JP2009063398A - Apparatus and method for detecting oil film - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect the presence of the formation of an oil film by removing the brightness variation caused by waves and extracting only the brightness variation caused by the oil film regardless of whether or not a rainbow interference pattern is formed by an oil film on the water surface. <P>SOLUTION: The apparatus comprises: an imaging device (4) for taking an image of the water surface; and an image processing unit (5) for processing the image signal. The image processing unit (5) includes: a Fourier transform means (M<SB>1</SB>) for outputting a Fourier signal (FS<SB>1</SB>) based on a frequency component included in the image signal (IMG<SB>1</SB>) inputted from the imaging device (4); a high-pass filter (M<SB>2</SB>) for passing a frequency component that is included in the Fourier signal and higher than the brightness variation caused by waves on the water surface; an inverse Fourier transform means (M<SB>3</SB>) for performing an inverse Fourier transform on a Fourier signal (FS<SB>2</SB>) passing through the high-pass filter (M<SB>2</SB>) to return it to an image signal (IMG<SB>3</SB>); and an oil film extraction means (M<SB>4</SB>) for extracting an oil film portion by binarizing the image signal (IMG<SB>3</SB>) subjected to the inverse-Fourier transform. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an oil film detection device that detects the presence or absence of an oil film formed on a water surface.

In recent years, there have been frequent water quality accidents in which oil and harmful substances flow into rivers. In particular, oil spill accidents account for nearly 80% of water quality accidents in rivers, which are raw water supply, and cause serious damage to water treatment plants and downstream ecosystems.
Oil diffuses easily, and the diffused oil becomes an oil film and spreads several kilometers downstream of the river, so it has a great impact on the water environment.
Therefore, an oil film detection system for rivers that can be continuously monitored for 24 hours with high sensitivity is desired.

For this reason, various oil film detection devices have been proposed.
For example, a fluorescence analysis method (Patent Document 1) that irradiates the water surface with excitation light having a wavelength of 200 to 300 nm, selectively receives fluorescence having a wavelength of 300 to 400 nm by the oil film, and determines the presence or absence of the oil film based on the received light intensity (Patent Document 1), water The reflectance measurement method using the fact that the reflectance of the oil film becomes higher than that of the above, the brightness change detection method (Patent Document 2) for picking up the water surface as an image and detecting the brightness change as an oil film, the light beam from the laser light source There is known a polarization analysis method (Patent Document 3) or the like that discriminates the presence or absence of an oil film by performing two-dimensional scanning or diffusing in a planar shape to irradiate the surface of water and performing polarization analysis of the reflected light.
JP 2002-214140 A Japanese Patent Laid-Open No. 08-292185 JP-A-11-326188

However, the fluorescence analysis method is not only difficult to detect in a bright environment where disturbance light such as sunlight and illumination light is present in addition to the excitation light, but also requires a light receiving means for selectively receiving the fluorescence wavelength. Therefore, there is a problem that the optical system becomes complicated.
The reflectance measurement method has a problem that it is difficult to measure the oil spill area and the amount of spill because there is a risk of malfunction because it detects other than oil if the reflectance is higher than water. .
In the luminance change detection method, if there is a luminance change, the cause is detected even if it is not oil, so there is a risk of malfunction.
Further, the ellipsometry has a problem that not only the light projecting system and the light receiving system are complicated, but also the detection range is as narrow as 10 cm × 10 cm.

For this reason, the Applicant has taken a color image of the rainbow-colored interference pattern that occurs when an oil film is formed on the water surface, and analyzed the colors contained in the image, so that the presence or absence of the oil film is accurately formed over a wide range. We have proposed an oil film detector that can detect spillage and can also measure the area of oil spilled into the river.
JP 2007-147448 A

However, as a result of taking images of rivers where oil is actually spilled and experimenting with waste oil flowing in an outdoor experimental facility, even if an oil film is formed on the water surface, it cannot be measured accurately unless conditions are met. There was found.
The following can be considered as the cause.
(1) The oil that has flowed out vaporizes or diffuses over time and becomes thinner than the film thickness that produces an iridescent interference pattern, or depending on the water temperature, the oil film becomes thick and does not become iridescent.
(2) The iridescent interference pattern does not occur depending on the type and viscosity of the oil.
(3) A general surveillance camera has a small number of pixels of about 300,000 pixels, and the resolution is lower than that of a 10 million pixel digital camera used in the experiment, so the accuracy of color information is low.
(4) When capturing the image, the specular reflection light is removed by the polarizing filter. However, when an oil film is actually detected using a surveillance camera or the like outdoors, depending on the incident angle of the sunlight and the shooting angle. By using a polarizing filter, interference fringes often disappear. If a polarizing filter is not used, the sky and wall color are strongly reflected in the oil film, making it difficult to extract iridescent interference fringes.

On the other hand, when an oil film is formed on the water surface, the brightness of the oil film portion is higher than that of the water surface regardless of whether or not an iridescent interference pattern is formed.
Therefore, if there is no noise, it is possible to identify the oil film by changing the brightness, but a wave is generated on the actual water surface, and the brightness change that the light reflected on the wavefront becomes noise and cannot be removed by the polarization filter Therefore, it is difficult to specify the portion where the oil film is formed from the image signal.
As a result of extensive research by the present inventors, it has been found that the luminance change caused by the formation of the oil film has a higher frequency than the luminance change caused by the formation of the wave.

  Therefore, the present invention has been made based on such inventor's knowledge, and when an oil film is formed on the water surface, the frequency of the luminance change regardless of whether or not an iridescent interference pattern is formed. Based on this difference, it is a technical problem to detect the presence or absence of the formation of an oil film by removing the brightness change due to waves and extracting only the brightness change due to the oil film.

  In order to solve this problem, the present invention is an oil film detection device that detects an oil film formed on the water surface, and includes an imaging device that images the water surface and an image processing device that processes the image signal. The image processing device passes a Fourier transform means for outputting a Fourier signal based on a frequency component included in an image signal input from the imaging device, and a frequency component higher than a luminance change caused by a wave on the water surface included in the Fourier signal. A high-pass filter that performs an inverse Fourier transform on the Fourier signal that has passed through the high-pass filter and returns the image signal to an image signal, and an oil film extraction that binarizes the image signal that has been subjected to the inverse Fourier transform and extracts an oil film portion And a means.

According to the present invention, when the water surface is imaged by the imaging device, the image signal is input to the image processing device.
In the image processing apparatus, first, an image signal is Fourier transformed, a frequency component included in the image signal is analyzed, and output as a Fourier signal.
Then, according to the conditions at the observation place such as the light source and the sunshine, the luminance change caused by the wave is cut off and passed through the high-pass filter in which the cutoff frequency is determined so that the luminance change caused by the oil film can be transmitted. Luminance changes caused by waves are eliminated.
Then, when the signal is subjected to inverse Fourier transform and returned to the image signal, an image in which the luminance change due to the wave is removed from the original image is obtained.
Finally, if binarization is performed with an appropriate threshold value, the part where the oil film is formed is white and the part where there is no oil film is projected black, so if there is a white part, it can be determined that the oil film is formed. If necessary, the amount of oil leakage can be calculated based on the area of the white portion.

  In this example, regardless of whether a rainbow interference pattern is formed, based on the difference in frequency of the luminance change, the luminance change due to the wave is removed, and only the luminance change due to the oil film is extracted. In order to achieve the purpose of detecting the presence or absence of formation, an oil film detection device for detecting an oil film formed on the water surface, comprising: an imaging device for imaging the water surface; and an image processing device for processing the image signal The image processing apparatus includes: Fourier transform means for outputting a Fourier signal based on a frequency component included in the image signal input from the imaging apparatus; and a frequency component higher than a luminance change caused by a wave on the water surface included in the Fourier signal. A high-pass filter that passes through the high-pass filter, an inverse Fourier transform means that performs an inverse Fourier transform on the Fourier signal that has passed through the high-pass filter and returns the image signal, and an inverse Fourier transform Characterized in that a oil film extracting means for extracting the oil film portion by binarizing the image signal.

Hereinafter, the present invention will be described based on embodiments shown in the drawings.
FIG. 1 is an explanatory view showing an example of an oil film detection apparatus according to the present invention, FIG. 2 is an explanatory view showing an image of a water surface and a process of image processing this, FIG. 3 is a flowchart showing a processing procedure of the present invention, and FIG. It is explanatory drawing which shows the processed image.

The oil film detection apparatus 1 shown in FIG. 1 images the water surface of the monitoring area E set in the irrigation channel 2 and detects whether an oil film has flowed in by detecting whether an oil film R is formed on the water surface. The monitoring area E, which is to be determined, is covered with a roof or the like so as not to be affected by direct sunlight, and the lighting device 3 that irradiates the water surface with illumination light, and imaging that performs color imaging of the water surface of the monitoring area E A device 4 and an image processing device 5 for processing the image signal are provided. The imaging device 4 is equipped with a polarizing filter (not shown) that cuts the reflected light from the liquid surface.
The illumination device 3 irradiates at least a part of the visible light region with illumination light having a continuous spectral distribution. In this example, a plurality of fluorescent lamps are arranged in parallel, and a milky white acrylic scattering plate is provided in front of the fluorescent light. Is used.

  This illumination light has a generally continuous spectrum although some peaks exist in the visible light region having a wavelength of 400 nm to 700 nm, and contains many wavelength components of visible light. When light is reflected, the light reflected by the oil film R and the light passing through the oil film surface and reflected by the boundary surface between the oil film R and the water surface interfere with each other, and light of different wavelengths depends on the thickness of the oil film R. An iridescent interference pattern is observed, and depending on the type of oil, an iridescent pattern is not formed and an oil film pattern having a brightness different from that of the water surface is observed.

The image processing device 5 includes a computer to which a display device 6 is connected, and includes Fourier transform means M ₁ that outputs a Fourier signal FS ₁ based on a frequency component included in the image signal IMG ₁ input from the imaging device 4, and Fourier A high-pass filter M ₂ that passes a higher frequency component than a luminance change caused by a wave on the water surface included in the signal FS ₁ , and a Fourier signal FS ₂ that has passed through the high-pass filter M ₂ are subjected to inverse Fourier transform to generate an image signal IMG ₃ . Inverse Fourier transform means M _{3 for} returning, oil film extraction means M ₄ for outputting the binarized image signal IMG ₅ obtained by binarizing the image signal IMG ₃ subjected to inverse Fourier transform and extracting the oil film portion, and binarization Judgment means M ₅ for judging whether or not the oil film R is formed based on the image signal IMG ₅ is provided.
Further, when the determination means M ₅ in oil film R is determined to be formed, and a runoff area calculating means M ₆ for calculating the outflow area of the oil on the basis of the shape of the portion where the oil film R is formed .

When the image signal IMG ₁ output from the imaging device 4 is a color signal, the Fourier transform unit M ₁ performs a color reduction process for reducing the color of the image signal IMG ₁ to a grayscale image signal IMG ₂ having 256 gradations. When the Fourier transform process is performed and the image signal IMG ₁ output from the imaging device 4 is a monochrome signal (256 gray scale image signal), the Fourier transform is set as it is without performing the color reduction process. Yes.

FIG. 2 shows a comparison of image processing processes when (a) there is no wave and no oil film, (b) there is a wave and no oil film, and (c) there is a wave and oil film.
Figure 2 (a-1) (b -1) (c-1) represents the image by the image signal IMG ₂ gray scale, FIG. 2 (a-2) (b -2) (c-2) is the Fourier signal It is an image generated by FS ₁ .
Fourier signal FS ₁ image, horizontal frequency components horizontal axis, vertical axis is vertical frequency component, the concentration indicates the magnitude of the amplitude.
According to this, when there is neither a wave nor an oil film in FIG. 2 (a-2), it is bright overall, and when there is a wave in FIG. 2 (b-2) and there is no oil film, it is dark near the origin where the low frequency component appears. When there is a point and both the wave and the oil film in FIG. 2 (c-2) are present, the dark spot is observed up to a position away from the origin, and it can be seen that the luminance change of the high frequency component due to the oil film appears.

High pass filter M ₂ separates the frequency component of the luminance change caused by the waves, the frequency components of the luminance change caused by oil film R.
The frequency of the luminance change caused by the oil film R is generally higher than the frequency of the luminance change caused by the wave.
Therefore, if you set higher than the frequency of the luminance change caused by the waves the cutoff frequency of the high-pass filter M ₂ (cutoff frequency = 50 Hz in this example), from the Fourier signal FS ₁ output from the Fourier transform unit M ₁ A low-frequency component such as a luminance change caused by a wave is cut, and a Fourier signal FS ₂ in which only a high-frequency component such as a luminance change caused by the oil film R is transmitted is output.

Inverse Fourier transform means M ₃ are, the process is performed to recover the Fourier signal FS ₂ to lower component frequency is cut into the image signal IMG _3.
Since the frequency component of the brightness change caused by the wave is removed from the Fourier signal FS ₂ , when this is returned to the image signal IMG ₃ , the brightness change of the wave is removed and only the brightness change due to the oil film R is removed. Becomes an image left as a luminance change.

Since the low frequency component is cut in all the images based on the Fourier signal FS ₂ , the vicinity of the origin of the images is white as shown in FIGS. 2 (a-3) to (c-3). The Fourier images shown in (a-3) and (b-3) are substantially equal, and only the Fourier image shown in FIG.
Therefore, when the inverse Fourier transform is performed, the image signal IMG ₃ in FIGS. 2A-4 and 2B-4 almost eliminates the luminance change caused by the waves, and the image signal IMG in FIG. _The luminance change due to the oil film is reproduced only in the image of ₃ .

The oil film extraction means M ₄ binarizes the image signal IMG ₃ with an appropriate threshold value so that, for example, a portion where the oil film R is formed is displayed black and a portion where the oil film R is not formed is displayed white.
In this case, the image signal IMG ₃ subjected to Fourier transform is subjected to edge detection processing by a first-order differential (difference) filter such as a Sobel filter or a Prewitt filter as necessary. If the edge image signal IMG ₄ is output and binarized, the boundary between the portion where the oil film is formed and the portion where the oil film is not formed becomes clearer, so that errors caused by binarization are small.
Then, when the binarization processing is completed, the binarized image signal IMG ₅ from which the oil film portion has been extracted is output.
Thereby, it can be determined that the oil film R is formed in the black part, and the oil film R is not formed in the white part.

Determination means M ₅ is to determine the presence or absence of formation of the oil film R on the basis of the binary image signal IMG _5.
Then, when the number of pixels in the black portion is detected more than a preset number, it is determined for the first time that it is an oil film, so that some warning signal is output, or an alarm device (not shown) is activated to Inform the spill.
Since it is difficult to completely remove the luminance change due to the wave by high pass filter M _2, which makes it possible to avoid your operation of the alarm device.

Then, in the outflow area calculating means M _6, when the oil film determination unit M ₅ is determined to have been formed, based on the virtual grid formed on the water surface at a pitch corresponding to the resolution (e.g., 2 mm to 10 mm) The oil outflow area is calculated by counting the number of black grids on which the oil film R is formed.

The above is one configuration example of the present invention, and the operation thereof will be described with reference to FIGS. 4 and 5.
FIG. 4 is a flowchart showing a processing procedure of the image processing apparatus 5.
When oil film detection is performed, when the imaging device 4 is turned on with the illumination device 3 turned on and the image processing program is executed, a color image is obtained from the imaging device 4 at predetermined time intervals (for example, 10 seconds) in step STP1. signal IMG ₁ is captured after being Genshoku grayscale image signal IMG ₂ of 256 gradations in step STP2, the Fourier transform processing is performed at step STP3, Fourier signal FS ₁ is output.
5 (a) is an image by the image signal IMG ₂ gray scale, FIG. 5 (b) is a Fourier transformed image.

Then, at step STP4, it is transmitted through the high-pass filter M _2, and outputs the Fourier signal FS ₂ obtained by cutting the luminance change caused by the waves.

Then, the inverse Fourier transform processing on at step STP5, Returning Fourier signal FS ₂ to the image signal IMG _3, the luminance change due to waves from an image of the image signal IMG ₃ is to be cut.
Figure 5 (c) is the image by the image signal IMG _3.

Next, edge detection processing is performed in step STP6, and the output edge image signal IMG ₄ is binarized in step STP7, and a binarized image signal IMG ₅ is output.
FIG. 5D shows an image based on the binarized image signal IMG ₅ .

Then, the number of dark spot pixels displayed in the image obtained in step STP8 is counted, and when this number is smaller than a preset number, it is determined that no oil film is formed, and an alarm signal is generated. Returning to step STP1 without outputting, when the number of dark spot pixels is larger than a preset number, it is determined that an oil film R has been formed, and the routine proceeds to step STP9 where a predetermined alarm signal is output.
Next, the process proceeds to step STP10, the oil outflow film area is calculated, and the process returns to step STP1.

Incidentally, in the above-described processing procedure, processing steps STP2 and 3 are performed by the Fourier transform processing means _{M 1,} the processing step STP4 is performed by the high-pass filter _{M 2,} step STP5 is performed by the inverse Fourier transform unit _{M 3} that process, process step STP6 and 7 is executed by the oil film extracting unit _{M 4,} process step STP8 and 9 is executed by the determination means _{M 5,} is in the process step STP10 runs outflow area calculating means _{M 6} .

  In the above description, the portion where the oil film R is formed is displayed in black, and the portion where the oil film R is not present is displayed in white.

  As described above, the present invention can be applied to the use of detecting the presence or absence of an oil film formed when oil flows out to the surface of a river or the like.

Explanatory drawing which shows an example of the oil film detection apparatus which concerns on this invention. Explanatory drawing which shows the process which image-processed the image of the water surface, and this. The flowchart which shows the process sequence of this invention. Explanatory drawing which shows the processed image.

DESCRIPTION OF SYMBOLS 1 Oil film detector 2 Water channel E Monitoring area R Oil film 3 Illuminating device 4 Imaging device 5 Image processing device 6 Display device M ₁ Fourier transform means M ₂ High pass filter M ₃ Inverse Fourier transform means M ₄ Oil film extraction means M ₅ Determination means

Claims (6)

An oil film detection device for detecting an oil film formed on a water surface,
An imaging device for imaging the water surface, and an image processing device for processing the image signal;
The image processing device passes a Fourier transform means for outputting a Fourier signal based on a frequency component included in an image signal input from the imaging device, and a frequency component higher than a luminance change caused by a wave on the water surface included in the Fourier signal. A high-pass filter that performs an inverse Fourier transform on the Fourier signal that has passed through the high-pass filter and returns the image signal to an image signal, and an oil film extraction that binarizes the image signal that has been subjected to the inverse Fourier transform and extracts an oil film portion And an oil film detecting device.   2. The oil film detection device according to claim 1, wherein when the image signal is a color image, the Fourier transform means performs a color reduction process on the gray scale image signal before Fourier transforming the image signal.   The oil film detection device according to claim 1, wherein the oil film extraction unit performs binarization processing after performing edge detection processing on the inverse Fourier transformed image signal. An oil film detection device for detecting an oil film formed on a water surface,
The image signal obtained by imaging the water surface is subjected to Fourier transform, and the obtained Fourier signal is passed through a high-pass filter to output only frequency components higher than the luminance change caused by the wave on the water surface, and inverse Fourier transform is performed to the Fourier signal that has passed through the high-pass filter. The oil film detection method is characterized in that an oil film portion is extracted by performing binarization processing on the image signal that has been subjected to inverse Fourier transform and returning to an image signal.   The oil film detection method according to claim 4, wherein when the image signal is a color image, a color-reduction process is performed on the gray-scale image signal before Fourier transform. The oil film detection method according to claim 4, wherein the binarization process is performed after the edge detection process is performed on the inverse Fourier transformed image signal.