JP4887130B2 - Farm district drawing data creation system - Google Patents

Description

  The present invention relates to a geographic image processing system that analyzes observation images obtained by photographing a ground surface from an altitude using a satellite, an aircraft, and the like, and creates data relating to land sections on the ground surface, and in particular, agriculture included in geographic images. The present invention relates to a system for performing image processing for facilitating site extraction.

  In recent years, the use of satellite images and aerial photograph images has become widespread, and techniques relating to the creation of maps using these images and the analysis of the ground surface have been developed. Among these, in order to analyze a photograph of a farmland part, it is essential to extract a land section in the image and further extract an area corresponding to the farmland from the extracted land sections.

  Conventionally, the extraction of land parcels and the extraction of farmland from the extracted land parcels have been performed manually, so that it has been a problem that a great human cost is required. Therefore, a method for automating these processes by a computer has been proposed. As an example, in Patent Document 1, an edge is first extracted from an image obtained by photographing the ground surface, an intersection of a straight line and an edge extending radially from an arbitrary point on the edge image is obtained, and a plurality of intersections are connected. A process has been proposed for extracting a polygonal segmented area. By this process, the land parcel can be automatically extracted.

JP 2003-256807 A J. Canny, `` A Computational Approach to Edge Detection '' IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.PAMI-8, No6, November 1986 Mamoru Onoe, “Image Processing Handbook”, Shoshoku Shuppan, 1987 Masami Kaneko, 5 others "Study on remote sensing method for marsh vegetation classification", Hokkaido Environmental Science Research Center, No. 29, 2002, pp.53-58

  However, although it is possible to extract a partitioned area that is a candidate for farmland from a geographic image by the method of Patent Document 1, it is determined by the operator that the extracted partitioned area is a farmland or a non-farmland and is classified by the operator. Actually, it is necessary to look at the image of each divided area. This operation is not automated and requires a great amount of human cost, and the operator requires a great deal of concentration and time. In addition, in order to work on a wide range of images in a limited time, a plurality of operators are required. In this case, the classification results vary depending on the subjective judgment criteria of each operator. There is also a fear.

  The present invention has been made in view of such circumstances, and intends to provide an agricultural district image data creation system capable of easily creating land parcel data such as farmland from geographic images such as satellite images and aerial photograph images. To do.

  As a result of diligent research in view of the above problems, the inventor automatically calculated parameters such as the amount of noise, vegetation index, and section shape for the extracted section area, and automatically determined the farmland-likeness. And non-agricultural land, and the above problem was solved.

  That is, the present invention is a system for extracting and classifying a partition area included in a geographic image, extracting a contour of an agricultural district image such as a field in the geographic image, that is, an edge, and dividing the partition area based on the extracted edge. And a zone classification unit that analyzes the characteristics of each zone to represent the characteristics of farmland and determines land use applications such as “agricultural land” or “non-agricultural land”.

  More specifically, the present invention relates to an agricultural district image data creation system that analyzes geographic image data obtained by photographing the ground surface and creates agricultural district image data, and relates to an area extraction unit that extracts an area from geographic image data. Noise determining means for calculating the amount of noise in the partitioned area, vegetation determining means for calculating the vegetation index of the partitioned area, shape determining means for determining the shape feature of the partitioned area, noise amount, vegetation index, and shape feature And a farmland judging means for judging whether or not the partition area is a farmland. The noise determination unit calculates a noise evaluation value that is a noise amount with respect to the area of the extracted partition area. Here, the noise evaluation value indicates that the smaller the value, the higher the agricultural land quality. In addition, the shape determining means checks the angle formed by each vector constituting the extracted partitioned area. The angle made here indicates that the closer to 0 degrees or 90 degrees, the higher the farmland-likeness.

  The farmland judging means multiplies each of the noise amount, the vegetation index, and the shape feature by different weighting coefficients, calculates the sum of them, and when the sum is larger than a predetermined threshold, the partition area is farmland Is determined. The farmland determination means may make the weighting coefficient variable according to the photographing time of the geographic image data.

  Further features of the present invention will become apparent from the best mode for carrying out the present invention and the accompanying drawings.

  As described above, according to the geographic image processing system of the present invention, the work of extracting and classifying land sections from geographic images such as satellite images and aerial photograph images can be made much more efficient. In addition, since the work process that previously relied on the skill of the operator is automated, even if the work is performed by a large number of operators, a uniform land parcel is not affected by differences in the skill level of the operators. Data can be obtained.

  The best mode for carrying out the geographic image processing system of the present invention will be described below in detail with reference to the accompanying drawings. 1 to 10 are diagrams illustrating an embodiment of the present invention. In these drawings, parts denoted by the same reference numerals represent the same items, and the basic configuration and operation are the same. Note that the present invention is not limited to this embodiment. A person skilled in the art can make various changes and improvements to the configuration and function of the invention according to the embodiments without departing from the gist of the present invention.

<System configuration>
FIG. 1 is a block diagram schematically showing the configuration of a geographic image processing system of the present invention. In FIG. 1, this system includes a storage device 20 including a processing device 10 including a personal computer, a workstation and the like, and an auxiliary storage device such as a RAM (Random Access Memory) and a magnetic disk storage device used as a main memory. And an input / output device 30.

  The input / output device 30 includes an input device 31 including a pointing device such as a keyboard and a mouse, a display device 32 such as a CRT display device, and a printer 33. The input device 31 is used for parameter input by a user, activation of a command, and the like, and land parcel data generation using a geographic image subjected to image processing by the present system. The display device 32 and the printer 33 are used to present a geographic image and land parcel data handled by this system to the user. Note that the display device may include only one of the display device 32 and the printer 33.

  The processing device 10 includes a geographic image processing program 40. The geographic image processing program 40 includes, as program modules, a section extraction unit 100 that extracts sections of farmland and the like included in the geographic image, and a section classification unit 200 that classifies each extracted section area for each land use purpose. Including.

  In addition, the subdivision classification unit 200 uses, as a submodule, a noise determination unit 201 that extracts noise in each subdivision region using edges extracted from a geographical image to determine the likelihood of farmland, and a vegetation index (described later) using the geographical image. Each of the vegetation determination unit 202 that calculates the agricultural land-likeness of each divided area, the shape determination unit 203 that evaluates the shape of each divided area and determines the agricultural-landiness, the noise determination unit, the vegetation determination unit, and the shape determination unit A farmland determination unit 204 that performs classification (classification / determination of farmland or non-farmland) of each divided area based on the evaluation value is included.

  The storage device 20 stores geographic image data 21, partition area data 22, noise data 23, vegetation data 24, section shape data 25, and section classification data 26. Among these, the geographic image data 21 is obtained from an artificial satellite or an aerial photograph image, and is data stored in advance before executing the processing by the present system. On the other hand, the section area data 22, noise data 23, vegetation data 24, section shape data 25, and section classification data 26 are data generated based on the geographic image data 21 in this system.

<Outline of geographic image processing>
FIG. 2 is a diagram illustrating an overview of image processing performed by the system. FIG. 2A shows geographic image data 21 stored in the storage device 20 in advance. By performing section extraction on this geographic image data, a section area as shown in FIG. 2D is obtained. In FIG. 2 (d), one polygon represents one section area. One partition area is data in which a plurality of vectors are connected. This data is stored in the storage device 20 as partitioned area data 22. Further, the edge image shown in FIG. 2C is obtained by performing edge detection processing on the geographic image data. A noise determination process is performed on the edge image to determine the farmland-likeness. The edge detection process can be performed using various edge filters such as the Canny method (see Non-Patent Document 1) and a Sobel filter (see Non-Patent Document 2). Data of the evaluation value of the noise amount of each partition area obtained by the noise determination process is stored in the storage device 20 as noise data 23. Moreover, the vegetation image shown in FIG.2 (b) is obtained by calculating a vegetation index with respect to geographic image data. The vegetation image is an image that displays the value of the vegetation index of each pixel of the geographic image. The vegetation evaluation value data of each section area obtained by the vegetation determination process is stored in the storage device 20 as vegetation data 24. In addition, the angle between adjacent vectors in each partitioned area of the partitioned area data is calculated. FIG. 10A is a schematic view at that time. An angle formed by adjacent vectors in each divided area is calculated, and an evaluation value representing the agricultural land quality is calculated. Evaluation value data based on the shape of each partition area at this time is stored in the storage device 20 as partition shape data 25. Using the noise data 23, the vegetation data 24, and the section shape data 25, the farmland-likeness of each section area is determined, and the classification result of each section area is output to the storage device 20 as the section classification data 26. FIG. 2E is an example of the section classification data 26. In the prior art, the process up to the extraction of the partition area as shown in FIG. 2D is automated and the classification of the partition area is performed manually. However, in the present invention, the classification of the partition area can also be performed automatically.

<Details of geographic image processing>
The details of the section extraction process, noise determination process, vegetation determination process, shape determination process, and section determination process in the geographic image processing system of the present invention will be described below. In the present invention, when the geographic image processing program 40 is activated, the segment extraction unit 100 and the segment classification unit 200 are activated in order. In the section classification unit, a noise determination unit 201, a vegetation determination unit 202, a shape determination unit 203, and an agricultural land determination unit 204 are activated. However, the processes of the noise determination unit 201, the vegetation determination unit 202, and the shape determination unit 203 are in no particular order, and the farmland determination unit 204 is activated after these. The details of the processing by each will be described below.

(1) Section Extraction Processing FIG. 3 is a flowchart showing the flow of processing by the section extraction unit 100 of the geographic image processing program 40. In FIG. 3, the section extraction unit 100 reads geographic image data 21 from the storage device 20 (S301), extracts edges such as farmland existing in the image (S302), and various well-known processes for extracting edges. Technology is applicable. Any method may be used as long as the contour of the land parcel can be extracted. In this system, the Canny method is applied. By this process, an edge image as shown in FIG. 2C is generated.

  Next, the section extraction unit 100 performs region extraction processing on the edge image in order to extract regions that are farmland candidates (S303). A linear intersection method (Patent Document 1) is used for region extraction at this time. This process will be described with reference to FIG. This process is a process of extracting a partitioned area including a seed point from a given edge image and a coordinate of one point called a seed point. The seed point refers to a coordinate serving as a starting point when extracting a partitioned area. The seed points are determined by a method such as interspersed in a lattice pattern or randomly arranged. Assume that an edge image and a seed point are given in advance as shown in FIG. First, a straight line is extended in all directions at equal angular intervals from the seed point (FIG. 8B). And the intersection of each straight line and the edge pixel in an edge image is calculated | required, and the obtained intersection is combined in order (FIG.8 (c)). A polygonal region extracted by this processing is defined as a partitioned region (FIG. 8D). The number of vertices of the partition area depends on the number of straight lines extending from the seed point. For example, in FIG. 8, the number of straight lines is eight. By increasing the number of straight lines, it is possible to extract a more precise partition region. However, when a lot of noise is included in the edge image, noise may be extracted as an intersection, so an appropriate number is set. The above processing is performed for all seed points.

Next, the section extraction unit 100 performs adjustment processing on the extracted section area as necessary (S304, S305). The adjustment process is performed in order to check the partition area obtained in the process so far and to correct unnatural partition area data (for example, an unnaturally sharp point). The partition area data is checked mainly by checking the size and checking the shape and the angle formed by the shape (refer to Patent Document 1 for details). Accordingly, for example, in the case of land data such as bag land, there is a possibility that the passage leading to the road may be extracted as an acute portion, but only the shape portion excluding the passage is extracted at the stage of the section extraction processing.
The partition extraction unit 100 stores the vector data of each partition region thus obtained in the storage device 20 as the partition region data 22 (S306).

(2) Noise Determination Processing FIG. 4 is a flowchart showing the flow of processing by the noise determination unit 201 in the section classification unit 200 of the geographic image processing program 40. In FIG. 4, the noise determination unit 201 reads the geographic image data 21 and the partitioned area data 22 from the storage device 20 (S401), and extracts an edge in the geographic image (S402). Then, based on the edge amount in the area corresponding to each divided area, the farmland-likeness (whether it is farmland) of the divided area is determined. Similar to the partition extraction process, various well-known techniques can be applied as the process of extracting an edge. In this system, the Canny method is applied. The edge image generated by this processing is called a noise determination edge image. Then, a noise amount calculation process for each partition area is performed (S403). The amount of noise is calculated using Equation 1.

  Where N is the noise evaluation value in the target partition area, g (x, y) is the luminance value at the coordinates (x, y) of the edge image (that is, the luminance of the location detected as an edge), and S is the target Is the area of the partition area. This calculation is performed for each partitioned area in the noise determination edge image, and the noise evaluation value of each partitioned area is calculated.

  Next, noise determination processing is performed to determine the farmland-likeness of each divided area. In general, an area of farmland has a small amount of edge because there is little change in luminance within the area. On the other hand, in buildings and parking lots, the outlines of buildings and outlines such as white lines tend to be strong as edges, and the amount of edges tends to increase. Therefore, it is possible to distinguish between farmland and buildings to some extent by evaluating the edge amount. FIG. 9 is a diagram illustrating examples of edge data of farmland and non-farmland in the noise determination process. FIG. 9A shows a paddy field with a small amount of edges. On the other hand, FIG. 9B is a section where a building stands, and there are many edges of the building. In such a case, when determination is made according to the present invention, FIG. 9A can be distinguished as “agricultural land”, and FIG. 9B can be distinguished as “non-agricultural land”.

  The noise determination unit 201 stores the noise evaluation value of each partition area thus generated in the storage device 20 as the noise data 23 (S404).

(3) Vegetation Determination Processing FIG. 5 is a flowchart showing a flow of processing by the vegetation determination unit 202 in the section classification unit 200 of the geographic image processing program 40. In FIG. 5, the vegetation determination unit 202 reads the geographic image data 21 and the section area data 22 from the storage device 20 (S501), and calculates a vegetation index in the geographic image (S502). The vegetation index is an index indicating the presence / absence / something / activity of vegetation obtained from multiple bands of remote sensing (see, for example, Non-Patent Document 3). A typical vegetation index is NDVI (Normalized Difference Vegetation Index), which is obtained by Equation 2.

  Here, IR is the reflectance of the near-infrared band, and R is the reflectance of the visible red band. IR is about 0.7 to 1.1 μm, and R is about 0.5 to 0.7 μm. Although the wavelength band varies depending on the surface image used, IR corresponds to the near infrared region where the reflectance is high due to the cell structure of the leaf of the plant, and R is strong around 0.64 to 0.67 μm due to chlorophyll contained in the leaf It corresponds to the absorption band (red of visible light). In other words, IR and R are two bands sandwiching the red edge. For this reason, the value of NDVI becomes high, so that there are many leaves of a plant on the ground surface corresponding to a pixel. Even in the same farmland, there are periods when the NDVI value is high and low depending on the season. For example, in the case of paddy fields, one side is green and the NDVI value is high from the rice planting to the summer, but the green is low and the NDVI value is low at the harvest time in the autumn and after the harvest. Therefore, an optimum determination can be made by determining whether to consider the NDVI value as an important factor for farmland determination according to the season (for example, changing a weighting factor described later according to the season).

  In general, the vegetation index is high in farmland, and the vegetation index is low in water areas such as ponds, buildings, and parking lots. Utilizing this fact, the present invention determines the farmland-likeness of each section area. The calculation of the vegetation evaluation value representing the farmland-likeness of each divided area is performed using Equation 3.

  However, V is a vegetation evaluation value in the target partition area, and NDVI (x, y) is a vegetation index at the coordinates (x, y) of the geographic image. S is the area of the target partition area. This calculation is performed for each partitioned area in the geographic image, and the vegetation value of each partitioned area is calculated.

  The vegetation determination unit 202 stores the vegetation evaluation value of each partition area generated in this way as the vegetation data 24 in the storage device 20 (S503).

(4) Shape Determination Processing FIG. 6 is a flowchart showing the flow of processing by the shape determination unit 203 in the section classification unit 200 of the geographic image processing program 40. In FIG. 6, the shape determination unit 203 reads the partitioned area data 22 from the storage device (S601), and quantifies the shape feature by calculating the angle formed by the adjacent vectors of each partitioned area (S602). FIG. 10 is a diagram showing an outline of this processing. FIG. 10A is a diagram when the partitioned area is represented by an octagon, and angles formed by adjacent vectors are represented by θ ₁ to θ ₈ . In general, farmland often has a rectangular shape or a parallelogram shape close to a rectangle. Therefore, it can be said that the angle formed by the adjacent vectors in the partitioned area is close to either 0 degrees or 90 degrees. Therefore, as shown in FIGS. 10B and 10C, many evaluation values are given when the angle is close to either 0 degrees or 90 degrees, and evaluation is performed when the angle is far from either 0 degrees or 90 degrees. Give a small value. The evaluation value of the angle of the adjacent vector in each division area is calculated | required, and the average is used as the evaluation value showing the agricultural land likeness of the division area. Specifically, it is calculated by Equation 4.

Here, A is the shape evaluation value in the target partition area, E (θ _j ) is the evaluation value of each angle, and n is the number of vertices.
The shape determination unit 203 stores the shape evaluation value of each partition area generated in this way in the storage device 20 as the partition shape data 25 (S603).

(5) Agricultural Land Determination Unit FIG. 7 is a flowchart showing a flow of processing by the agricultural land determination unit 204 in the section classification unit 200 of the geographic image processing program 40. In FIG. 7, the farmland determination unit 204 reads the noise data 23, the vegetation data 24, and the section shape data 25 from the storage device 20 (S701). Next, the farmland-likeness of each section area is evaluated by Expression 5 (S702).

  However, E is a farmland evaluation value representing the farmland-likeness of each division area, and the larger the farmland, the more likely it is farmland. N, V, and A represent a noise evaluation value, a vegetation evaluation value, and a shape evaluation value, respectively, in each partition area. These are obtained from the noise data 23, the vegetation data 24, and the section shape data 25, respectively. α, β, and γ are constants representing the weights of the noise evaluation value, the vegetation evaluation value, and the shape evaluation value, respectively. With this E, it is possible to obtain the farmland-likeness and classify the partitioned areas such as farmland, vacant land, and pond included in the geographic image. The threshold value of E for distinguishing between farmland and non-farmland may be determined appropriately according to the user's request. In addition, even if it says farmland, there are orchards from paddy fields and fields. Therefore, the tendency of the values of N, V, and A varies depending on the properties of each farmland. For example, in the case of an orchard, there is a high possibility that there will be more edges and more noise than paddy fields. Therefore, the N-value weight coefficient may be set lower than that of paddy fields, or individual indicators for orchards (provided separately). The final determination may be made in comparison with the index).

  FIG. 2E is an example of the partition area and classification result for the geographic image of FIG. 2A and the partition area image of FIG. In FIG. 2D, the divided areas of (1), (2), and (4) have less noise as can be confirmed from FIG. 2 (c). Moreover, the vegetation index is high as can be confirmed from FIG. Further, as can be confirmed from FIG. 2D, the shape is close to a rectangle. Therefore, E in Equation 5 is a large value. On the other hand, in FIG. 2D, the partitioned area of (3) is noisy as can be confirmed from FIG. Moreover, the vegetation index is low as can be confirmed from FIG. Further, the shape is slightly distorted as can be confirmed from FIG. Therefore, E in Equation 5 is a small value. As described above, in FIG. 2D, the areas (1), (2), and (4) are determined as “farmland”, and the area (3) is determined as “non-farmland”.

  The farmland determination unit 204 stores the classification result of each division area generated in this way in the storage device 20 as the division classification data 26 (S703).

<Summary>
As described above, the embodiment of the present invention relates to an agricultural district image data creation system that analyzes geographic image data obtained by photographing a ground surface and creates agricultural district image data. The system includes a partition extraction unit (module) that extracts a partition region from geographic image data, a noise determination unit (module) that calculates the amount of noise in the partition region, and a vegetation determination that calculates a vegetation index of the partition region. Using a part (module), a shape determination unit (module) for determining a shape characteristic of a partition area, and a farmland determination unit (module) for determining whether the partition area is farmland using a noise amount, a vegetation index, and a shape feature And). As a result, it is possible to automate the farmland determination process, prevent subjective determination and variations in the determination, and implement objective farmland determination.

  The noise determination unit calculates a noise evaluation value that is a noise amount with respect to the area of the extracted partition region. Here, the noise evaluation value indicates that the smaller the value, the higher the agricultural land quality. In addition, the shape determining means checks the angle formed by each vector constituting the extracted partitioned area. The angle made here indicates that the closer to 0 degrees or 90 degrees, the higher the farmland-likeness. By doing in this way, the index of farmland determination becomes clear and it can contribute to more objective determination.

  In addition, the farmland determination unit multiplies each of the noise amount, the vegetation index, and the shape feature by different weighting factors, calculates a sum of them, and if the sum is greater than a predetermined threshold value, It is determined that If the weighting factor is changed, adaptive farmland determination can be performed. For example, if the weighting coefficient for V is variable according to the photographing time of geographic image data, more appropriate farmland determination can be executed in accordance with the season.

  The present invention can also be realized by a program code of software that realizes the functions of the present embodiment. In this case, a storage medium in which the program code is recorded is provided to the system or apparatus, and the computer (or CPU or MPU) of the system or apparatus reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the program code itself and the storage medium storing the program code constitute the present invention. As a storage medium for supplying such program code, for example, floppy (registered trademark) disk, CD-ROM, DVD-ROM, hard disk, optical disk, magneto-optical disk, CD-R, magnetic tape, non-volatile A memory card, ROM, or the like is used.

  Also, based on the instruction of the program code, an OS (operating system) running on the computer performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing. May be. Further, after the program code read from the storage medium is written in the memory on the computer, the computer CPU or the like performs part or all of the actual processing based on the instruction of the program code. Thus, the functions of the above-described embodiments may be realized.

  In addition, the program code of the software that realizes the functions of the embodiment is distributed via a network, so that it is stored in a storage means such as a hard disk or memory of a system or apparatus or a storage medium such as a CD-RW or CD-R. It may also be achieved by the computer (or CPU or MPU) of the system or apparatus reading and executing the program code stored in the storage means or the storage medium.

It is a block diagram which shows roughly the structure of the geographic image processing system of this invention. It is a figure which illustrates the outline | summary of the image processing performed by the geographic image processing system of this invention. It is a flowchart which shows the flow of a process by the division extraction part of the geographic image processing program shown in FIG. It is a flowchart which shows the flow of a process by the noise determination part of the geographic image processing program shown in FIG. It is a flowchart which shows the flow of a process by the vegetation determination part of the geographic image processing program shown in FIG. It is a flowchart which shows the flow of a process by the shape determination part of the geographic image processing program shown in FIG. It is a flowchart which shows the flow of a process by the farmland determination part of the geographic image processing program shown in FIG. It is a figure which illustrates the outline | summary of a division extraction process. It is a figure which shows the example of the edge data of each farmland and non-farmland in a noise determination process. It is explanatory drawing of a shape determination process. (A) is a figure which illustrates the angle | corner which the adjacent vector makes. (B) is a figure which illustrates the evaluation value of each angle with respect to the angle which a vector makes. (C) is a table illustrating the evaluation value E (θ) of each angle with respect to the angle formed by the vector.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Processing apparatus 20 Storage apparatus 21 Geographic image data 22 Section area data 23 Noise data 24 Vegetation data 25 Section shape data 26 Section classification data 30 Input / output apparatus 31 Input apparatus 32 Display apparatus 33 Printer 40 Geographic image processing program 100 Section extraction part 200 Section classification unit 201 Noise determination unit 202 Vegetation determination unit 203 Shape determination unit 204 Farmland determination unit

Claims (3)

An agricultural district image data creation system that analyzes geographical image data obtained by photographing the ground surface and creates agricultural district image data,
Section extracting means for extracting a section area from the geographic image data;
Noise determination means for calculating an amount of noise in the partition area by extracting an edge image of the partition area and dividing a luminance value of the edge image by an area of the partition area;
The value obtained by subtracting the reflectance of the visible red band from the reflectance of the near-infrared band in the partitioned area, and the value obtained by adding the reflectance of the near-infrared band and the reflectance of the visible red band in the partitioned area. A vegetation judging means for calculating a vegetation index of the partition area by dividing by
Shape determining means for determining a shape feature of the partition area;
Using the amount of noise, the vegetation index, and the shape feature, and including farmland determination means for determining whether the section area is farmland ,
The shape determining means checks an angle formed by each vector constituting the extracted divided area, and the formed angle indicates that the farmland-likeness is higher as the angle is closer to 0 degrees or 90 degrees,
The farmland determination means multiplies the noise amount, the vegetation index, and the shape feature by different weighting factors, calculates a sum of them, and when the value of the sum is larger than a predetermined threshold, farmland partition data creation system area, characterized that you determined to be farmland. The noise determination means calculates a noise evaluation value that is a noise amount with respect to the area of the extracted partition region,
The farm area drawing data creation system according to claim 1, wherein the noise evaluation value indicates that the smaller the value, the higher the farmland quality. The farm district image data creation system according to claim 1 or 2 , wherein the farmland determination means makes the weighting coefficient variable according to the photographing time of the geographic image data.

Images