JP2005117262A - Method and apparatus for discriminating dirt of lens - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and apparatus for accurately discriminating dirt of a lens in an imaging apparatus used for an image processing apparatus for processing an unspecified image and capable of quickly and efficiently providing maintenance to a required part. <P>SOLUTION: Zoom-out control is applied to the imaging apparatus, and an input image from the imaging apparatus is binarized to convert the image into a black region and white region. When the black region having the number of pixels equal to or larger than those of the dirt region and has scenes equal to or larger than a specified number exist in the same region in the input image, it is discriminated that there is dirt on the lens of the imaging apparatus. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a lens dirt determination method and apparatus, and more particularly, to a method and apparatus for determining lens dirt of a camera used in an imaging apparatus.

  An imaging device is used in an image processing device or the like for detecting and recognizing a specific character or pattern from a moving body such as a vehicle. The camera lens or its front panel dirt (hereinafter collectively referred to as a lens) in the imaging device. ) Or the like may cause a reduction in image processing capability.

  The following is a conventional method and apparatus for determining dirt on a lens in such an imaging apparatus.

  (1) In coating film defect detection, the previous image captured from the CCD camera is compared with the image captured this time, and if there is a defect at the same position in both images, it is determined that the lens is dirty ( For example, see Patent Document 1.)

(2) At the time of vehicle detection, an image captured with a lens with no problem such as dirt is stored in advance, the brightness difference with the image captured during operation is compared for the same area, and when there is a large brightness difference, What is determined to be dirty (for example, see Patent Document 2).
Japanese Patent Laid-Open No. 6-148083 JP-A-2001-52185

  However, in such a prior art, when performing lens dirt determination using an image that becomes unspecified or unstable due to changes in the environment, such as outdoors, there is an object in the background or lens dirt. Since it is difficult to accurately determine whether or not it is, it is necessary for a human to perform periodic inspection and maintenance, and accordingly, a great number of inspection and maintenance man-hours are required. In addition, depending on the interval of periodic inspections of the entire system, the system may be in a state where it is operating with reduced performance.

  Accordingly, an object of the present invention is to accurately determine lens contamination in an imaging apparatus used in an image processing apparatus that processes an unspecified image so that necessary portions can be maintained quickly and efficiently.

  In order to achieve the above object, a method for determining contamination of a lens according to the present invention includes a first step of performing zoom-out control of an image pickup apparatus, and binarizing an input image from the image pickup apparatus into a black area and a white area. A second step of converting, and a third step of determining that the lens of the imaging device is contaminated when the black region is present in the same region in the input image by a number equal to or greater than the number of smudged region pixels. It is characterized by having.

  That is, according to the present invention, the contamination of the camera lens (or its front panel) in the imaging device is determined using a plurality of images taken in the process of zooming out the imaging device (hereinafter sometimes referred to as zoom-out images). It is something to try. This concept will be described below with reference to FIGS.

  Now, as shown in FIG. 1 (1), when the imaging apparatus 1 images the front objects 5 and 6, an image in a normal state where the lens of the imaging apparatus 1 is not contaminated is shown in FIG. ) And (3).

  In particular, FIG. (2) shows an image when the imaging apparatus 1 zooms in on the objects 5 and 6, and thus the objects 5 and 6 are enlarged and appear close to each other. In the zoom-out image shown in FIG. 5, the objects 5 and 6 are reduced and appear far away.

  In this case, in the zoomed-out image shown in (3) of FIG. 3, the depth of field becomes deep (the focal distance is long) to widen (zoom out) the angle of view of the lens of the imaging device 1, and the vicinity of the lens Focus on dirt.

  That is, as shown in FIG. 2 (1), when the imaging device 1 captures an object 6, when zooming in as shown in FIG. 2 (2), it is close to the imaging surface like dirt 7 near the lens. The image will be out of focus and blurred (shown as ash area in the drawing), but when zoomed out as shown in (3) in the figure, the dirt 7 near the lens will be in focus. 7 is a clearly visible image (shown as a black area in the drawing). In this zoomed-out image, an image having the same number of pixels as the dirty area is obtained in a plurality of sheets.

  Therefore, in the lens dirt determination method according to the present invention, as described above, the zoom-out control of the imaging device is performed in the first step, and the zoom-out image input from the imaging device is binarized in the second step to obtain a black region. In the third step, if the above black area is present in the same area in the input image more than the specified number of pixels in the third step, the lens of the imaging device is dirty. It becomes possible to judge.

  In the first step, pan control (or tilt control) may be performed instead of zoom-out control. This will be described with reference to FIGS.

  When the imaging apparatus 1 images the objects 5 and 6 in the state shown in FIG. 3 (1), if the lens is not contaminated, the objects 5 and 6 are panned before panning as shown in FIG. Is located in the right half, but after panning, the image moves to the left half of the screen as shown in FIG.

  On the other hand, when the imaging device 1 images the object 6 as shown in FIG. 4 (1), if dirt 7 is attached to the lens, in the image before pan shown in FIG. Since the black area of the dirt 7 shown in the figure shows that the object 6 has moved to the left in the image after panning shown in FIG. If this black area is present in the same area of the input image as the number of pixels in the dirty area and the specified number or more, it can be determined that the lens is dirty.

  If it is determined in the third step that the lens is dirty, the lens can be cleaned in the fourth step.

  Furthermore, the image processing apparatus 2 can notify the center apparatus 3 of the determination result in the third step in the fifth step, for example, via the network line 4.

  Further, if it is determined that the lens is still dirty in the third step even after the first to fourth steps are repeated a predetermined number of times, it may be determined that the lens is abnormal in the sixth step.

  On the other hand, when carrying out the dirt check of the lens at night, if the LED for nighttime (light emitting diode) is directly applied to the lens, a black object will shine white due to strong light hitting nearby objects. On the other hand, the stain on the lens determined to be a black area becomes a white area.

  By using this property, in the method for determining contamination of a lens according to the present invention, after the nighttime LED illumination is directly applied to the lens of the imaging device, the input image from the imaging device is binarized in the same manner as described above to obtain a black region. When the white area is converted into a white area and the above white area is present in the same area of the input image as the number of smudge area pixels or more and the specified number or more, it can be determined that the lens is smudged.

  As an apparatus for realizing the lens dirt determination method, the imaging apparatus and the imaging apparatus are zoomed out, the input image from the imaging apparatus is binarized and converted into a black area and a white area, The image processing apparatus can be configured to determine that the lens of the imaging apparatus is dirty when the black area exists in the same area in the input image with the number of pixels of the dirty area being equal to or greater than the specified number.

  This image processing apparatus can also perform pan or tilt control instead of zoom-out control.

  Further, when the image processing apparatus determines that the lens is contaminated, a means for removing the lens dirt may be further provided.

  The image processing apparatus may further include means for notifying a determination result to the center apparatus.

  Furthermore, when the above-described image processing apparatus repeats the process a predetermined number of times and determines that the lens is still dirty, a means for determining that the lens is abnormal may be further provided.

  Further, according to the present invention, the image pickup device, the night LED illumination, and the lens of the image pickup device are directly applied, the input image from the image pickup device is binarized and converted into a black region and a white region, and the white region is a dirty region. It is also possible to provide a lens dirt determination device including an image processing device that determines that a lens is dirty when the number of pixels is equal to or greater than a specified number and is present in the same area of the input image.

  In the present invention, the following effects can be obtained when determining lens contamination.

・ Accurate stain detection even when processing unspecified images such as outdoor images with large environmental changes.

-By automatically operating the wiper when determining lens dirt, dirt attached to the lens can be automatically removed.

-By automatically notifying the user of lens abnormality via the center unit, it can be dealt with quickly without leaving the system performance degraded, so the system can be maintained in high quality.

・ Automatic detection of lens dirt and abnormalities eliminates the need for regular maintenance and inspection by humans, enabling efficient maintenance.

  FIG. 5 shows an embodiment of an apparatus for realizing the lens dirt determination method according to the present invention. In this embodiment, the imaging device 1 is connected to the image processing device 2 as shown in FIGS. 1 (1) to 4 (1). The image processing device 2 is connected via the network line 4. Are connected to the center device 3.

  The imaging apparatus 1 includes a camera 11, a wiper 12 for removing dirt on a lens (not shown) of the camera 11, and an LED 13 for night illumination of the camera 11.

  In addition, the image processing device 2 includes an image capturing unit 21 that captures the image signal T1 from the imaging device 1, and a plurality of images captured by the image capturing unit 21 in an area that is equal to or less than a luminance threshold (hereinafter, black region). A region dividing unit 22 that divides the region into luminance regions or more (hereinafter referred to as white regions), a calculation image storage unit 23 that stores the images divided by the region dividing unit 22, and the calculation image Based on the image stored in the storage unit 23, the contamination determination unit 24 for performing lens contamination determination, the image captured by the image capture unit 21, and the control signal T2 is generated based on the determination result by the contamination determination unit 24 and imaged. An image pickup apparatus control unit 25 that controls the apparatus 1 is configured.

  The center device 3 also receives a notification capturing unit 31 that captures the determination result notification T3 sent from the dirt determination unit 24 via the network line 4, and a notification that displays the notification captured by the notification capturing unit 31. And a display unit 32.

  FIG. 6 is a flowchart showing the operation of the lens dirt determination apparatus according to the present invention shown in FIG. Hereinafter, the operation of the embodiment of FIG. 5 will be described with reference to the flowchart of FIG.

  First, the imaging device control unit 25 of the image processing device 2 starts control of the camera 11 of the imaging device 1 (step S1 in FIG. 6). In this case, zoom-out control may be performed as shown in FIG. 2 (3), or pan control (or tilt control) may be performed as shown in FIG. 4 (3).

  Thereafter, the image processing device 2 captures the zoomed-out image signal T1 from the camera 11 in the image capturing unit 21 (step S2).

  Then, the area dividing unit 22 performs binarization conversion of the image signal T1 and divides the image signal T1 into a black area and a white area (step S3). As shown in FIG. 7, when the binarization conversion is performed based on the threshold value of the luminance of the blacked out portion in the actual zoom-out image of FIG. A binarized converted image (black / white region divided image) set to “0” is obtained as shown in FIG.

  The converted image of FIG. 7 (2) obtained in this way is stored in the calculated image storage unit 23 (step S4). Then, the zoom-out image is stored for a predetermined number of times (for example, the number of sampling times) (step S5).

  In this way, after storing the binarized converted image for the predetermined number of times stored in the arithmetic image storage unit 23, the imaging device control unit 25 stops controlling the camera 11 (step S6), and the state before the camera control (Step S7).

  That is, since zoom-out or pan / tilt control is performed in step S1, it is necessary to return to the normal state before the start of camera control.

  Thereafter, it is determined whether or not a dirty area exists (step S8). First, set the numerical values as follows.

・ Image save count: 5 times ・ Dirt judgment count: 4 sheets ・ Wiper operation count: 5 times ・ Lens abnormality judgment count: 3 times ・ Dirt area judgment pixel count: 3 areas FIG. 8 is stored in the computation image storage unit 23 FIG. 7B shows the binary conversion image of FIG. 7 (2) as shown above and partially shown. In this way, among the converted images stored in the five images, the lens of the imaging device is contaminated, so that a black region whose luminance value is lower than the threshold value (“1” region) is indicated by a bold line in FIG. 8, for example. If there is more than a certain number of the same area (stained judgment number “4” above) and the size of the black area is more than a certain value (stained pixel judgment pixel number “3” above), there is dirt and dirt. The determination unit 24 makes the determination.

  As described above, when it is determined that there is dirt, if the determination result of “dirty” does not exceed the specified number of times (lens abnormality determination number “3”) (step S9), the imaging device control unit 25 Activates the wiper 12 (step S10), and removes dirt on the lens. However, if the specified number of times is exceeded (step S9), dirt or scratches cannot be removed. The contamination determination unit 24 determines (step S11 in the same), and sends a lens abnormality notification as a determination result notification T3 to the center device 3, for example.

  In the example of FIG. 8, there are 6 identical black areas of the binarized value “1”, and they exist over 4 sheets. Therefore, it is determined that the lens is dirty, and the wiper is operated five times to remove the dirt. If it is determined that “dirt” is detected even if this process is performed three times continuously, it is determined as “lens abnormality” and an alarm is given to the center device 3 via the network line 4.

In the center device 3, this lens abnormality notification can be sent from the notification capturing unit 31 to the notification display unit 32 to notify the user of the lens abnormality, for example.
(Appendix 1)
A first step of zooming out the imaging device;
A second step of binarizing the input image from the imaging device and converting it into a black region and a white region;
A third step of determining that there is dirt on the lens of the imaging device when the black area is present in the same area in the input image with a specified number of dirt areas or more and a specified number or more;
A method for determining contamination of a lens, comprising:
(Appendix 2) In Appendix 1,
A method for determining contamination of a lens, wherein panning or tilting control is performed instead of zooming out control in the first step.
(Appendix 3) In Appendix 1,
A lens dirt judging method, further comprising a fourth step of removing the lens dirt when it is judged that the lens dirt is found in the third step.
(Appendix 4) In Appendix 1,
A method for determining contamination of a lens, further comprising a fifth step of notifying the center device of a determination result in the third step.
(Appendix 5) In Appendix 3,
The lens dirt, further comprising a sixth step of determining that the lens is abnormal when it is determined that the lens is still dirty in the third step after repeating the first to fourth steps a predetermined number of times. Judgment method.
(Appendix 6)
Directing night LED lighting directly onto the lens of the imaging device;
Binarizing an input image from the imaging device and converting it into a black region and a white region;
A step of determining that the lens is soiled when the white region is present in the same region of the input image as the number of pixels in the soiled region is greater than or equal to the number of pixels in the soiled region;
A method for determining contamination of a lens, comprising:
(Appendix 7) In any one of Appendices 1 to 6,
A lens dirt determination method, wherein the lens includes a front panel thereof.
(Appendix 8)
An imaging device;
Zoom-out control of the imaging device, binarization of the input image from the imaging device and conversion into a black region and a white region, and the black region is more than the number of dirty region pixels and more than a specified number of pixels in the input image An image processing device that determines that there is dirt on the lens of the imaging device, if present in the same region;
An apparatus for determining contamination of a lens, comprising:
(Appendix 9) In Appendix 8,
A lens dirt determination device, wherein the image processing device performs pan or tilt control instead of zoom-out control.
(Appendix 10) In Appendix 8,
A lens dirt determination apparatus, further comprising means for removing the lens dirt when the image processing apparatus judges that the lens is dirty.
(Appendix 11) In Appendix 8,
A lens dirt determination apparatus, wherein the image processing apparatus further comprises means for notifying a determination result to a center apparatus.
(Appendix 12) In Appendix 10,
The image processing apparatus further comprises means for determining that the lens is abnormal when it is determined that the lens is still dirty even if the processing until the lens is removed is repeated a predetermined number of times. Judgment device.
(Appendix 13)
An imaging device;
Nighttime LED lighting,
Directly apply to the lens of the image pickup device, binarize the input image from the image pickup device and convert it into a black region and a white region, An image processing device that determines that there is dirt on the lens if present in the same region;
An apparatus for determining contamination of a lens, comprising:
(Supplementary note 14) In any one of Supplementary notes 8 to 13,
A lens dirt judging device, wherein the lens includes a front panel thereof.

It is the figure which showed the zoom-in image and zoom-out image in case a lens stain | pollution | contamination does not adhere when an imaging device uses a zoom function. It is the figure which showed the zoom-in image and zoom-out image when lens dirt has adhered when an imaging device uses a zoom function. It is the figure which showed the zoom-in image and zoom-out image in case a lens stain | pollution | contamination does not adhere when an imaging device uses a pan / tilt function. It is the figure which showed the zoom-in image and zoom-out image when lens dirt has adhered when an imaging device uses a pan / tilt function. It is the block diagram which showed one Example of the apparatus for implement | achieving the stain | pollution | contamination determination method of the lens which concerns on this invention. FIG. 6 is a flowchart showing an operation example of the image processing apparatus shown in FIG. It is the figure which showed the example which converts the zoomed-out image imaged with the imaging device into a binarization conversion image. It is the figure which illustrated a part of image at the time of the dirt adhesion preserve | saved by binarization conversion.

Explanation of symbols

1 Imaging device
2 Image processing device
3 Center equipment
4 Network line
5,6 object
11 Camera
12 Wiper
13 Night lighting LED
21 Image capture unit
22 Image segmentation
23 Calculation image storage
24 Dirt judgment part
25 Imaging device controller
31 Numerical data acquisition
32 Numerical display section In the figure, the same reference numerals indicate the same or corresponding parts.

Claims (5)

A first step of zooming out the imaging device;
A second step of binarizing the input image from the imaging device and converting it into a black region and a white region;
A third step of determining that there is dirt on the lens of the imaging device when the black area is present in the same area in the input image with a specified number of dirt areas or more and a specified number or more;
A method for determining contamination of a lens, comprising: In claim 1,
A method for determining contamination of a lens, wherein panning or tilting control is performed instead of zooming out control in the first step. Directing night LED lighting directly onto the lens of the imaging device;
Binarizing an input image from the imaging device and converting it into a black region and a white region;
A step of determining that the lens is soiled when the white region is present in the same region of the input image as the number of pixels in the soiled region is greater than or equal to the number of pixels in the soiled region;
A method for determining contamination of a lens, comprising: An imaging device;
Zoom-out control of the imaging device, binarization of the input image from the imaging device and conversion into a black region and a white region, and the black region is more than the number of dirty region pixels and more than a specified number of pixels in the input image An image processing device that determines that there is dirt on the lens of the imaging device, if present in the same region;
An apparatus for determining contamination of a lens, comprising: An imaging device;
Nighttime LED lighting,
Directly apply to the lens of the image pickup device, binarize the input image from the image pickup device and convert it into a black region and a white region, An image processing device that determines that there is dirt on the lens if present in the same region;
An apparatus for determining contamination of a lens, comprising:

Images