KR101924278B1 - Apparatus and method for extraction of tidal creeks in tidal flat zones using drone - Google Patents

Abstract

The apparatus for extracting the tugboat information in the tidal-flat area using the drone according to an embodiment of the present invention includes a ground reference point surveying unit for performing the ground reference point survey of the tidal flat area using the DGPS (Differential Global Positioning System) For the purpose of triangulation, a drone for capturing aerial images of a tidal flat area at regular intervals, a numerical surface model generating unit for generating a numerical surface model using aerial images taken, a generated numerical surface model, A gradient image generating unit for generating a gradient image by using a difference in brightness value between adjacent pixels in the generated orthoimage image, a gradient image generating unit for generating a gradient image using the generated gradient image, An image segmentation unit for dividing the image into a plurality of image segments by applying a watershed algorithm to the image, For each piece, if the spectral characteristics of adjacent image pieces are equal to or greater than a certain degree of similarity, the image piece merging unit for merging the adjacent image pieces and the characteristics of the pixels constituting each piece of the image pieces obtained through the image piece merging unit are extracted The proposed method is based on the recognition criteria of the water based classification criterion for learning the water classification criterion and classifies the water area in the acquired image fragments by using the learned water classification criterion and merges the image fragments of the separated water area with the object adjacency analysis And a tidal flat contour extracting unit for extracting a contour line of the tidal flat area by superimposing the extracted water boundary on a time basis.

Description

[0001] APPARATUS AND METHOD FOR EXTRACTION OF TIDAL CREEK IN TIDAL FLAT ZONES USING DRONE [0002]

The present invention relates to an apparatus and method for extracting tidal information in a tidal flat area using a dron. A tidal flat area is measured by using a drone, and a numerical surface model and a normal image The present invention relates to an apparatus and method for extracting tex1 information in a tidal flat area using a dron which obtains texel information by analyzing the features of the obtained numerical surface model and orthoimage image, and dividing and merging the features.

Recently, thanks to the development of drone technology and the development of image processing technology, there are increasing cases of making and using topographical information by using drone in remote sensing field. In particular, when the terrain information is constructed using the drone, it is possible to construct a high-resolution remote sensing data having a resolution of several centimeters at a desired time in a desired area at a relatively low cost.

In addition, remote sensing using a drone is easier than shooting with a conventional aircraft or satellite, and it is easy to adjust the gas to a desired region and can be taken at an altitude that is not influenced by clouds It has advantages. However, because of the use of the drone, it is possible to limit the camera area due to limitation of the battery capacity, limit the camera performance according to the weight limitation of the payload, difficulty in processing large capacity data, and difficulty in acquiring the ground reference point. Nevertheless, because of the advantages of drones in areas that are difficult to use in conventional remote sensing, remote sensing using drones is increasing. Examples include cultural property monitoring, forest and crop monitoring, red tide and greenery monitoring, monitoring for traffic control, disaster area monitoring, digital mapping and cadastral mapping.

In the coastal marine surveying field, there is an increasing tendency to apply remote sensing using drone. Coastal marine survey includes areas such as coastal survey, coastal change monitoring, water depth survey, tidal flat survey, and red tide monitoring. Among these tidal flats, various studies have been conducted due to the recent social interest, Guidelines for conservation and restoration and utilization are being prepared. Especially, in case of coastal west coast, the necessity of periodic monitoring is constantly evolving as the tidal-flat erosion phenomenon and damage are continuously caused by the periodic change of sea level by tide, the influence of wave invasion by seasons, Has been raised. In addition, since the tidal flats formed along the waterways in and around the tidal flats are the main points of occurrence of drowning accidents, which account for more than half of the marine safety accidents every year, construction and monitoring of terrain data, including tidal information, There is a need.

The present invention has been developed to solve the above-mentioned problems. The present invention provides an aerial triangulation method for a tidal flat area using a drone, obtains a numerical surface model and a normal image using the aerial triangulation results, The object of the present invention is to provide an apparatus and a method for extracting tibia information in a tidal flat area using a dron which analyzes the characteristics of pixels of a surface model and an orthoimage and divides and merges the tibia information to acquire the tibia information.

The apparatus for extracting the tugboat information in the tidal-flat area using the drone according to an embodiment of the present invention includes a ground reference point surveying unit for performing the ground reference point survey of the tidal flat area using the DGPS (Differential Global Positioning System) For the purpose of triangulation, a drone for capturing aerial images of a tidal flat area at regular intervals, a numerical surface model generating unit for generating a numerical surface model using aerial images taken, a generated numerical surface model, A gradient image generating unit for generating a gradient image by using a difference in brightness value between adjacent pixels in the generated orthoimage image, a gradient image generating unit for generating a gradient image using the generated gradient image, An image segmentation unit for dividing the image into a plurality of image segments by applying a watershed algorithm to the image, For each piece, if the spectral characteristics of adjacent image pieces are equal to or greater than a certain degree of similarity, the image piece merging unit for merging the adjacent image pieces and the characteristics of the pixels constituting each piece of the image pieces obtained through the image piece merging unit are extracted The proposed method is based on the recognition criteria of the water based classification criterion for learning the water classification criterion and classifies the water area in the acquired image fragments by using the learned water classification criterion and merges the image fragments of the separated water area with the object adjacency analysis And a tidal flat contour extracting unit for extracting a contour line of the tidal flat area by superimposing the extracted water boundary on a time basis.

In addition, the image fragment dividing unit may determine a threshold value by applying a scale level to the generated gradient image, and calculate a gradation value higher than the determined threshold value without considering a local minimum point having a gradient value lower than the determined threshold value The image can be divided into a plurality of image pieces using the local minimum point having the smallest point.

The image fragment merging unit may calculate an average spectral vector of a plurality of segmented image segments, measure the similarity, compare the measured similarity with a predetermined threshold value, and merge the image segments having a degree of similarity larger than a predetermined threshold value .

In addition, the water-based reference learning unit may extract at least one of the average brightness value, the area, and the texture property of the pixels constituting each piece, and analyze the water based criterion by analyzing using the supervisory classification method.

The method of extracting the tugboat information in the tidal flat area using the dron according to an embodiment of the present invention includes the steps of performing a ground reference point survey of the tidal flat area through a DGPS (Differential Global Positioning System) survey, A step of capturing an aerial image of a tidal flat area at a predetermined time interval using the captured aerial image, a step of generating a numerical surface model and an orthographic image using the aerial image taken, dividing the generated gradient image into a plurality of image segments by applying a watershed algorithm to the generated gradient image, and if the spectral characteristics of adjacent image segments are equal to or greater than a certain degree of similarity Merging adjacent image fragments, combining the image fragments obtained through the merging step, Extracting and analyzing the characteristics of the pixels constituting the image, and learning the water classification criterion. Using the learned water classification criterion, the water region is divided into the acquired image fragments, Extracting the boundary of the water system by merging through the analysis, and extracting the contour line of the tidal flat area by superimposing the extracted water system boundary on a time basis.

The step of dividing the generated gradient image into a plurality of image segments by applying a watershed algorithm may include the steps of determining a threshold value by applying a scale level to the generated gradient image, And dividing the image into a plurality of image pieces using a local minimum point having a gradient value higher than the determined threshold without considering a local minimum point having a gradient value.

If the spectral characteristics of adjacent image segments are equal to or greater than a certain degree of similarity for each of the plurality of image segments, the step of merging adjacent image segments may include calculating an average spectral vector of the plurality of segmented image segments, And comparing the similarity to a predetermined threshold value to merge the image pieces having a degree of similarity larger than a predetermined threshold value.

Also, the step of extracting and analyzing the characteristics of the pixels constituting each piece and learning the water classification criterion may include extracting at least one of the average brightness value, the area and the texture characteristic of the pixels constituting each piece, And a step of learning the water classification criterion.

Meanwhile, as an embodiment of the present invention, a computer-readable recording medium on which a program for causing the computer to execute the above-described method may be provided.

According to an apparatus and method for extracting tex1 information in a tidal flat area using a dron according to an embodiment of the present invention, a numerical surface model and an orthoimage can be obtained by using a rotor blade dron and a camera.

According to the apparatus and method for extracting the tugboat information in the tidal flat area utilizing the drones according to the embodiment of the present invention, it is possible to economically and efficiently extract the terrain information in the coastal area and acquire the tugboat information, .

1 is a block diagram showing a configuration of an apparatus 1000 for extracting tex1 information in a tidal area using a dron according to an embodiment of the present invention
2 is a view showing a rotor blade dron 20 and a camera 22 according to an embodiment of the present invention.
FIG. 3 is a view showing a coastal area located at Wansanri, Cheonggye-myeon, Muan-gun, Jeollanam-do, which is an aerial photographing area of a dron according to an embodiment of the present invention.
4 is a view illustrating a flight path of a drones according to an embodiment of the present invention.
FIG. 5 is a flowchart illustrating an operation of aerial triangulation with respect to a tidal flat area utilizing the drones 20 according to an embodiment of the present invention.
FIG. 6 is a diagram showing an orthoimage image and a numerical surface model (DSM) generated by photographing times (four times in total) according to an embodiment of the present invention.
FIG. 7A is an orthoimage image of the ground surface according to an embodiment of the present invention, and FIG. 7B is a diagram illustrating a gradient image generated by using an orthoimage image of the ground surface.
FIG. 8A is a diagram illustrating a height difference (gradation operation value) of a pixel of a gradient image according to an exemplary embodiment of the present invention, and FIG. 8B is a diagram illustrating a gradient value according to an exemplary embodiment of the present invention (Region 1 and region 2) by applying a watershed algorithm in a gradient image.
9A is a view showing a gradient image according to an embodiment of the present invention. FIG. 9B is a diagram illustrating a gradient image according to an exemplary embodiment of the present invention, in which a watershed algorithm is applied by applying a high- 9C is a diagram illustrating a watershed algorithm applying a low scale level to a gradient image according to an exemplary embodiment of the present invention.
10 is a view showing a result of dividing an image fragment into a watershed algorithm by applying a scale level of 50% to a gradient image of a tidal flat area according to an embodiment of the present invention.
FIG. 11 is a diagram showing a result of merging image fragments having a similarity level (merge level 70%) corresponding to the upper 70% by measuring similarity after applying the watershed algorithm according to an embodiment of the present invention.
FIG. 12 is a diagram showing the boundaries of the water boundaries extracted by the water boundary extracting unit 90 according to an embodiment of the present invention, by time (time interval).
FIG. 13 is a diagram showing the distribution of tidal area contour lines obtained by superimposing a water boundary at each time point according to an embodiment of the present invention.
14 is a flowchart illustrating a method of extracting tibia information in a tidal flat area using a dron according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

The terms used in this specification will be briefly described and the present invention will be described in detail.

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiments. Also, in certain cases, there may be a term selected arbitrarily by the applicant, in which case the meaning thereof will be described in detail in the description of the corresponding invention. Therefore, the term used in the present invention should be defined based on the meaning of the term, not on the name of a simple term, but on the entire contents of the present invention.

When an element is referred to as " including " an element throughout the specification, it is to be understood that the element may include other elements as well, without departing from the spirit or scope of the present invention. Also, the terms " part, " " module, " and the like described in the specification mean units for processing at least one function or operation, which may be implemented in hardware or software or a combination of hardware and software . In addition, when a part is referred to as being "connected" to another part throughout the specification, it includes not only "directly connected" but also "connected with other part in between".

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing a configuration of an apparatus 1000 for extracting tex1 information in a tidal area using a dron according to an embodiment of the present invention

Referring to FIG. 1, the apparatus 1000 for extracting tidal information in a tidal flat area using a drone includes a ground reference point surveying unit 10, a drone 20, a numerical surface model generating unit 30, an orthoimage image generating unit 40 A boundary image extracting unit 90, and a tidal flat contour extracting unit 100. The gradient image extracting unit 100 extracts the contour of the tidal flat contour, . ≪ / RTI >

The ground reference point surveying unit 10 according to an embodiment of the present invention can perform a ground control point (GCP) survey of a tidal area using a DGPS (Differential Global Positioning System) survey result.

A ground reference point (GCP) is a reference point used to obtain a coordinate conversion formula between an image coordinate system and a map coordinate system in remote sensing. For remote sensing, a ground reference point for coordinate conversion should be selected first. In other words, since the coordinate point which is commonly known in both coordinate systems of the image and the map should be selected as the ground reference point, the position of the pixel capable of grasping the clear point, the point where it can be grasped regardless of the shooting time, Ground reference points should be selected.

Ground reference point surveying refers to the horizontal position of the ground reference point and the survey performed to obtain the elevation. That is, the ground reference point survey is to measure the three-dimensional coordinates of the ground reference point required for aerial triangulation. Based on the national reference point scorecard of the Korea Geographical Information Institute, it is possible to use the DGPS (Differential Global Positioning System) Coordinates can be obtained.

That is, the terrestrial reference point surveying unit 10 can obtain the three-dimensional coordinates of the terrestrial reference point using the measurement results of the counterpart method (DGPS) using two or more GPS receivers with respect to the predetermined terrestrial reference point.

The drone 20 according to an embodiment of the present invention can take aerial images of the tidal flat area at predetermined time intervals. That is, in order to perform aerial triangulation on the tidal flat area, it is possible to take an aerial image of the tidal flat area at a predetermined time interval using the drones 20.

The dron 20 according to an embodiment of the present invention may include a fixed wing dragon used for professional map production or a rotor wing dragon used for general leisure purposes. In the case of fixed-wing drones used for professional map production, it is possible to carry out measurement by using a DSLR camera because of its wide shooting range and high loading capacity. However, it is very expensive and it is necessary to secure sufficient space for take- It has the disadvantage of high risk in coastal marine surveying due to the characteristics of the marine environment that it is difficult to recover if the gas is lost due to dots, gusts or heavy rainfall.

In the case of the rotor-blade drones, which are relatively used for leisure purposes, the vertical take off and landing is possible and the fact that it is easy to adjust is also advantageous compared to the fixed-wing drones in that the sudden weather change can be immediately recovered. In the present invention, the tibia information is extracted using the rotor blade drones in consideration of the advantages of the rotor blade drums, but the scope of the present invention is not limited thereto.

2 is a view showing a rotor blade dron 20 and a camera 22 according to an embodiment of the present invention. As the rotor blade drones (20) for aerial triangulation, the same specification model as the currently available Phantom4 model of DJI company was used and a camera mounted on the aircraft was used. Table 1 below shows specifications of the rotor blade drones 20 and the camera 22 according to an embodiment of the present invention.

FIG. 3 is a view showing a coastal area located in Noan-ri, Chunggye-myeon, Muan-gun, Jeollanam-do, which is an aerial image capturing area of a dron according to an embodiment of the present invention, and FIG. 4 is a view showing a flight path of a dron according to an embodiment of the present invention .

Referring to FIGS. 3 and 4, it is assumed that a single shooting time for the corresponding area using the drone 20 took 20 minutes on average. For shooting blocks of about 0.5 km ² per shooting time, 80% We captured about 40% or more and got about 350 aerial images per shot block.

FIG. 5 is a flowchart illustrating an operation of aerial triangulation with respect to a tidal flat area utilizing the drones 20 according to an embodiment of the present invention.

Referring to FIG. 5, the operation sequence of the aerial triangulation with respect to the tidal flat area using the dron 20 is performed by selecting an area to be photographed to be surveyed prior to photographing the aerial image (S10) A field survey is performed (S20), a flight plan such as a flight path and an altitude is established (S30), and a step S40 of acquiring a photographing permission of an aerial image using a drone is performed.

Next, the ground reference point is measured using the ground reference point surveying unit 10 (S50), the aviation image of the tidal flat area is photographed using the drones 20 (S60), and the image noise removal (S70), a digital surface model (DSM) and an orthoimage image are acquired using the preprocessed aerial image and the measured ground reference point coordinates (S80), and the obtained numerical surface (DSM) and the step of evaluating and verifying the accuracy of the orthoimage (S90), the aerial triangulation can be performed on the tidal flat area.

The numerical surface model generation unit 30 may generate a digital surface model (DSM) using the aerial image photographed by the drones 20. [

A numerical surface model (DSM) is a numerical model representing real world topographic information, and refers to a model expressed in three-dimensional coordinates including the height values of the terrain including actual terrain, trees, buildings, and artificial structures .

That is, the numerical surface model generation unit 30 performs preprocessing such as image noise removal and distortion correction on the aerial image using the aerial image photographed by the drones 20, Extraction and interpolation can be performed to generate a numerical surface model (DSM) expressed in three-dimensional coordinates in an aerial image.

Also, the numerical surface model generation unit 30 automatically determines the conjugate points in the two aerial images using the two matching aerial images through the preprocessing process, automatically matching the two images, We can generate the numerical surface model (DSM) expressed in three-dimensional coordinates in the aerial image by determining the ground coordinates of the conjugate point extracted through matching and determining the space intersection method and performing interpolation or the like.

The orthoimage image generating unit 40 can generate orthoimages using the numerical surface model and the ground reference point. That is, the orthoimage image generation unit 40 uses the ground reference point coordinates acquired through the ground reference point measurement to position the ground reference point in the aerial image, and to correct the undulation displacement, which is the height difference distortion of the object deviated from the lens center, The orthoimage can be generated by combining surface models and performing a rectification process. For example, the orthoimage image generation unit 40 combines the brightness values of the pixels of the aerial image with the numerical surface model to correct the deviation of the image due to the altitude difference of the ground in the aerial image The orthoimage can be generated.

In addition, the orthoimage image generating unit 40 may include a method of interpolating neighboring pixel values to correct a deviation appearing in an image due to an altitude difference of a ground, or a method of interpolating a pixel value corrected through a deviation correction, An orthogonal image can be generated by using an image resampling method or the like.

FIG. 6 is a diagram showing an orthoimage image and a numerical surface model (DSM) generated by photographing times (four times in total) according to an embodiment of the present invention, This is a table showing the accuracy and utility evaluation of the generated orthoimage and numerical surface model (DSM).

6 and Table 2, it can be seen that the spatial resolution of the orthoimage generated in FIG. 6 is about 5 cm on an average, and thus it is possible to generate data having a very high level of spatial resolution compared to the conventional remote sensing technique . In addition, the shooting range for 20 minutes shooting time is 0.5 km ^{2 on} average, which satisfies the general working efficiency for geospatial information production. In case of the position accuracy, it can be confirmed that the data having a level of accuracy enough to construct the geospatial information can be generated because the root mean square error (RMSE) indicating the precision is less than 0.1 m.

The gradient image generator 50 according to an exemplary embodiment of the present invention may generate a gradient image using the difference in brightness value between neighboring pixels in the generated orthoimage.

That is, in order to acquire the boundary lines of different indicator materials, such as the boundary line between the sea and the land, the gradient image generator 50 may calculate brightness values of pixels of a specific portion and surrounding pixels surrounding the specific portion The gradient image can be generated using the difference. A method of generating a gradient image from an orthoimage can be expressed as Equation (1) below.

FIG. 7A is an orthoimage image of the ground surface according to an embodiment of the present invention, and FIG. 7B is a diagram illustrating a gradient image generated by using an orthoimage image of the ground surface.

Referring to FIGS. 7A and 7B, it can be seen that, in the gradient image, the coordinate point where the difference in the brightness value with respect to the neighboring pixel is large appears brighter than the coordinate point with the difference in brightness value between the neighboring pixel and the neighboring pixel. For example, if the brightness value of a pixel is 100 in the orthoimage and the brightness value of the neighboring pixel is 110, the brightness difference (slope) is 10, and if the brightness value of the neighboring pixel is 200, 100, when the brightness difference (slope) is 100 in the gradient image, the brightness difference (slope) can be expressed more brighter than when the difference is 10. That is, since the gradient image is a diagram showing a difference in brightness value (degree of gradient) of a pixel, a coordinate point of a pixel having a large gradient (brightness gradient) is brightly displayed by a difference in brightness value (degree of gradient) ) Can be expressed darkly at the coordinate point of a small pixel.

E.g. In the case of a pixel located at the boundary line between the sea and the land, the coordinate value of the boundary line between the sea and the land can be expressed brightly in the gradient image because the image brightness value with the surrounding pixels is large. Also, in the case of a pixel positioned at the center of a region having the same physical properties as the sea or the center of the land, the difference in brightness value between the surrounding pixels and the neighboring pixels may be small, so that the coordinate point in the center of the sea or land can be darkened in the gradient image. Accordingly, by connecting the pixels of the coordinate points that are brightly expressed in the gradient image, it is possible to obtain a boundary line of a point having different physical properties, for example, a boundary between the sea and the land.

The image segment dividing unit 60 may divide the generated gradient image into a plurality of image segments by applying a watershed algorithm. That is, the image segment dividing unit 60 uses a watershed algorithm to divide an image into pieces having similar characteristics in a gradient image. This algorithm uses a watershed algorithm in which the brightness difference (gradation operation value) Assuming that the local minimum is filled with water, the image is divided according to the shape of the water due to the difference in height.

FIG. 8A is a diagram illustrating a height difference (gradation operation value) of a pixel of a gradient image according to an exemplary embodiment of the present invention, and FIG. 8B is a diagram illustrating a gradient value according to an exemplary embodiment of the present invention (Region 1 and region 2) by applying a watershed algorithm in a gradient image.

8 (b), the watershed algorithm assumes that the brightness difference value (gradient operation value) of each pixel in the gradient image is the contour line of the indicator, and divides the watershed between the ridge line and the ridge line into the respective areas For example, region 1 and region 2), it is possible to extract and divide into image pieces having similar characteristics.

In other words, the watershed algorithm searches the local lowest points from the gradient image, and then merges pixels with high altitudes around each other according to the shape of the water at each local trough, to be. For example, when a search originating at a different lowest point in a pixel meets each other, the pixels may be connected to extract a boundary, and a piece may be formed along the extracted boundary to divide the image.

However, since the watershed algorithm extracts too many regions when the number of local low points is excessively large in the gradient image, in the case of a high resolution image having a spatial resolution of several centimeters (cm) The shadow or the ground object of the image is identified in the image, so that an excessive image fragment is extracted and divided. In order to solve this problem, the image segment dividing unit 60 can extract a proper number of image fragments from the gradient image by determining a threshold value by applying a scale level in the application process of the watershed algorithm.

9A is a view showing a gradient image according to an embodiment of the present invention. FIG. 9B is a diagram illustrating a gradient image according to an exemplary embodiment of the present invention, in which a watershed algorithm is applied by applying a high- 9C is a diagram illustrating a watershed algorithm applying a low scale level to a gradient image according to an exemplary embodiment of the present invention.

9 (a), 9 (b), and 9 (c), application of a scale level to a watershed algorithm determines a specific threshold by applying a scale level, A watershed algorithm is applied to extract only a local minimum point having a gradient value higher than a specific threshold without considering a local minimum point having a gradient value.

That is, the determination of the specific threshold value applying the scale level is a method of determining the local lowest point value of the upper% as a specific threshold value when all the regional lowest point values are accumulated in order of magnitude. For example, if the scale level is set to 10%, the local lowest point value corresponding to the upper 10% of all the regional lowest points is determined as the specific threshold value, and the local lowest point value corresponding to 10% A threshold value of 10%), the image segment can be extracted by applying the watershed algorithm.

10 is a view showing a result of dividing an image fragment into a watershed algorithm by applying a scale level of 50% to a gradient image of a tidal flat area according to an embodiment of the present invention.

Referring to FIG. 10, the image segmentation unit 60 may extract and divide a plurality of image segments by setting a scale level of 50%, which corresponds to a median value, in a gradient image of a tidal flat region, and applying a watershed algorithm. That is, in order to prevent the image from being divided into an excessive number, the image level of the tidal area is divided by applying the scale level of 50% corresponding to the median value. However, the scale level value can be changed as needed, But is not limited thereto.

The image merge unit 70 according to an embodiment of the present invention may merge adjacent image pieces when the spectral characteristics of adjacent image pieces are equal to or greater than a certain degree of similarity for each of the plurality of image pieces.

Referring to FIG. 10, even when the scale level is set to 50% and the watershed algorithm is applied to the gradient image, the ship and artifacts are still excessively divided as compared with the tidal-flat topography to be analyzed. Therefore, when the spectral characteristics of the adjacent image pieces are similar, for example, when the spectral characteristics are equal to or greater than a certain degree of similarity, a process of merging adjacent image pieces is required.

That is, the image merge unit 70 calculates an average spectral vector of a plurality of divided image fragments, measures the similarity, compares the measured similarity with a predetermined threshold value, and outputs the image fragments having a degree of similarity larger than a predetermined threshold value Can be merged.

For example, the image merge unit 70 measures the similarity of a plurality of image fragments using the Full Lambda Schedule technique, accumulates the measured similarity in the order of magnitude, and sets the similarity having an upper specific percentage (% And merges the image fragments having a degree of similarity larger than the determined predetermined threshold value.

A method of measuring the similarity of a plurality of image fragments using the full lambda schedule technique of the image fragment merging unit 70 can be expressed as Equation (2) below.

FIG. 11 is a diagram showing a result of merging image fragments having a similarity level (merge level 70%) corresponding to the upper 70% by measuring similarity after applying the watershed algorithm according to an embodiment of the present invention.

Referring to FIGS. 10 and 11, it can be seen that the image pieces that have been fragmented in the tidal-flat topography, the coastal vessel, and the artificial structure are merged by the image fragment merging unit 70. FIG.

The water-based discrimination reference learning unit 80 extracts and analyzes the characteristics of the pixels constituting each piece according to the image pieces obtained through the image-carving unit 70, can do.

For example, the water-based discrimination reference learning unit 80 may extract at least one of the average brightness value, the area, and the texture characteristic of the pixels constituting each piece, analyze it using the supervisory classification method, have. In other words, the water-based discrimination reference learning unit 80 may use at least one of the average brightness value, the area, and the texture property to learn the water-based discrimination criterion using the spectroscopic characteristics, geometric characteristics, surface entropy characteristics, One can be extracted and analyzed using the supervised classification method, so that the water classification criterion can be learned. For example, in order to extract the characteristics of the water system in the tidal-flat area, the water-based criterion learning unit 80 uses the characteristics of the image pieces superimposed on the water system of the tidal area and the surrounding reservoir area at the lowest water level, Can be used to learn water classification criteria.

The water boundary extracting unit 90 according to an embodiment of the present invention classifies the water area in the acquired image fragments by using the learned water based criterion, the water boundary can be extracted by merging through the adjacency analysis.

FIG. 12 is a diagram showing the boundaries of the water boundaries extracted by the water boundary extracting unit 90 according to an embodiment of the present invention, by time (time interval).

Referring to FIG. 12, the water boundary extracting unit 90 may extract the water boundary by merging extracted and merged image segment clusters using the learned water based criterion. Here, the image segment is a set of polygon objects obtained by image processing, and the water boundary extraction unit 90 performs object adjacency analysis, which is a type of spatial analysis, The boundary of the water system can be extracted.

The tidal flat contour extracting unit 100 according to an embodiment of the present invention can extract the contour lines of the tidal flat area by superimposing the water boundary extracted by the water boundary extracting unit 90 on a time basis.

FIG. 13 is a diagram showing the distribution of tidal area contour lines obtained by superimposing a water boundary at each time point according to an embodiment of the present invention.

Referring to FIG. 13, the tidal flat contour extracting unit 100 can extract the contour lines of the tidal flat area by superimposing the water boundary images extracted by the water boundary extracting unit 90 on a time basis.

14 is a flowchart illustrating a method of extracting tibia information in a tidal flat area using a dron according to an embodiment of the present invention.

Referring to FIG. 14, in step S100, the ground reference point surveying unit 10 can perform ground reference point survey of the tidal flat area through DGPS (Differential Global Positioning System) surveying. For example, the ground reference point surveying unit 10 acquires the three-dimensional coordinates of the ground reference point based on the DGPS (Differential Global Positioning System) survey results based on the national reference point scorecard of the National Geographical Information Institute for the previously selected ground reference points can do.

In step S200, for the aerial triangulation, the drones 20 can be used to take aerial images of the tidal flat area at regular intervals. In other words, in order to acquire the numerical surface model and orthoimage image for the analysis of the tidal flat area, the tidal flat area can be photographed at a predetermined time interval using the drone.

In step S300, a numerical surface model and an orthoimage can be generated using the aerial image photographed. For example, the numerical surface model generating unit 30 performs preprocessing such as image noise removal and distortion correction on the aerial image using the aerial image photographed by the dron 20, (DSM) represented by three-dimensional coordinates in the aerial image, and the ortho-image generation unit 40 generates the numerical surface model (DSM) The orthoimage can be generated by using the numerical surface model generated by the geomagnetic sensor and the measured ground reference point coordinates. For example, an orthoimage can be generated through processes such as rectification, occlusion correction, and color matching.

In step S400, the gradient image generator 50 may generate a gradient image using the difference in brightness value between neighboring pixels in the orthoimage image. For example, in order to acquire the boundary lines of different indicator materials, such as the sea-to-land boundary line, the gradient image generator 50 may generate a gradient image of pixels of a specific part and neighboring pixels surrounding the specific part The gradient image can be generated using the brightness difference.

In step S500, the image segment dividing unit 60 may divide the generated gradient image into a plurality of image segments by applying a watershed algorithm. For example, the image segment dividing unit 60 searches the local lowest points from the gradient image, and sequentially merges pixels having high altitudes in accordance with the shape of the water at each local lowest point to extract boundaries A gradient image can be divided into a plurality of image fragments by applying a watershed algorithm that divides the image into fragments.

The image segment dividing unit 60 determines a threshold by applying a scale level to the generated gradient image, and does not consider a local minimum point having a gradient value lower than the determined threshold value, The watershed algorithm can be applied to use only local low points having a higher gradient value to divide into multiple image segments.

In step S600, for each of the plurality of image fragments, the image fragment merging unit 70 may merge adjacent image fragments when the spectral characteristics of adjacent image fragments are equal to or greater than a certain degree of similarity. For example, the image merge unit 70 may calculate an average spectral vector of a plurality of divided image fragments, measure the similarity, compare the measured similarity with a predetermined threshold value, and calculate a similarity degree larger than the predetermined threshold value You can merge video fragments.

In step S700, the water-based discrimination reference learning unit 80 may learn the water-based discrimination criteria by extracting and analyzing the characteristics of the pixels constituting each piece of the image piece obtained through the image-carving unit 70. [ For example, the water-based discrimination reference learning unit 80 may extract at least one of the average brightness value, the area, and the texture characteristic of the pixels constituting each piece, analyze it using the supervisory classification method, have.

In step S800, the water-system boundary extraction unit 90 classifies the water-based areas in the acquired image pieces using the learned water-based criterion, merges the image pieces of the separated water-based areas through object proximity analysis, Can be extracted.

In step S900, the tidal flat contour extracting unit 100 can extract the contour lines of the tidal flat area by superimposing the water boundary extracted by the water boundary extracting unit 90 on a time slot basis.

As to the method of extracting the tugboat information in the tidal flat area using the dron according to the embodiment of the present invention, the content of the tugbo information extracting apparatus 1000 in the tidal flat area using the dron may be applied. Therefore, with respect to the method of extracting the tugboat information in the tidal flat area using the drones, the description of the same contents as the contents of the tugbo information extraction device 1000 in the tidal flat area using the drones is omitted.

One embodiment of the present invention may also be embodied in the form of a recording medium including instructions executable by a computer, such as program modules, being executed by a computer. Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. The computer-readable medium may also be embodied in a computer-readable medium, which may be volatile and / or non-volatile, implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules, Nonvolatile, removable and non-removable media.

It will be understood by those skilled in the art that the foregoing description of the present invention is for illustrative purposes only and that those of ordinary skill in the art can readily understand that various changes and modifications may be made without departing from the spirit or essential characteristics of the present invention. will be. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. For example, each component described as a single entity may be distributed and implemented, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is defined by the appended claims rather than the detailed description and all changes or modifications derived from the meaning and scope of the claims and their equivalents are to be construed as being included within the scope of the present invention do.

10: ground reference point surveying unit 20: drones
30: Numerical surface model 40: Orthographic image generating section
50: gradient image generation unit 60: image segmentation unit
70: video image merge unit 80: water system discrimination reference learning unit
90: Water boundary extraction unit 100: Tidal flat contour extraction unit
1000: Tidal information extraction device in a tidal flat area using drone

Claims (9)

In a tide information extracting apparatus in a tidal flat area utilizing a drone,
A ground reference point surveying unit for performing land reference point survey of tidal flat area using DGPS (Differential Global Positioning System) survey results;
A drones for capturing aerial images of the tidal flat area at predetermined time intervals for aerial triangulation;
A numerical surface model generation unit for generating a numerical surface model using the photographed aerial image;
An orthoimage image generation unit for generating an orthoimage image using the generated numerical surface model and the ground reference point;
A gradient image generation unit for generating a gradient image using a brightness value difference between a pixel in a predetermined region and a surrounding pixel surrounding the predetermined region in the generated orthoimage;
An image segmentation unit for dividing the generated gradient image into a plurality of image segments by applying a watershed algorithm;
A merging unit for merging the adjacent image fragments when the spectral characteristics of adjacent image fragments are equal to or greater than a certain degree of similarity for each of the plurality of image fragments;
A water level discrimination reference learning unit for extracting and analyzing characteristics of pixels constituting each piece of the image piece obtained through the image carving unit and learning the water classification criterion;
Using the learned water-based criterion, water areas are classified in the obtained image fragments, image fragments of the separated water areas are merged through object adjacency analysis, and water boundary is extracted A water boundary extracting unit; And
And a tidal flat contour extracting unit for extracting contour lines of the tidal flat area by superimposing the extracted water boundary on a time basis,
The image merge unit calculates an average spectral vector of the divided image segments using the Full Lambda Schedule technique to measure the similarity, accumulates the measured similarity in order of magnitude, and calculates an upper predetermined percentage (%) value Determining a degree of similarity having the similarity degree as the threshold value, merging image pieces having a degree of similarity larger than the determined threshold value,
The Full Lambda Schedule technique is expressed by the following equation,

remind

I < / RTI >

Quot;

Area,

Quot;

Average spectral characteristic vector < RTI ID = 0.0 &

Quot;

Wow

And the length of the boundary of the tidal flats in the tidal flat area.
The method according to claim 1,
Wherein the image fragment dividing unit comprises:
A threshold value is determined by applying a scale level to the generated gradient image and a local minimum point having a gradient value higher than the determined threshold value is considered without considering a local minimum point having a gradient value lower than the determined threshold value, A device for extracting tibia information in a tidal flat area using a dron which divides the image into a plurality of image fragments by using the drones.
delete The method according to claim 1,
Wherein the water-
Extracting at least one of an average brightness value, an area, and a texture characteristic of the pixels constituting each piece, analyzing it using a supervisory classification method,
In a method for extracting tibia information in a tidal flat area using a drone,
Performing a ground reference point survey of a tidal flat area through a DGPS (Differential Global Positioning System) survey;
Capturing aerial images of the tidal flat area at predetermined time intervals using a drone for aerial triangulation;
Generating a numerical surface model and an orthoimage using the photographed aerial image;
Generating a gradient image using a brightness difference between a pixel in a predetermined region and a surrounding pixel surrounding the predetermined region in the orthoimage;
Dividing the generated gradient image into a plurality of image segments by applying a watershed algorithm;
Merging the adjacent image fragments when the spectral characteristics of adjacent image fragments are greater than or equal to a certain degree of similarity, for each of the plurality of image fragments;
Extracting and analyzing characteristics of pixels constituting each piece of the image piece obtained through the merging step, and learning a water classification criterion;
Extracting a water body region from the acquired image pieces using the learned water based criterion and extracting a water body boundary by merging the image pieces of the separated water body region through object adjacency analysis ; And
And extracting contour lines of the tidal flat area by superimposing the extracted water boundary on a time zone basis,
The merging of the adjacent image fragments may include calculating a mean spectral vector of the plurality of divided image fragments using the Full Lambda Schedule technique to measure the similarity, accumulating the measured similarity in order of magnitude, Determining a degree of similarity having a value (%) as a threshold value, and merging image pieces having a degree of similarity larger than the determined threshold value,
The Full Lambda Schedule technique is expressed by the following equation,

remind

I < / RTI >

Quot;

Area,

Quot;

Average spectral characteristic vector < RTI ID = 0.0 &

Quot;

Wow

And the length of the boundary of the tidal flat using the dron.
6. The method of claim 5,
Wherein dividing the generated gradient image into a plurality of image segments by applying a watershed algorithm comprises:
A threshold value is determined by applying a scale level to the generated gradient image and a local minimum point having a gradient value higher than the determined threshold value is considered without considering a local minimum point having a gradient value lower than the determined threshold value, And dividing the image into a plurality of image segments by using the drones.
delete 6. The method of claim 5,
The step of extracting and analyzing the characteristics of the pixels constituting each of the fragments extracts at least one of the average brightness value, the area and the texture characteristic of the pixels constituting each piece, And extracting the tide information in the tidal flat area using the drone.
A computer-readable recording medium having recorded thereon a program for implementing the method of any one of claims 5, 6, and 8.

Images