KR102476565B1 - A floating waste detection system for lake and swamp using statistical characteristics of SAR radar images and methods using the same - Google Patents

Abstract

The present invention relates to a lake floating waste detection system using the statistical characteristics of SAR radar images, which automatically analyzes a change in a lake in time series and displays floating wastes, and a method thereof. To this end, the lake floating waste detection system using statistical characteristics of SAR radar images comprises: an image acquisition unit which acquires a plurality of SAR images for each time period; a pre-processing unit which matches or corrects the acquired SAR images; a hypothesis verification unit which verifies whether images of different time zones belong to the same category by using the plurality of SAR images; and a floating matter display unit which displays pixels determined to be a floating object in the SAR image of the time zone determined not to belong to the same category by the hypothesis verification unit with a predetermined color. Therefore, according to the present invention, floating wastes in a lake are accurately detected and monitored through image analysis using the SAR images regardless of atmospheric conditions such as clouds, fog, smog, or rainfall, thereby facilitating the management of floating waste and reducing concerns of environmental pollution.

Description

A floating waste detection system for lake and swamp using statistical characteristics of SAR radar images and methods using the same}

The present invention relates to a system and method for detecting floating waste in lakes using statistical characteristics of SAR radar images. It relates to a system and method for detecting floating waste in lakes using characteristics.

Large multi-purpose dams have been built in the upper reaches of major rivers in Korea and have played an important role in the efficient maintenance and management of water resources. Recently, as rainfall increases due to rapid climate change, floating waste in lakes is increasing. Floating waste and waterside pollutants caused by concentrated rainfall rapidly deteriorate water quality over time, and waste monitoring is required from the initial stage of generation to determine whether waste is generated, establish a collection plan, and budget.

In this regard, according to Patent Registration No. 10-2316387 (floating object detection method and device) according to the prior art, a first high-resolution image labeled with a position where a floating object exists and a region corresponding to the first high-resolution image, including Generating a first artificial neural network by machine learning a first artificial neural network using a captured first low-resolution image; obtaining a second low-resolution image; and estimating, for each pixel in the second low-resolution image, an amount of floating matter present in a cell region corresponding to a corresponding pixel by inputting the input to an intelligent neural network.

However, since this technology uses an optical image based on visible light, there are many days when it is impossible to secure an image depending on weather conditions such as a lot of clouds, so there is a limit to continuous monitoring.

Therefore, studies using remote sensing data such as Sentinel-I SAR images are needed to monitor particulate matter in a wide area. The eutrophication of water quality due to the neglect of waste is due to the absence of spatial information such as changes in the extent of waste generation. Therefore, a floating waste detection methodology that provides continuous monitoring information by searching and extracting change areas from SAR (Synthetic Aperture Radar) images is needed.

The present invention is to solve the above problems, and SAR can facilitate management of floating waste and reduce the concern of environmental pollution by accurately detecting and monitoring floating waste of appeal through image analysis at all times. It is to provide a system and method for detecting floating waste in lakes using statistical characteristics of radar images.

In order to achieve the above object, the present invention is an image acquisition unit for acquiring a plurality of SAR images for each time period; a pre-processing unit that matches or corrects the acquired SAR image; a hypothesis verification unit verifying whether images of different time zones belong to the same category by using the plurality of SAR images; The appeal unit using the statistical characteristics of the SAR radar image, which includes a floating object display unit that displays the color of a pixel determined to be a floating object in a predetermined color in the SAR image of the time zone determined not to belong to the same category in the hypothesis verification unit. Provides an oily waste detection system.

The pre-processing unit uses a dual-polarization GRD image that has undergone basic processing, and includes thermal noise removal, border noise removal, speckle filtering, and Doppler region correction (Range Doppler Terrain Correction). ), Conversion to dB process may be performed.

The hypothesis verification unit sets a hypothesis that all SAR images of multiple time zones belong to the same category, and each pixel sets the measured intensity by the VV band and the VH band as S _VV and S _VH , respectively, based on the polarization direction, and the S _VV , S _VH is calculated as an element of the covariance matrix, and whether the hypothesis is accepted or rejected is determined based on a probability statistic using the determinant of the covariance matrix.

The hypothesis verification unit determines whether the hypothesis is suitable or rejected based on the total likelihood statistic. It is desirable to judge sequentially.

Preferably, the hypothesis verification unit determines whether the hypothesis is accepted or rejected by comparing a change measurement value based on a probability statistic in each SAR image with a preset change reference value.

The hypothesis verification unit may obtain a distribution of the change measurement values in each SAR image and determine whether to remove noise by calculating a ratio of values smaller than the change reference value.

According to the present invention, it is easy to manage floating waste by accurately detecting and monitoring floating waste in lakes through image analysis at all times regardless of atmospheric conditions such as clouds, fog, smog, or rainfall using SAR images. It has the advantage of reducing the risk of environmental pollution.

1 is a block diagram of a floating waste detection system according to the present invention;
Figure 2 is a conceptual diagram of the operation of the floating waste detection system according to the present invention;
3A and 3B are exemplary diagrams showing the signal distribution of radar waves of SAR images captured on August 31, 2018 and April 9, 2020, respectively;
4a is an aerial photograph of the study area
Figure 4b is a result diagram showing the distribution of change measurement values (p-value) in the study area;
Figure 4c is a result display diagram for displaying the results according to the present invention;
5 is a field photograph showing an image of a study area taken by a drone.

The configuration and operation of specific embodiments according to the present invention will be described in detail with reference to the drawings.

Referring to FIG. 1, the floating waste detection system 100 of the lake using the statistical characteristics of the SAR radar image according to this embodiment may be provided in the form of a server, an image acquisition unit 110, a pre-processing unit 120 , a hypothesis verification unit 130, and a floating object display unit 140.

The image acquisition unit 110 receives an image of a target area to be monitored, and the image is preferably a synthetic aperture radar (SAR) image captured using a satellite radar.

The monitoring method using optical images using visible light has a low image frequency that can be obtained due to meteorological conditions such as no sunlight or cloudy weather, and it is difficult to analyze images due to atmospheric turbidity. Not suitable for a single monitoring system.

On the other hand, the SAR image has the advantage of being able to be monitored at all times regardless of the presence or absence of sunlight and atmospheric conditions by using an active sensor.

The image acquisition unit 110 may continuously receive SAR images captured in the same area at different times and receive images captured multiple times at regular time intervals.

For example, the image acquisition unit may receive images captured in the 3 o'clock band and the 15 o'clock band twice a day, and the 3 o'clock and 15 o'clock bands may be SAR images captured 100 times for 10 seconds.

The pre-processing unit 120 serves to transform the received SAR image so that it can be analyzed.

Specifically, the pre-processing unit includes a trajectory correction unit, a noise removal unit, a radiation correction unit, and a terrain correction unit.

The orbit correction unit serves to update orbit information that has not been input when initial data is generated, and, for example, accurately updates satellite position and speed information.

The noise removal unit removes thermal noise and speckle of the SAR image.

The inside of the satellite communication system is composed of various elements, and as the movement of electrons becomes active, heat is generated, which causes noise in the communication system. The noise removal unit extracts the information signal and the output signal, arranges them according to time and input level, calculates distortion through comparison, calculates distortion compensation values, and stores these compensation values in a table in a storage unit.

These compensation values may be updated in real time, and the noise removal unit removes thermal noise by adding compensation values.

In addition, the noise removal unit removes speckle by performing spatial filtering by removing or emphasizing a specific frequency in the spatial domain with respect to the video signal.

The radiation correction unit serves to convert raw data into surface reflectivity. That is, by using the metadata of the satellite itself, the pixel value is corrected with the pure reflectivity value of the object by using the correction coefficient.

The topography correction unit performs a geometrical correction process to remove and correct distortions caused by a change in the position of an image due to a change in attitude of a satellite and acquisition of an inclination distance image by SAR. That is, the topographic correction unit determines a reference image and performs geometric transformation by matching a plurality of SAR images configured in time series with the reference image.

The hypothesis verification unit 130 establishes a hypothesis and determines whether or not the established hypothesis is appropriate. For example, a hypothesis is set that "a plurality of SAR images for each time period are included in the identity category", and the hypothesis verification unit determines whether the hypothesis is suitable for all SAR images or individual SAR images.

Referring to FIG. 2, the radar wave transmitted from the SAR satellite is reflected from the surface of the water and maintains a time-sequentially stable signal strength. At this time, if the floating waste is floating on the surface of the water, the radar wave transmitted from the SAR satellite is scattered by the floating waste and the signal strength is changed. When this change is detected, the location of the floating waste can be detected.

However, referring to FIGS. 3A and 3B, even in the same place, SAR images may appear differently depending on natural phenomena such as foliage growing thick or waves waving over time, and since the SAR satellite itself moves, the shooting angle and Depending on time, the SAR image of the same place may appear differently.

The hypothesis verification unit determines whether the changed SAR image is due to such a natural change or floating waste.

In this embodiment, the hypothesis verification unit assumes that the population of radar wave signals appears in the form of a Wishart distribution, and the standard deviation and average of the radar wave distribution depend on the shooting time, satellite orbit, shooting angle, and change in the ground surface. The variance is different, and if floating waste is generated, it is predicted that it will affect the normal distribution with a larger error than the changed pixel above.

In addition, assuming that the signal characteristics of the ground surface reflected from the radar wave follow a continuous random variable, the ground surface radar wave reflected with the probability density function can be simulated, and the changed signal strength can be statistically estimated from the predicted probability density function. do.

In addition, in the case of lakes (swamps and lakes) where the influence of wind on the water surface perpendicular to the direction of gravity is relatively small, the signal strength is very stable and will always maintain a constant average value, and this characteristic is the signal strength when changes occur on the surface of the lake It is predicted that it may appear as a change in

Hereinafter, a method for evaluating the goodness of fit of the probability density distribution by the hypothesis verification unit 130 will be described.

If the measured intensities of the VV and VH bands of the polarized signals at one pixel are Svv and Svh, respectively, the covariance matrix for pixel m can be expressed as follows.

< > means the ensemble average.

Sampling number of SAR image i = 1,2,… , k, and ENL (equivalent number of looks) n, an independent random variable X _i =n<C> _i , an expected value E{X _i /n}=Σ _i , and an independent random variable X _i Assuming that follows the complex order distribution Wc(p, n, Σ _i ), the hypothesis that SAR images for each time period belong to the identity category is Σ _{1 =} Σ _{2 = … =} Σ _k ,

In the hypothesis verification unit, the total likelihood statistic Q for the plurality of SAR images is set as follows, and the total likelihood statistic is evaluated to determine whether the hypothesis is suitable or rejected.

이고, | |는 행렬식을 나타낸다.here,

and | | represents a determinant.

In the case of data with three polarizations, p = 3, in the case of two polarizations, p = 2, and in the case of only one band, p = 1.

The total likelihood statistic Q has a value of [0,1], and indicates perfect equality when Q=1.

Also, taking the logarithm on both sides, it can be expressed as follows.

here,

, the probability of a value of -2ρlnQ smaller than z(=-2ρlnq) (q is an observed value) is as follows.

Here, χ ² (f) represents the chi-square probability distribution function with the degree of freedom f.

Meanwhile, the partial likelihood statistic R _j for the j-th SAR image can be set as follows.

임을 알 수 있다.in other words,

It can be seen that

here,

f = p ²

, the approximate probability density distribution of -2ρ _j lnR _j is as follows.

In the hypothesis verification unit 130, the change measurement value (p-value)

It is calculated as Pr(Q<q)=1-Pr(-2ρlnQ ≤ -2ρlnq) (q is an observed value) and compared with the change reference value, which is a significance level of 0.01 to 0.005, to determine whether the hypothesis is suitable or rejected. It is preferable that the change reference value is converted into a DB and stored in the storage unit, and the hypothesis verification unit 130 searches and determines the DB according to the set significance level.

That is, if the change measurement value is greater than the change reference value, the hypothesis verification unit determines that the corresponding pixel is included in the identity category, and if the change measurement value is smaller than the change reference value, the corresponding pixel is not included in the identity category and floats are generated on the surface of appeal. judge that this has changed.

Next, a method for detecting floating waste in lakes using statistical characteristics of SAR radar images according to the present invention will be described.

The image acquisition unit 110 receives a Synthetic Aperture Radar (SAR) image captured using a satellite radar, and the received SAR image is matched or corrected in the pre-processing unit 120 and transformed so that the SAR image can be analyzed.

Thereafter, the hypothesis verification unit 130 hypothesizes that the SAR images for each time period belong to the same category, and calculates the change measurement value (p-value) for the total likelihood statistic Q

Pr(Q<q)=1-Pr(-2ρlnQ ≤ -2ρlnq) is calculated, and a change reference value with a significance level of 0.005 stored in the storage unit is detected and compared with the change measurement value.

As a result of comparison, if the change measurement value is greater than the change reference value, all SAR images for each time period belong to the same category, so the hypothesis is judged to be suitable and marked as no float. On the other hand, if the change measurement value is smaller than the change reference value, the hypothesis is rejected and it is determined that there is floating waste, and two SAR images are extracted from each SAR image to sequentially perform hypothesis verification.

That is, the hypothesis verification unit sequentially calculates Pr(R _j < r _j ) = 1-Pr(-2ρlnR _j ≤ -2ρlnR _j ), detects the change reference value with a significance level of 0.001 stored in the storage unit, and calculates the change measurement value. Compare with At this time, if the pixel signals of the two SAR images are s1 and s2, respectively, the hypothesis verification unit calculates s1/s2 and can be used as an input for the probability statistic Q.

As a result of comparison, if the change measurement value for each pixel is greater than the change reference value, it is judged that the hypothesis is appropriate because all SAR images for each time period belong to the same category, and it is marked as no float. On the other hand, if the change measurement value is smaller than the change reference value, the hypothesis is rejected and it is determined that there is floating waste in the corresponding pixel.

On the other hand, the hypothesis verification unit calculates the change measurement value for each pixel in each SAR image to obtain the distribution of the change measurement value, calculates the ratio of pixels smaller than the change reference value among all pixels, and judges it as noise if the ratio is less than a certain ratio. and remove it from the float. In addition, if the location of a pixel smaller than the change reference value is not the surface of the water (appeal) by using the location information of the pixel, it is determined as noise and removed from the float.

The floating object display unit 140 displays pixels determined to be floating objects by the hypothesis verification unit 130 in a preset color (for example, red) on the map. In addition, in the floating matter display unit, if an area judged to be floating waste at a certain time is determined to be non-floating waste at another time using the change measurement value calculated in time series, the area is displayed in a different color (for example, blue). can give it

As a result of actual execution, as a result of calculating the change measurement value for the region in FIG. 4a and obtaining the distribution, the same result as in FIG. 4b was obtained, and the final result as in FIG. 4c was obtained by judging the significance level of 0.001. In FIG. 4C, the area marked in red is determined as the area corresponding to the floating waste.

Referring to FIG. 5, it was confirmed that there was a floating object when checking the image taken by the drone at the same time.

Although the above has been described with reference to the embodiments of the present invention, those skilled in the art can variously modify and modify the present invention within the scope not departing from the spirit and scope of the present invention described in the claims below. You will understand that it can be changed.

100: floating waste detection system 110: image acquisition unit
120: pre-processing unit 130: hypothesis verification unit
140: float display unit

Claims (5)

an image acquisition unit that acquires a plurality of SAR images for each time period;
a pre-processing unit that matches or corrects the acquired SAR image;
a hypothesis verification unit verifying whether images of different time zones belong to the same category by using the plurality of SAR images;
A float display unit for displaying the color of pixels determined to be floating in a predetermined color in the SAR image of the time zone determined not to belong to the same category by the hypothesis verification unit,
The hypothesis verification unit establishes a hypothesis that the SAR images for each time period belong to the category of identity, sets the measured intensities by the VV band and the VH band to S VV and S VH, respectively, based _on the polarization direction for each pixel, and sets the S _VV , S _VH Calculate S _VH as an element of the covariance matrix C,
The total likelihood statistic Q for multiple SAR images

(

)
set as
The change measurement value is calculated as Pr(Q<q) = 1-Pr(-2ρlnQ ≤ -2ρlnq) (q is an observed value), the change reference value stored in the storage unit is detected and compared with the change measurement value,
If the change measurement value is greater than the change reference value, all SAR images for each time period belong to the same category and are judged to be free of floating matter, and if the change measurement value is smaller than the change reference value, the hypothesis is rejected of floating waste detection system.
delete According to claim 1,
The hypothesis verification unit determines whether the hypothesis is suitable or rejected based on the total likelihood statistic. A system for detecting floating waste in lakes using statistical characteristics of SAR radar images, characterized in that they are determined sequentially.
delete According to claim 1,
The hypothesis verification unit calculates the distribution of the change measurement values in each SAR image and calculates the ratio of values smaller than the change reference value to determine whether or not to remove the noise. waste detection system.

Images