position in the quasi-luminance value storage unit 23₀The difference between
Whether the absolute value satisfies equation (4) is determined by a new reference value t.₁
(FIG. 1: P4). Pixel change detector 26
Is | a₂-A₀| Is t₁Current image when greater than
It is determined that a significant difference has occurred in the pixel position of the data.
Set. For example, luminance value a₂Is small due to deterioration due to dirt
When it comes | a ₂-A₀Is also small, but t₁Also luminance value
a₂Does it get smaller as it gets smaller?
Then, the difference is detected. The image processing unit 27
The difference between pixel positions equal to or greater than a predetermined value
Is determined (FIG. 1: P5), it is indicated by the pixel position.
The envelope and calculate its size
(FIG. 1: P6). The result shows that the envelope and its
When the size is within a predetermined vehicle determination range,
The vehicle can be detected.

[0026]

As described above in detail, according to the present invention, even if the image data is deteriorated due to dirt attached to the lens surface of the image input device or the filter surface protecting the lens, the data of the data can be obtained. Since the reference value for determining whether there is a difference in the brightness values is constantly changed according to the brightness value of the image data during operation, even if a certain amount of dirt adheres, the vehicle can be detected and the dirt can be cleaned. The intervals can be set at relatively long intervals.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating one embodiment of a vehicle detection method according to the present invention.

FIG. 2 is a functional block diagram illustrating an embodiment of a vehicle detection device according to the present invention.

FIG. 3 is a diagram illustrating an example of an installation state of a vehicle detection device according to the present invention.

FIG. 4 is an image in an initial state where no dirt is attached to an imaging surface.

FIG. 5 is an image in a case where light stains are attached to the entire imaging surface in an initial state.

FIG. 6 is an image in an initial state where no dirt is attached to the imaging surface and no vehicle exists.

[Explanation of symbols]

21: image input device (TV camera), 22: fixed reference value storage unit, 23: reference luminance value storage unit, 24: reference pixel luminance value storage unit, 25: new reference value calculation unit, 26: pixel change detection unit 27: image processing unit, 28: control unit, 31: vehicle detection processing unit, 32: vehicle, 33: imaging range, 34: road.

 ──────────────────────────────────────────────────続 き Continued on front page F term (reference) 2F065 AA67 BB15 CC11 DD13 EE03 FF01 FF42 FF61 JJ03 JJ09 JJ26 KK02 QQ03 QQ24 QQ25 QQ28 QQ42 RR07 RR08 TT04 UU05 2F112 AD10 BA06 CA05 FA19 FA35 5H180 AA01 CC01A01 CC DA02 EA05 FA32 FA46 FA54 GA08 GA41 GA51 HA03 JA09

Claims (3)

[Claims] 1. An image processing type vehicle detection method comprising: acquiring image data by an image input device; and performing a vehicle detection process when there is a difference in luminance value between the image data and reference image data that is larger than a reference value. A vehicle detection method characterized by changing a reference value according to a luminance value of image data during operation. 2. Image data is acquired by an image input device.
And the difference between the image data and the reference image data is larger than a reference value.
Image processing that performs vehicle detection processing when there is a
In a logical vehicle detection method, a fixed reference value t for capturing a difference in luminance value at an initial stage.₀To
Filter for protecting the lens surface or lens of the image input device
An image of a vehicle taken at an initial time when the surface is clean
Luminance value a at elementary position₁Is stored for each pixel of the entire screen, and the luminance value a at each pixel position is₁And during operation
The luminance value a at the pixel position ₂With the fixed reference value t₀
Is obtained for all screens, and
The average value is calculated and a new reference value t is calculated.₁At a certain pixel position, the new reference value t₁Bigger
When the difference between the brightness values is detected, the pixel
And a vehicle detection method. 3. Acquisition of image data by an image input device
And the difference between the image data and the reference image data is larger than a reference value.
Image processing that performs vehicle detection processing when there is a
In a logical vehicle detection device, a fixed reference value t for capturing a difference in luminance value at an initial stage.
₀And a filter for protecting the lens surface or the lens of the image input device.
An image of a vehicle taken at an initial time when the surface is clean
Luminance value a at elementary position₁Is stored for each pixel of the entire screen
A reference pixel luminance value storage unit, and a luminance value a at each pixel position₁And during operation
The luminance value a at the pixel position ₂With the fixed reference value t₀
Is obtained for all screens, and
The average value is calculated and a new reference value t is calculated.₁New reference value meter
And a new reference value t at a certain pixel position₁Bigger
When the difference between the brightness values is detected, the pixel
A pixel change detection unit that determines that
And an image processing unit for performing image processing for detection.
Vehicle detection device.