KR20220138698A - Method and apparatus for rainfall computation - Google Patents

Abstract

A method and an apparatus for calculating rainfall by analyzing a rainfall image are provided. As a result, the present invention can accurately determine day and night weather phenomena in a local area using cheaper observation equipment. The method for calculating rainfall includes steps of: converting a series of frames of a rainfall image into at least one grayscale image; separating at least one rain streak layer from the at least one grayscale image; and determining a rainfall intensity of the rainfall image.

Description

METHODS AND APPARATUS FOR RAINFALL COMPUTATION

The present invention relates to a method and apparatus for calculating rainfall, and to a method and apparatus for calculating rainfall by analyzing a rainfall image.

The content to be described below is only provided for the purpose of providing background information related to the embodiment of the present invention, and the content to be described does not naturally constitute the prior art.

Precipitation generally refers to the phenomenon of water droplets falling to the ground from clouds in the atmosphere. At this time, precipitation has various forms. When water droplets fall in a liquid state, it is defined as rain, and when it falls in a solid state, it is defined as snow or hail.

In general, organizations such as the Korea Meteorological Administration operate ground meteorological observation equipment nationwide to observe rainfall phenomena. However, the current automatic weather observation equipment is mostly dependent on expensive imported sensors and systems that are large in size or complex to apply to local weather phenomena, so the distribution and utilization of the system is limited.

In addition, there is a limitation in that it is difficult to confirm the rainfall phenomenon in an area where automatic weather observation equipment is not installed because automatic weather observation equipment is not installed nationwide.

On the other hand, the above-mentioned prior art is technical information possessed by the inventor for derivation of the present invention or acquired in the process of derivation of the present invention, and it cannot be said that it is necessarily known technology disclosed to the general public before the filing of the present invention. .

An object of the present invention is to provide a method and apparatus for calculating rainfall by analyzing day and night image data.

It is an object of the present invention to provide a rainfall calculation method and apparatus for more accurately observing a local rainfall phenomenon while using an economical weather sensor.

The object of the present invention is not limited to the above-mentioned problems, and other objects and advantages of the present invention that are not mentioned may be understood by the following description, and will be more clearly understood by the embodiments of the present invention. It will also be appreciated that the objects and advantages of the present invention may be realized by means of the instrumentalities and combinations thereof indicated in the claims.

A rainfall calculation method according to an embodiment of the present invention includes converting a series of frames of a rainfall image into at least one grayscale image, and using a raindrop identification function defined based on a change in pixel value of the at least one grayscale image. separating the at least one rain layer from the at least one grayscale image and determining the rainfall intensity of the rain image based on rainfall characteristic information of the rain region displayed in the at least one rain layer. .

A rainfall calculation apparatus according to an embodiment of the present invention includes a memory and a processor for storing a rainfall image including a series of frames, wherein the processor converts the series of frames into at least one grayscale image, and The at least one rain layer is separated from the at least one gray scale image using a rain rain identification function defined based on a change in the pixel value of the at least one gray scale image, and the rainfall characteristic of a rain region displayed in the at least one rain layer and determine the rainfall intensity of the rainfall image based on the information.

Other aspects, features, and advantages other than those described above will become apparent from the following drawings, claims, and detailed description of the invention.

According to the embodiment, it is possible to calculate the amount of precipitation in the observation area by analyzing the images obtained using CCTV cameras installed at various locations nationwide.

According to the embodiment, it is possible to accurately determine a meteorological phenomenon in a local area where there is no precipitation observation data using cheaper observation equipment.

According to the embodiment, it is possible to measure the amount of rainfall by using the rain stream of the rainfall image captured by the CCTV camera, and it is possible to calculate the amount of rain falling at night through the infrared photographed image.

Effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is an exemplary diagram of an operating environment in which a rainfall calculation process according to an embodiment is performed.
2 is a block diagram of an apparatus for calculating rainfall according to an embodiment.
3 is a flowchart of a method for calculating rainfall according to an embodiment.
4 is a view for explaining a rain layer separation process of the rainfall calculation method according to the embodiment.
5 is a flowchart of a rainfall intensity calculation process according to an embodiment.
6 is a flowchart of a rain layer separation process according to an embodiment.
7 is a view for explaining an index used in a rainfall intensity calculation process of the rainfall amount calculation method according to the embodiment.
8 exemplarily shows a rain layer separation result of a rainfall calculation process according to an embodiment.
9 exemplarily shows a rain layer separation result of a rainfall calculation process according to an embodiment.
10 is a graph exemplarily showing a rainfall calculation result according to an embodiment.

Hereinafter, the present invention will be described in more detail with reference to the drawings. The present invention may be embodied in several different forms, and is not limited to the embodiments described herein. In the following embodiments, parts not directly related to the description are omitted in order to clearly explain the present invention. . In addition, the same reference numerals are used for the same or similar elements throughout the specification.

In the following description, terms such as first, second, etc. may be used to describe various components, but the components should not be limited by the terms, and the terms refer to one component from another component. It is used only for distinguishing purposes. Also, in the following description, the singular expression includes the plural expression unless the context clearly dictates otherwise.

In the following description, terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification is present, but one or more other It should be understood that this does not preclude the possibility of addition or presence of features or numbers, steps, operations, components, parts, or combinations thereof.

Hereinafter, the present invention will be described in detail with reference to the drawings.

1 is an exemplary diagram of an operating environment in which a rainfall calculation process according to an embodiment is performed.

The camera 200 is disposed on the target site to measure the amount of rainfall, and transmits the captured field image to the rainfall calculation apparatus 100 according to the embodiment through the network 300 . For example, the camera 200 includes a camera capable of capturing an infrared image. For example, the camera 200 may include a general CCTV camera or an infrared camera.

The camera 200 may store the captured image in a memory (eg, SD card, DVR, etc.) of the camera 200 . The camera 200 may transmit the field image to the rainfall calculation device 100 in a real-time streaming method (eg, NTSC and LAN communication).

The rainfall amount calculation apparatus 100 according to the embodiment calculates the rainfall amount from the input rainfall image by executing a series of processes according to the rainfall amount calculation method according to the embodiment.

In an example, the rainfall calculation device 100 may be implemented as a server device. For example, the rainfall calculation apparatus 100 may be implemented as a part of a weather control system. The rainfall amount calculation apparatus 100 may receive rainfall images from a plurality of cameras 200 disposed in various destinations through the network 300 , respectively, and calculate the rainfall amount of each destination.

In one example, the rainfall calculation device 100 may be implemented integrally with the camera 200 or directly connected to the camera 200 or may be disposed near the camera 200 and connected to the camera 200 by wireless communication. . In an example, the camera 200 may directly calculate the amount of rainfall in the target area where the camera 200 is located by directly executing the method of calculating the amount of rainfall according to the embodiment, and transmit the calculation result to the weather control system.

The rainfall calculation apparatus 100 or the weather control system may transmit the rainfall calculation result to the terminal 400 through the network 300 . The terminal 400 may receive the rainfall calculation result of the rainfall calculation apparatus 100 according to the embodiment from the rainfall calculation apparatus 100 or the weather control system.

In an example, the terminal 400 may provide rainfall information to a user based on a rainfall calculation result according to an embodiment by executing an application or an app that provides weather information. For example, the terminal 400 may provide the user with rainfall amount information of the user's current location or a location of interest set by the user.

Network 300 is a wired and wireless network, such as a local area network (LAN), a wide area network (WAN), the Internet (internet), intranet (intranet) and extranet (extranet), and mobile networks, such as It may be any suitable communication network, including cellular, 3G, LTE, 5G, WiFi networks, ad hoc networks, and combinations thereof.

Network 300 may include connections of network elements such as hubs, bridges, routers, switches, and gateways. Network 300 may include one or more connected networks, eg, multiple network environments, including public networks such as the Internet and private networks such as secure enterprise private networks. Access to network 300 may be provided via one or more wired or wireless access networks.

2 is a block diagram of an apparatus for calculating rainfall according to an embodiment.

The rainfall calculation apparatus 100 according to the embodiment may include a processor 110 and a memory 120 . Such a configuration is exemplary, and the rainfall calculation device 100 may include some of the configurations shown in FIG. 2 , or additionally include a configuration necessary for the operation of the device although not shown in FIG. 2 .

The rainfall calculation apparatus 100 may include a memory 120 and a processor 110 for storing a rainfall image including a series of frames.

The processor 110 is a kind of central processing unit, and may control the operation of the rainfall calculation apparatus 100 by executing one or more instructions stored in the memory 120 .

The processor 110 may include any type of device capable of processing data. The processor 110 may refer to, for example, a data processing device embedded in hardware having a physically structured circuit to perform a function expressed as a code or an instruction included in a program.

As an example of the data processing device embedded in the hardware as described above, a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated (ASIC) circuit) and a processing device such as a field programmable gate array (FPGA), but is not limited thereto. The processor 110 may include one or more processors.

The processor 110 may control each configuration of the rainfall amount calculation apparatus 100 to analyze the rainfall image to calculate the rainfall amount of the target site.

The processor 110 converts a series of frames into at least one grayscale image, and uses a raindrop identification function defined based on a change in pixel value of the at least one grayscale image from the at least one grayscale image to at least one raindrop layer and to determine the rainfall intensity of the rainfall image based on rainfall characteristic information of the rain region displayed in the at least one rain layer.

The processor 110 may convert a series of frames into at least one grayscale image. The processor 110 may separate the at least one rain layer from the at least one gray scale image by using a rain stream identification function defined based on a change in pixel values of the at least one converted gray scale image.

In one example, the rain streak identification function is an optimization function using, as a determining variable, a first matrix having pixel values of a rain layer for a grayscale image as a component, and a plurality of operation terms and a plurality of operation terms associated with pixel value change It may contain modulating variables.

Here, the plurality of operation terms are a first operation term for spatial density of the first matrix, a second operation term for vertical smoothness of the first matrix, and pixel values of the background layer corresponding to the remainder of separating the rain layer from the grayscale image. It may include a third operation term for the horizontal continuity of the second matrix having as a component, and a fourth operation term for the temporal dependence of the second matrix. For example, each of the first operation term, the second operation term, the third operation term, and the fourth operation term may be an L1-norm operation term associated with the first matrix or the second matrix.

In one example, the processor 110 is configured to separate the at least one rain layer, a first adjustment variable for the first operation term, a second adjustment variable for the second operation term, and the second adjustment variable for the first operation term that minimizes the rain identification function value. The combination of the third adjustment variable for the 3 operation terms and the fourth adjustment variable for the fourth operation term is determined, and the first matrix that minimizes the value of the rain identification function by the determined combination is used as the rain layer for the given grayscale image. can be configured to determine.

Grayscale image conversion and rain layer identification will be described in detail in steps S1 and S2 with reference to FIG. 3 .

The processor 110 may determine the rainfall intensity of the rainfall image based on rainfall characteristic information of the rain region displayed in the separated at least one rain layer.

In an example, the rainfall characteristic information may include an actual size of rain corresponding to a rain region displayed in the at least one rain layer, the speed of the rain, and a rainfall particle size distribution for the at least one rain layer.

In one example, the processor 110 estimates an actual distance between a lens photographing a rainfall image and a rain shower corresponding to a rain region displayed in at least one rain layer to determine the rainfall intensity, and based on the estimated actual distance and determine an actual size of the rain stream, and determine a speed of the rain stream based on the determined actual size of the rain stream.

In one example, the processor 110 determines an image depth of a rainfall image based on a parameter of a lens photographing the rainfall image in order to determine the rainfall intensity, and an effective rainfall volume for at least one rain layer based on the image depth. determine the rainfall particle size distribution of the at least one raindrop layer with respect to the effective rainfall volume, and determine the rainfall intensity for the rainfall image based on the actual size of the raindrop, the velocity of the rainstorm, and the rainfall particle size distribution. can

The specific rainfall intensity determination will be described in step S3 with reference to FIG. 3 .

The memory 120 may store a program including one or more commands for calculating a rainfall amount from a rainfall image acquired through the communication unit 130 . The processor 110 may execute the rainfall calculation process according to the embodiment based on the program and instructions stored in the memory 120 .

The memory 120 may further store intermediate data and calculation results that are calculated by the rainfall calculation algorithm and generated in the calculation process for calculating the rainfall amount.

Memory 120 may include internal memory and/or external memory, and may include volatile memory such as DRAM, SRAM, or SDRAM, one time programmable ROM (OTPROM), PROM, EPROM, EEPROM, mask ROM, flash ROM, NAND Flash memory or non-volatile memory such as NOR flash memory, SSD, compact flash (CF) card, SD card, Micro-SD card, Mini-SD card, Xd card, or flash drive such as a memory stick; Alternatively, it may include a storage device such as an HDD. The memory 120 may include, but is not limited to, magnetic storage media or flash storage media.

The rainfall calculation apparatus 100 according to the embodiment may further include a communication unit 130 .

The communication unit 130 includes a communication interface for transmitting and receiving data of the rainfall calculation device 100 . The communication unit 130 may connect the rainfall calculation device 100 with the network 300 by providing various types of wired/wireless communication paths to the rainfall calculation device 100 .

The rainfall amount calculation apparatus 100 may transmit/receive input rainfall image data, resultant rainfall amount, and intermediate data through the communication unit 130 . The communication unit 130 may be configured to include, for example, at least one of various wireless Internet modules, a short-range communication module, a GPS module, a modem for mobile communication, and the like.

The wireless Internet module refers to a module for wireless Internet access. The wireless Internet module is configured to transmit and receive wireless signals in a communication network according to wireless Internet technologies.

As wireless Internet technology, for example, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), High Speed Downlink Packet Access (HSDPA), High Speed Uplink Packet Access (HSUPA), Long Term Evolution (LTE), Long Term Evolution-Advanced (LTE-A), and the like.

The short-range communication module is for short-range communication, and is Bluetooth (Bluetooth??), RFID (Radio Frequency Identification), infrared communication (Infrared Data Association; IrDA), UWB (Ultra Wideband), ZigBee, NFC (Near) Field Communication), Wi-Fi (Wireless-Fidelity), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus) technology may be used to support short-distance communication.

The rainfall calculation apparatus 100 may further include a bus 140 that provides a physical/logical connection path between the processor 110 , the memory 120 , and the communication unit 130 .

The rainfall calculation apparatus 100 according to the embodiment confirms the presence of rain using a grayscale change in pixels of an input image, and applies an optimization technique to extract only the rain from the image background to select an optimal parameter in a specific target site do.

Thereafter, the rainfall calculation apparatus 100 according to the embodiment determines the rainfall intensity by using physical characteristics such as the speed and size of the rain stream included in the input image, and the distance between the camera lens and the raindrops, and calculates the rainfall amount therefrom.

Hereinafter, a rainfall calculation method according to an embodiment will be described in more detail with reference to FIGS. 3 to 6 .

3 is a flowchart of a method for calculating rainfall according to an embodiment.

The rainfall calculation method according to the embodiment includes the steps of converting a series of frames of a rainfall image into at least one grayscale image (S1), at least one using a rain stream identification function defined based on a change in pixel value of the at least one grayscale image Separating at least one rain layer from the grayscale image of (S2) and determining the rainfall intensity of the rainfall image based on rainfall characteristic information of the rain region displayed in the at least one rain layer (S3).

The rainfall calculation apparatus 100 according to the embodiment confirms the existence of the rain stream using the grayscale change in the pixels of the input image by step S1, and an optimization technique to extract only the rain stream from the image background by step S2 is applied to select the optimal parameters in a specific target site.

In step S1 , the processor 110 may convert a series of frames of the rainfall image into at least one grayscale image.

The rainfall image includes a series of frames generated by the camera 200 photographing the periphery of the target site on which the camera 200 is disposed for a predetermined time with reference to FIG. 1 .

In an example, the camera 200 may periodically transmit a rainfall image to the rainfall calculation apparatus 100 . In an example, the processor 110 may perform a rainfall calculation process by dividing the input image received from the camera 200 at a predetermined time period. For example, the time period may be 30 seconds.

In one example, the camera 200 may be a Closed Circuit Television (CCTV) camera that is disposed at a predetermined location to photograph a target site and its surroundings. For example, the camera 200 may be a CCTV camera capable of capturing an infrared image.

The rain stream identification algorithm implemented in the rainfall calculation method according to the embodiment uses a rainfall image of an observation area to distinguish a rain layer from a background layer without rain. Specifically, the rain stream identification algorithm separates the rain stream and the background through a pixel change of raindrops in a plurality of frames of a photographed image.

In step S1 , the rainfall calculation apparatus 100 may convert a rainfall image including a series of frames acquired by the camera 200 into at least one grayscale image. Here, each frame of the rainfall image is an image having an RGB channel, and the grayscale image means a grayscale image.

For example, assuming there are three consecutive frames, the RGB image of each frame is converted into an 8-bit grayscale image (0 black-255 white).

For example, the processor 110 may acquire at least one grayscale image by converting a series of frames of the rainfall image to have the same luminance (or relative luminance) as the original color image using a photometric method or a color measurement method. For example, for grayscale conversion, a tool that converts RGB of a rainfall image into luminance and chroma (eg, matlab's rgb2ycbcr function) can be used.

Step S1 may include a preprocessing process for a series of frames of the rainfall image.

For example, the processor 110 may convert at least one frame obtained by sampling a series of frames at a constant period into a grayscale image. For example, the processor 110 may convert at least one frame generated by grouping and averaging a series of frames at regular intervals into a grayscale image. That is, the processor 110 may generate a grayscale image from a number of frames smaller than the number of frames in a series of frames.

Additionally, step S1 may further include a pre-processing of setting a rainfall region in the rainfall image.

For example, the processor 110 may set a region in the image that excludes a region in which other objects can move and a region in which raindrops are not recognized due to raindrops on the camera lens as the rainfall region. For example, the processor 110 may set a region of an image having the least movement of an object and the darkest background as a rainfall region. For example, the processor 110 may set the lower center of the image as the rainfall area.

In step S2 , the processor 110 may separate the at least one rain layer from the at least one gray scale image by using a rain stream identification function defined based on a change in the pixel value of the at least one gray scale image.

에 의해 정의된다.The grayscale image including raindrops in one frame converted in step S1 is a two-dimensional matrix in which all pixel elements are composed of grayscale values.

is defined by

)는 빗방울이 없는 배경 레이어(

)와 빗줄기만 있는 레이어(

)로 나뉘게 된다.procession(

) is the background layer without raindrops (

) and a layer with rain only (

) is divided into

In step S2 , the processor 110 executes a rain stream identification algorithm to separate a rain layer and a background layer without rain from the at least one grayscale image converted in step S1 .

Hereinafter, step S2 will be described in detail with reference to FIGS. 4 and 5 .

4 is a flowchart of a rain layer separation process according to an embodiment.

Referring to FIG. 3 , in step S2 , the processor 110 separates the at least one rain layer from the at least one gray-scale image by using a rain rain identification function defined based on a change in the pixel value of the at least one gray-scale image.

To this end, in step S2, the first adjustment variable for the first operation term, the second adjustment variable for the second operation term, the third adjustment variable for the third operation term, and the fourth operation for minimizing the rain stream identification function value Determining the combination of the fourth adjustment variable for the term (S21) and determining the first matrix that minimizes the rain stream identification function value by the determined combination as the rain stream layer for the grayscale image (S22) have.

In step S21, the processor 110 generates a first adjustment variable for the first operation term, a second adjustment variable for the second operation term, a third adjustment variable for the third operation term, and A combination of the fourth control variable for the fourth operation term may be determined.

First, let's look at the rain stream identification function.

The rainfall calculation method according to the embodiment designs an optimization model in consideration of spatial density, vertical continuity, horizontal discontinuity, and time dependence of a rain layer separated from at least one grayscale image converted from a rainfall image, and a rain stream identification function for this make up In addition, the rainfall calculation method according to the embodiment selects an optimal adjustment variable (ie, step S21), and separates the rain layer from the background layer without rain in the rainfall image (ie, step S22)) carry out

The rain streak identification function is an optimization function using, as a decision variable, a first matrix having pixel values of a rain streak layer with respect to a grayscale image as a component. Equation 1 below is an example of a rain stream identification function used in the method for calculating rainfall according to an embodiment.

Here, O denotes a matrix having pixel values of a grayscale image as a component, and R denotes a first matrix having pixel values of a rain layer as a component.

는 최적화 모델의 결정변수 제약 조건을 나타내고 있으며 R값은 음수가 될 수 없다는 것을 나타낸다.In the optimization model represented by the rain stream identification function of Equation 1, R is a determinant variable of the optimization model.

represents the determinant constraint of the optimization model, and indicates that the R value cannot be negative.

은 매트릭스의 모든 요소의 절댓값의 합을 의미한다.

_,

_,

_,

항은 R(빗줄기 레이어)과 B(배경 레이어)의 특성들을 일반화한 항이다.The initial R matrix is composed of a two-dimensional matrix composed only of grayscale values from the initial image.

is the sum of the absolute values of all elements of the matrix.

_,

_,

_,

The term is a generalized term for the properties of R (rain layer) and B (background layer).

)는 복수의 연산항의 조절 변수로서, 0이 아닌 양의 값을 가진다.Multiple control variables (

) is a control variable of a plurality of operands, and has a non-zero positive value.

The rain stream identification function of Equation 1 includes four operation terms. Each operand is associated with spatial density, vertical continuity, horizontal continuity, and temporal dependence of the image.

That is, the plurality of operation terms of the rain stream identification function include a first operation term for the spatial density of the first matrix, a second operation term for the vertical smoothness of the first matrix, and the remainder of separating the rain layer from the grayscale image. It may include a third operation term for horizontal continuity of the second matrix having pixel values of the background layer as a component, and a fourth operation term for temporal dependence of the second matrix.

)은 제 1 행렬(R)의 공간 밀도와 연계된 연산항으로서, 제 1 행렬(R)의 픽셀값의 L1-norm과 제 1 조절 변수(

)에 기반하여(예를 들어 곱셈으로) 값이 결정된다.The first operand (

) is an operation term associated with the spatial density of the first matrix R, and L1-norm of the pixel values of the first matrix R and the first adjustment variable (

) (eg by multiplication), the value is determined.

)의 값을 최소화시킨다는 것은 행렬 내의 값이 대부분 0인 희소행렬(Sparse of matrix)을 나타내어 배경 레이어(B)와 빗방울 레이어(R)의 차이를 분명하게 한다는 것을 의미한다.The layer with only raindrops (R) has more zero grayscale values than the background layer without raindrops (B). Therefore, the first operand (

) means that the difference between the background layer (B) and the raindrop layer (R) is made clear by representing a sparse matrix in which the values in the matrix are mostly 0.

)은 제 1 행렬(R)의 수직 방향 연속도, 즉 수직 방향 평활도(smoothness)와 연계된 연산항으로서, 제 1 행렬(R)의 픽셀값의 y축 방향 변화량(

)에 대한 L1-norm과 제 2 조절 변수(

)에 기반하여(예를 들어 곱셈으로) 값이 결정된다.The second operand (

) is an operation term associated with the vertical continuity of the first matrix R, that is, the vertical smoothness, and the amount of change in the y-axis direction of the pixel value of the first matrix R (

L1-norm for ) and the second control variable (

) (eg by multiplication), the value is determined.

)은 세로 방향의 빗줄기 모양을 의미한다. 제 2 연산항의 델 연산자(

)는 Y 방향 변화를 나타내고, 제 2 연산항(

)은 빗방울 레이어(R)의 픽셀값의 Y방향 총 변화를 나타낸다.The second operand (

) means a vertical rain stream. The del operator of the second operand (

) represents the change in the Y direction, and the second operation term (

) represents the total change in the Y-direction of pixel values of the raindrop layer (R).

)을 최소화한다는 것은 세로 방향으로 내리는 비와 배경의 차이를 잘 감지하게 만들어 빗줄기와 배경을 효과적으로 분리한다는 것을 의미한다.The second operand (

) means that it effectively separates the rain from the background by making the difference between the vertical rain and the background well sensed.

)은 회색조 이미지의 픽셀값을 성분으로 행렬(O)과 제 1 행렬(R)의 픽셀값의 차이에 대응하는 제 2 행렬(

)의 수평방향 연속도, 즉 제 2 행렬의 픽셀값의 x축 방향 변화량(

에 대한 L1-norm과 제 3 조절 변수(

)에 기반하여(예를 들어 곱셈으로) 값이 결정된다.The third operand (

) is a second matrix (

) in the horizontal direction, that is, the amount of change in the x-axis direction of the pixel values of the second matrix (

L1-norm and the third control variable for

) (eg by multiplication), the value is determined.

)은 제 2 행렬(

)의 시간 의존도, 즉 시간 독립도와 연계된 항으로, 시간에 따른 제 2 행렬(

)의 픽셀값의 변화량(

)에 대한 L1-norm과 제 4 조절변수(

)에 기반하여(예를 들어 곱셈으로) 값이 결정된다.4th operation term (

) is the second matrix (

) is a term associated with the time dependence, i.e., time independence, of the second matrix (

) change in pixel value (

L1-norm for ) and the fourth moderator (

) (eg by multiplication), the value is determined.

That is, according to the fourth operation term, the background layer and the rain layer may be separated by considering the difference in pixel values of not only each still frame but also consecutive frames.

)은 배경 레이어(B)의 가로 방향 지속성 보존이 목적이고, 제 4 연산항(

)은 배경 레이어(B)의 시간 연속성 유지가 목적이다.The third operand (

) is for the purpose of preserving the horizontal persistence of the background layer (B), and the fourth operation term (

) is for the purpose of maintaining temporal continuity of the background layer (B).

) 및 제 4 연산항(

)을 최소화한다는 것은 X 방향의 변화 및 시간의 변화에 따라 배경 레이어(B)가 변하지 않는다는 것을 의미하므로, 이들 항을 최소화함으로써 확실한 빗줄기 레이어(R)와 배경 레이어(B)와의 분리가 가능해진다.The third operand (

) and the fourth operand (

) means that the background layer (B) does not change according to the change in the X direction and the change in time.

The rain stream identification function includes a plurality of operation terms associated with a change in a pixel value of at least one grayscale image, and adjustment variables for the plurality of operation terms. As shown in the example of Equation 1, for example, the rain stream identification function has a first operation term, a second operation term, and a third operation by the first adjustment variable, the second adjustment variable, the third adjustment variable, and the fourth adjustment variable. The weighted sum of the term and the fourth operation term is calculated as a function value.

In step S21, the processor 110 repeatedly calculates the rain stream identification function with respect to the given at least one grayscale image and the first matrix, and minimizes the function value of the rain stream identification function, a first adjustment variable, a second adjustment variable, A combination of a third conditioning variable and a fourth conditioning variable may be determined.

In one example, for the optimization of the rain stream identification function, a minimization technique that repeats until reaching the lowest point that can no longer go down by moving the adjustment variable from the current position to the direction in which the function value decreases among optimization techniques can be applied. , but is not limited thereto, and various optimization techniques may be used.

)는 복수의 연산항의 조절 변수로서, 0이 아닌 양의 값을 가진다. 빗줄기 식별 함수값을 최소화시키는 조절 변수(

)의 조합을 찾는 것은 강우 영상에서 배경(B)과 비(R)를 명확하게 분리하는 데에 중요한 요소이다.Multiple control variables (

) is a control variable of a plurality of operands, and has a non-zero positive value. An adjustment variable that minimizes the rainstorm identification function value (

) is an important factor in clearly separating the background (B) and the rain (R) in the rainfall image.

)의 값은 한 번에 각 연산항의 값이 얼마나 이동할 지를 정하는 조절 매개변수이며 이 값이 증가할수록 최소값으로 수렴하는 속도가 증가하지만, 값이 너무 커지면 발산할 수도 있기 때문에 적절한 매개변수의 선정이 배경 레이어(B)와 빗줄기 레이어(R)의 명확한 분리에 중요한 영향을 미친다.Each control variable (

) is a control parameter that determines how far the value of each operation term moves at a time. As this value increases, the speed of convergence to the minimum value increases, but if the value becomes too large, it may diverge. It has a significant impact on the clear separation of the layer (B) and the raindrop layer (R).

) 및 제 3 조절변수(

)를 조절한 후에 제 2 조절 변수(

) 및 제 4 조절 변수(

)를 조절할 수 있으며, 이에 제한되는 것은 아니다. 일 예에서, 조절 변수는 동시에 조절될 수 있다. 일 예에서 각 조절 변수는

조건을 만족하는 범위에서 조절될 수 있다.In one example, the adjustment variable may be adjusted sequentially. In one example, the adjustment variable may be adjusted according to a predetermined order. For example, first the first adjustment variable (

) and the third control variable (

) after adjusting the second adjustment variable (

) and the fourth adjustment variable (

) can be adjusted, but is not limited thereto. In one example, the adjustment variable may be adjusted simultaneously. In one example, each modulating variable is

It can be adjusted in a range that satisfies the condition.

The determined first adjustment variable, the second adjustment variable, the third adjustment variable, and the fourth adjustment variable become a unique combination for the target location in which the camera 200 capturing the rainfall image is located.

Meanwhile, the processor 110 may differently determine and manage an optimal combination of the first adjustment variable, the second adjustment variable, the third adjustment variable, and the fourth adjustment variable for the target site according to conditions. For example, the processor 110 may separately determine an optimal combination of the first adjustment variable, the second adjustment variable, the third adjustment variable, and the fourth adjustment variable according to the background of the image capturing point or the intensity of rainfall. It can be applied to the rain stream identification function.

In step S22 , the processor 110 may determine the first matrix that minimizes the value of the rain stream identification function by the optimal combination of the adjustment variables determined in step S21 as the rain layer for the grayscale image.

)을 변화시켜가면서 빗줄기 조절 함수를 반복적으로 연산하여 빗줄기 조절 함수값이 최소가 되는 제 1 행렬(

)을 회색조 이미지(

)에 대한 빗줄기 레이어로 결정할 수 있다.In step S22, the processor 110 generates a first matrix (

) by repeatedly calculating the rain control function while changing the first matrix (

) to a grayscale image (

) can be determined as the rain layer for

5 is a diagram for explaining a rain layer separation process of a method for calculating rainfall according to an embodiment.

Referring to FIG. 3 , in step S1 , the processor 110 generates at least one grayscale image O1 , O2 , ..., OT converted from a series of frames of a rainfall image.

Referring to FIG. 4 , in step S21 , the processor 110 determines the first adjustment variable for the first operation term, the second adjustment variable for the second operation term, and the third operation term for minimizing the rain stream identification function value. A combination of the third control variable and the fourth control variable for the fourth operation term is determined.

To this end, the processor 110 may determine a combination of the first adjustment variable, the second adjustment variable, the third adjustment variable, and the fourth adjustment variable that minimizes the function value of the rain identification function while repeatedly calculating the rain stream identification function. .

Referring to FIG. 4 , in step S22 , the processor 110 may determine the first matrix that minimizes the rain stream identification function value by the determined combination as the rain stream layer for the grayscale image.

As described above, the plurality of operation terms of the rain stream identification function are respectively associated with spatial density, vertical continuity, horizontal continuity, and temporal dependence of the image.

)에 대한 제 2 연산항, 회색조 이미지에서 빗줄기 레이어(R1, R2,..., RT)를 분리한 나머지에 대응하는 배경 레이어(B1, B2,..., BT)의 픽셀값을 성분으로 갖는 행렬의 수평 연속도(

)에 대한 제 3 연산항, 및 배경 레이어 행렬(B1, B2,..., BT)의 시간 의존도(

)에 대한 제 4 연산항을 포함할 수 있다.The plurality of operation terms of the rain identification function include a first operation term for spatial density of matrices R1, R2, ..., RT for the rain layer, and a matrix for the rain layer (R1, R2, ..., RT) of the vertical smoothness (

), the pixel value of the background layer (B1, B2,..., BT) corresponding to the remainder of separating the rain layer (R1, R2,..., RT) from the grayscale image as a component The horizontal continuity of a matrix with (

), and the temporal dependence of the background layer matrices (B1, B2,..., BT)

) may include a fourth operation term for .

In step S22, the processor 110 determines an optimal first matrix as a rain layer.

Returning to FIG. 3 again, step S3 is looked at.

With respect to the at least one rain layer separated in step S2, the rainfall calculation apparatus 100 according to the embodiment performs the physical properties such as the speed, size, and distance between the camera lens and the raindrops included in the input image in step S3. The rainfall intensity is determined using the phosphorus characteristics and the rainfall amount is calculated from this.

The rainfall intensity calculation algorithm implemented by the rainfall calculation method according to the embodiment includes the physical characteristics of raindrops (eg, terminal velocity, diameter, spatial distribution, etc.) of the raindrops and the set values of the camera (eg, F-value, focal length, Resolution, etc.), the actual length of the rain stream is calculated, and the rainfall is calculated by estimating the size distribution of raindrops. At this time, the rainfall intensity is estimated by combining the physical characteristics of the rain stream and raindrops of the separated binary image by applying the optimal adjustment variable selected in the above-described rain stream identification algorithm.

In operation S3 , the processor 110 may determine the rainfall intensity of the rainfall image based on rainfall characteristic information of the rain region displayed in the at least one rain layer. Here, at least one rain layer is separated in step S2. For example, the at least one rain layer in step S3 may be a binary image (eg, rain pixel 1, remaining pixel 0) of the rain layer separated in step S2 .

Hereinafter, step S3 will be described in detail with reference to FIG. 6 .

6 is a flowchart of a rainfall intensity calculation process according to an embodiment.

Referring to FIG. 3 , in step S3 , the processor 110 may determine a rainfall intensity of a rainfall image based on rainfall characteristic information of a rain region displayed in at least one rain layer.

Here, the rainfall characteristic information includes the actual size of the rain stream corresponding to the rain region displayed in the at least one rain layer, the speed of the rain stream, and a rainfall particle size distribution for the at least one rain layer. This may be estimated or determined in steps S31 to S36 to be described later.

Step (S3) is a step (S31) of estimating the actual distance between the lens that captured the rainfall image and the raindrop corresponding to the raindrop area displayed in at least one raindrop layer, and determining the actual size of the raindrop based on the estimated actual distance. It may include step S32 and determining the speed of the rain stream based on the determined actual size of the rain stream ( S33 ).

In step S31 , the processor 110 may estimate the actual distance s between the lens that captures the rainfall image and the rain that corresponds to the rain region displayed in the at least one rain layer.

The raindrop size is estimated based on the distance between each raindrop and the camera lens in the image, and the instantaneous raindrop velocity at close range is calculated by the camera exposure time and the actual raindrop size. However, since the distance (s) between the actual raindrop and the camera lens is unknown, it is necessary to estimate it before estimating the rainfall.

First, assuming that the raindrop is spherical, the trajectory length (L, mm) of the raindrop and the diameter (D, mm) of the raindrop can be known through the lens equation.

는 초점 거리(Focal distance, mm), f는 초점 길이(Focal length, mm),

와

는 각각 이미징 센서 활성 영역의 수직 및 수평 크기(mm),

와

는 각각 캡쳐된 이미지의 수직 및 수평 크기(pixel),

와

는 이미지에서 빗줄기의 길이와 너비를 뜻한다.7 and Equation 2, s is the distance from the raindrop to the lens (mm),

is the focal length (mm), f is the focal length (mm),

Wow

are the vertical and horizontal dimensions (mm) of the imaging sensor active area, respectively;

Wow

is the vertical and horizontal size of the captured image (pixel),

Wow

is the length and width of the rain stream in the image.

At this time, the falling velocity v(s) (m/s) of the raindrop is expressed as a function of the exposure time τ.

However, the fall velocity is generally assumed to be the instantaneous velocity of a raindrop on the ground, which means the highest velocity achieved during the fall due to the balance between gravity and atmospheric drag. The terminal velocity of raindrops approximated using this is as follows.

D is the diameter of the raindrop, which is the same as D(s) described in Equation 2 above.

When v(s) = v(D), the distance (s) between the raindrop and the lens required for calculating L(s) and D(s) can be obtained using Equation 5 below.

,

,

를 의미한다. w(·)는 람베르트 w 함수로, f(w)=we^w 의 역 관계 함수이다. 위의 수학식 5를 보면 s를 제외한 모든 변수는 도 7에 나타나 있는 변수로 변수들의 값을 입력하여 화면에 나타나는 빗줄기

의 s값을 계산할 수 있다.In this case, A, B, and C are each

,

,

means w(·) is a Lambert w function, which is an inverse relational function of f(w)=we ^w . In Equation 5 above, all variables except s are variables shown in FIG. 7, and the rain stream appears on the screen by inputting the values of the variables.

can calculate the value of s.

In step S32 , the processor 110 may determine the actual size of the rain stream based on the estimated actual distance.

In step S32 , the processor 110 may determine the actual size of the rain stream according to Equation 2 described above.

) 및 해당 빗방울의 직경(

)을 결정할 수 있다.For example, the processor 110 may measure the actual length (

) and the diameter of the corresponding raindrop (

) can be determined.

In step S33 , the processor 110 may determine the speed of the rain stream based on the determined actual size of the rain stream.

In step S33 , the processor 110 may determine the speed of the rain stream based on Equations 3 and 4 described above.

)를 결정할 수 있다.For example, the processor 110 determines the speed (

) can be determined.

Additionally, with reference to FIG. 3 , step S3 is a step of determining an image depth of a rainfall image based on a parameter of a lens photographing the rainfall image ( S34 ), and the validity of at least one rain layer based on the determined image depth Determining a rainfall volume (S35), determining a rainfall particle size distribution of at least one raindrop layer for the determined effective rainfall volume (S36) and based on the actual size of the raindrop, the speed of the rainstorm, and the rainfall particle size distribution The method may further include determining a rainfall intensity for the rainfall image ( S37 ).

In step S34 , the processor 110 may determine the image depth of the rainfall image based on the parameters of the lens photographing the rainfall image.

과 원거리

사이의 범위에서만 s 값이 추정 가능하다.The actual distance s between the lens that captured the rainfall image estimated in step S31 and the rain corresponding to the rain region displayed in the at least one rain layer is maintained only for rain within a specific range DoF. that is, near

and far

The value of s can be estimated only in the range between

는 카메라 기종에 따른 최대 허용 착란원(mm)이며, N은 카메라 렌즈의 F 수이다.

is the maximum allowable source of confusion (mm) according to the camera model, and N is the F number of the camera lens.

과 원거리

사이의 범위를 이미지 심도로 결정할 수 있다.In step S34, the processor 110 according to Equation 6

and far

The range between can be determined as the image depth.

In operation S35 , the processor 110 may determine an effective rainfall volume for at least one rain layer based on the determined image depth.

In step S35, the theoretical rainfall volume V(m ³ ) in the rainfall image means a rectangular pyramid-shaped volume within the depth of field (DoF with reference to FIG. 7 ) formed on the focal plane of the camera.

However, a difference may occur between the theoretical rainfall and actual rainfall calculated in this way, and the difference can be reduced by determining the particle size distribution and reflecting the rainfall intensity estimation in the following steps.

In operation S36 , the processor 110 may determine a rainfall particle size distribution of at least one raindrop layer with respect to the determined effective rainfall volume.

In step S36, the processor 110 calculates the rainfall intensity (ie, the amount of rainfall per hour) by the drop size distribution (DSD).

The particle size distribution is expressed as the concentration of the number of precipitation particles per unit volume per diameter interval of the precipitation particles and may show different characteristics depending on the precipitation system.

For example, a probability density function to which a gamma distribution is applied may be used, and the DSD may be expressed as follows.

는 a 일 때의 gamma 함수,

는 a-1,

는 1/b를 의미한다. D는 빗방울의 직경을 나타낸다.N(D) is the number of raindrops in the depth-of-field measurement volume, a is the shape of the probability density function, b is the scale parameter,

is the gamma function when a,

is a-1,

means 1/b. D represents the diameter of the raindrop.

을 얻는다.At this time, in order to reduce the uncertainty due to the limited image size, the sample distribution of N(D) for consecutive frames is averaged to reduce the uncertainty due to the limited image size.

get

은 j번째 프레임의 i번째 픽셀에 있는 빗방울의 표본 개수이다.

is the sample number of raindrops in the i-th pixel of the j-th frame.

In operation S37 , the processor 110 may determine the rainfall intensity for the rainfall image based on the actual size of the rain stream, the speed of the rain stream, and the rain particle size distribution.

The calculation formula of the rainfall intensity RF (mm/hr) may be expressed by Equation (10).

Therefore, the final rainfall intensity calculation formula is obtained by substituting the formula for the terminal velocity v(D) of the raindrop (see Equation 4) and the formula for DSD (refer to Equation 9) into the rainfall intensity RF calculation formula (Equation 10). can

In an example, the processor 110 may calculate the total rainfall by integrating the rainfall intensity of Equation 11 over time.

7 is a view for explaining an index used in a rainfall intensity calculation process of the rainfall amount calculation method according to the embodiment.

The actual rain stream (length L, diameter D) located between the near focal plane Sn and the far focal plane Sf is imaged as a rain region (length lp, diameter dp) on the sensor plane.

8 exemplarily shows a rain layer separation result of a rainfall calculation process according to an embodiment.

The image shown in FIG. 8 is a rain image separated by a rain stream identification algorithm in step S2 with reference to FIG. 3 of the method for calculating rainfall according to the embodiment. (a) shows the current image, (b) shows the detected rain stream, and (c) shows the binarized image.

From FIG. 8 , it can be seen that raindrops are effectively separated from the field images taken during the day by the rainfall calculation method according to the embodiment.

9 exemplarily shows a rain layer separation result of a rainfall calculation process according to an embodiment.

The image shown in FIG. 9 is a rain image separated by a rain stream identification algorithm in step S2 with reference to FIG. 3 of the method for calculating rainfall according to the embodiment. (a) shows the current image, (b) shows the detected rain stream, and (c) shows the binarized image.

From FIG. 9 , it can be seen that raindrops are effectively separated from an infrared image captured at night by the rainfall calculation method according to the embodiment.

10 is a graph exemplarily showing a rainfall calculation result according to an embodiment.

10 is a graph of a result of calculating a rainfall image generated on June 30, 2020 using a rainfall calculation algorithm according to an embodiment.

The experiment was conducted on the rooftop of the atmospheric environment observatory at Seoul National University Gwanak Campus, and both the CCTV (Example) and Gwanak AWS used for image analysis exist in the same location. Precipitation measured by CCTV and AWS (Automatic Weather System) of the Korea Meteorological Administration was expressed as 15-minute cumulative values, respectively.

It can be seen that the rainfall calculation method according to the embodiment can perform the nighttime rainfall calculation using the infrared camera image with high accuracy and can calculate the rainfall amount that can be quantitatively estimated. In addition, it is possible to perform rainfall state determination with high accuracy while using a cheaper sensor.

According to the embodiment, it is possible to check the amount of rainfall in a precipitation unmeasured area where there is no meteorological observation equipment by analyzing the stored image using video recording equipment already installed at various points nationwide, such as a CCTV camera. Through this, it is possible to accurately observe rainfall events not only in urban areas but also in local areas such as mountains, roads, and coastal areas using cheaper observation equipment.

In addition, according to the embodiment, since rainfall is produced in units of minutes and hours, it has dense spatial and temporal resolution. Therefore, it can be used as essential basic data for model development for weather forecasting and water resource forecasting, so that the quality of the output of the forecasting model in each field can be improved.

There is an advantage that can be used in all industrial fields that require local rainfall information at low cost because the existing CCTV is installed or the rainfall calculation technology according to the embodiment can be applied by installing the CCTV outside.

In particular, if the CCTV installed on the road is utilized, it will be helpful in the production or prediction of dangerous weather information on general roads and highways, and it can be used in the road hazardous weather information monitoring system.

In addition, it can be used as real-time monitoring data in the flood prediction system. Through this, it is possible to take a more detailed customized and preemptive preventive response, and it can be helpful in resolving flood-prone areas where accidents are high.

Meanwhile, the precipitation data generated by the embodiment may be utilized as high-resolution precipitation basic data for modeling for weather prediction (weather forecast).

In addition, since it is a technology that can secure rainfall data in the non-measured area where there is no rainfall data, it can be applied for effective disaster prevention.

The rainfall calculation method according to an embodiment of the present invention described above can be implemented as computer-readable codes on a medium in which a program is recorded. The computer-readable medium includes all kinds of recording devices in which data readable by a computer system is stored. Examples of computer-readable media include Hard Disk Drive (HDD), Solid State Disk (SSD), Silicon Disk Drive (SDD), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. There is this.

The description of the embodiment of the present invention described above is for illustration, and those of ordinary skill in the art to which the present invention pertains can easily transform into other specific forms without changing the technical spirit or essential features of the present invention. you will be able to understand that Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and likewise components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included in the scope of the present invention.

100: rainfall calculator
110: processor
120: memory
130: communication department

Claims (17)

converting a series of frames of the rainfall image into at least one grayscale image;
separating at least one raindrop layer from the at least one grayscale image using a raindrop identification function defined based on a change in pixel values of the at least one grayscale image; and
determining the rainfall intensity of the rainfall image based on rainfall characteristic information of the rain region displayed in the at least one rain layer
containing,
How to calculate rainfall.
The method of claim 1,
The rain stream identification function is,
It is an optimization function using the first matrix having the pixel values of the rain layer for the grayscale image as a component as a decision variable,
including a plurality of operation terms associated with the pixel value change and adjustment variables for the plurality of operation terms,
How to calculate rainfall.
3. The method of claim 2,
The plurality of operation terms are
a first operation term for the spatial density of the first matrix;
a second operation term for the vertical smoothness of the first matrix;
A third operation term for the horizontal continuity of the second matrix having, as a component, the pixel value of the background layer corresponding to the remainder of separating the rain layer from the grayscale image, and
A fourth operation term for the time dependence of the second matrix
containing,
How to calculate rainfall.
4. The method of claim 3,
The first operation term, the second operation term, the third operation term, and the fourth operation term are L1-norm operation terms associated with the first matrix or the second matrix, respectively,
How to calculate rainfall.
4. The method of claim 3,
Separating the at least one rain layer comprises:
A first adjustment variable for the first operation term, a second adjustment variable for the second operation term, a third adjustment variable for the third operation term, and the fourth operation for minimizing the function value of the rain stream identification function determining a combination of a fourth modulator for the term; and
determining the first matrix that minimizes the rain stream identification function value by the combination as the rain stream layer for the grayscale image
containing,
How to calculate rainfall.
The method of claim 1,
The rainfall characteristic information,
The actual size of the rain stream corresponding to the rain region indicated in the at least one rain layer, the speed of the rain stream, and the rainfall particle size distribution for the at least one rain layer
containing,
How to calculate rainfall.
The method of claim 1,
The step of determining the rainfall intensity,
estimating an actual distance between the lens photographing the rainfall image and rain corresponding to the rain region displayed in the at least one rain layer;
determining an actual size of the rain stream based on the distance; and
determining the speed of the rain stream based on the actual size of the rain stream,
How to calculate rainfall.
8. The method of claim 7,
The step of determining the rainfall intensity,
determining an image depth of the rainfall image based on a parameter of a lens photographing the rainfall image;
determining an effective rainfall volume for the at least one raindrop layer based on the image depth;
determining a rainfall particle size distribution of the at least one raindrop layer with respect to the effective rainfall volume; and
determining the rainfall intensity for the rainfall image based on the actual size of the rain stream, the speed of the rain stream, and the rain particle size distribution;
further comprising,
How to calculate rainfall.
A computer-readable recording medium recording a computer program comprising one or more instructions for executing the method for calculating rainfall according to any one of claims 1 to 8.
a memory for storing rainfall images including a series of frames; and
processor
Including, the processor,
converting the series of frames into at least one grayscale image;
separating at least one rain layer from the at least one gray scale image using a rain stream identification function defined based on a change in pixel values of the at least one gray scale image;
configured to determine the rainfall intensity of the rainfall image based on rainfall characteristic information of the rain region displayed in the at least one rain layer,
Rainfall calculator.
11. The method of claim 10,
The rain stream identification function is,
It is an optimization function using the first matrix having the pixel values of the rain layer for the grayscale image as a component as a decision variable,
including a plurality of operation terms associated with the pixel value change and adjustment variables for the plurality of operation terms,
Rainfall calculator.
12. The method of claim 11,
The plurality of operation terms are
a first operation term for the spatial density of the first matrix;
a second operation term for the vertical smoothness of the first matrix;
A third operation term for the horizontal continuity of the second matrix having, as a component, the pixel value of the background layer corresponding to the remainder of separating the rain layer from the grayscale image, and
A fourth operation term for the time dependence of the second matrix
comprising,
Rainfall calculator.
13. The method of claim 12,
The first operation term, the second operation term, the third operation term, and the fourth operation term are L1-norm operation terms associated with the first matrix or the second matrix, respectively,
Rainfall calculator.
13. The method of claim 12,
The processor is
In order to separate the at least one raindrop layer,
A first adjustment variable for the first operation term, a second adjustment variable for the second operation term, a third adjustment variable for the third operation term, and the fourth operation for minimizing the function value of the rain stream identification function determining a combination of a fourth modulator for the term,
configured to determine the first matrix that minimizes the rain stream identification function value by the combination as the rain stream layer for the grayscale image,
Rainfall calculator.
11. The method of claim 10,
The rainfall characteristic information is
The actual size of the rain stream corresponding to the rain region indicated in the at least one rain layer, the speed of the rain stream, and the rainfall particle size distribution for the at least one rain layer
containing,
Rainfall calculator.
11. The method of claim 10,
The processor is
To determine the rainfall intensity,
estimating the actual distance between the lens photographing the rainfall image and the rain corresponding to the rain region displayed in the at least one rain layer,
determining the actual size of the raindrop based on the distance,
configured to determine the speed of the rain stream based on the actual size of the rain stream;
Rainfall calculator.
17. The method of claim 16,
The processor is
To determine the rainfall intensity,
Determining the image depth of the rainfall image based on the parameters of the lens photographing the rainfall image,
determine an effective rainfall volume for the at least one raindrop layer based on the image depth;
determine a rainfall particle size distribution of the at least one raindrop layer with respect to the effective rainfall volume;
and determine rainfall intensity for the rainfall image based on the actual size of the rain stream, the velocity of the rain stream, and the rainfall particle size distribution.
Rainfall calculator.

Images