JP2013255473A - Vegetation mapping program, vegetation mapping apparatus, and vegetation mapping method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To create a vegetation map without additional site investigation.SOLUTION: A disclosed program, in one embodiment, causes a computer to execute processing to classify spectrum data of each tree type into patterns based on spectrum data whose tree type is determined to be the same type as already-known spectrum data and the tree type. Also, the disclosed program causes a computer to execute processing to specify the tree type in an area by comparing the spectrum data of the classified pattern with spectrum data of an area whose tree type is unknown.

Description

  The present invention relates to a vegetation drawing creation program, a vegetation drawing creation device, and a vegetation drawing creation method.

  Conventionally, a vegetation map has been created in a vegetation survey. A vegetation map is a map showing the distribution of tree species covering each area. As a method for creating a vegetation map, for example, there are a field survey in which an investigator who conducts a vegetation survey directly visits the site and visually discriminates tree species in each region, and a method of analyzing an image acquired from an aircraft or satellite. In recent years, hyperspectral sensors have been used in image analysis methods. A hyperspectral sensor is a sensor that acquires hyperspectral data having spectral data in each pixel of one image, for example.

  Here, an example of a technique for creating a vegetation map using a hyperspectral sensor will be described. In this technique, first, a region to be investigated is photographed with a hyperspectral sensor mounted on an aircraft or a satellite to obtain hyperspectral data. Then, the spectrum data of the tree species at the point visually determined by the investigator is extracted from the hyperspectral data as a reference spectrum. Then, by comparing the reference spectrum with the hyperspectral data, the tree species of each pixel in the captured image is specified, and a vegetation map is created.

  Other technologies related to vegetation surveys are as follows. For example, a technique for evaluating vitality, which is an index of the soundness and withering of trees using spectral data, and a technique for identifying tree species in forests using color images taken from aircraft and satellites are known.

JP 2002-360070 A JP-A-8-130984 JP 2006-85517 A JP 2001-357380 A JP 2000-207564 A

  However, the above prior art has a problem that it requires an additional field survey. For example, in the above-described conventional technology, there is a case where the tree type of each pixel in the captured image cannot be specified even if the reference spectrum and the hyperspectral data are collated. In this case, in order to visually identify unspecified tree species, an additional field survey was conducted in which the investigator went directly to an unspecified area.

  The disclosed technology has been made in view of the above, and provides a vegetation drawing creation program, a vegetation drawing creation device, and a vegetation drawing creation method capable of creating a vegetation map without performing an additional field survey. With the goal.

  In one aspect, the disclosed program causes a computer to execute processing for classifying spectral data of each tree species into patterns based on spectral data of known tree species and spectrum data determined to be the same species as the tree species. Further, the disclosed program causes the computer to execute processing for comparing the spectrum data of the classified pattern with the spectrum data of the area where the tree species is unknown, and specifying the tree species of the area.

  According to one aspect of the technique disclosed in the present application, there is an effect that a vegetation map can be created without performing an additional field survey.

FIG. 1 is a block diagram illustrating a functional configuration of the vegetation map creation apparatus according to the first embodiment. FIG. 2 is a diagram illustrating an example of the HSD. FIG. 3 is a diagram illustrating an example of spectrum data included in the HSD. FIG. 4A is a diagram for explaining processing by the collation unit. FIG. 4B is a diagram for explaining processing by the collation unit. FIG. 4C is a diagram for explaining processing by the collation unit. FIG. 4D is a diagram for explaining processing by the collation unit. FIG. 5A is a diagram for explaining processing by the first classification unit. FIG. 5B is a diagram for explaining processing by the first classification unit. FIG. 5C is a diagram for explaining processing by the first classification unit. FIG. 5D is a diagram for explaining processing by the first classification unit. FIG. 6A is a diagram for explaining processing by the first classification unit. FIG. 6B is a diagram for explaining processing by the first classification unit. FIG. 6C is a diagram for explaining processing by the first classification unit. FIG. 6D is a diagram for explaining processing by the first classification unit. FIG. 7 is a diagram for explaining processing by the first classification unit. FIG. 8A is a diagram for explaining processing by the first classification unit. FIG. 8B is a diagram for explaining processing by the first classification unit. FIG. 8C is a diagram for explaining processing by the first classification unit. FIG. 9A is a diagram for explaining processing by the first classification unit. FIG. 9B is a diagram for explaining processing by the first classification unit. FIG. 9C is a diagram for explaining processing by the first classification unit. FIG. 9D is a diagram for explaining processing by the first classification unit. FIG. 10 is a diagram for explaining processing by the second classification unit. FIG. 11A is a diagram for explaining processing by the second classification unit. FIG. 11B is a diagram for explaining processing by the second classification unit. FIG. 12 is a diagram for explaining processing by the specifying unit. FIG. 13 is a diagram for explaining processing by the specifying unit. FIG. 14 is a flowchart illustrating a processing procedure of the vegetation map creation apparatus. FIG. 15 is a flowchart illustrating a processing procedure of the vegetation map creation apparatus. FIG. 16 is a flowchart illustrating a processing procedure of the vegetation map creation apparatus. FIG. 17 is a diagram illustrating an example of a computer that executes a vegetation map creation program.

  Hereinafter, embodiments of a vegetation drawing creation program, a vegetation drawing creation device, and a vegetation drawing creation method disclosed in the present application will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiments. Each embodiment can be appropriately combined within a range in which processing contents do not contradict each other.

  An example of a functional configuration of the vegetation map creation apparatus according to the first embodiment will be described. FIG. 1 is a block diagram illustrating a functional configuration of the vegetation map creation apparatus according to the first embodiment. As illustrated in FIG. 1, the vegetation map creating apparatus 100 includes a storage unit 110 and a control unit 120. In addition, the vegetation map creating apparatus 100 is connected to an input apparatus 101 and an output apparatus 102. The vegetation map creating apparatus 100 corresponds to, for example, a personal computer.

  The input device 101 receives input of various information. For example, the input device 101 receives an HSD (Hyper-Spectral Data) 111 obtained by photographing an area to be investigated, and sends the received HSD 111 to the receiving unit 121 described later. The input device 101 receives position information indicating a point where each tree species to be investigated is vegetated, and sends the received position information to the receiving unit 121 described later. In addition, the input device 101 receives an input of an instruction to start processing by the user, and sends the input of the received instruction to the reception unit 121. The input device 101 corresponds to a keyboard, a mouse, a medium reading device, or various sensors such as a thermometer, a hygrometer, and a hyperspectral sensor. In addition, although the case where the vegetation map creation apparatus 100 creates a vegetation map with one HSD 111 as a processing target will be described here, the disclosed technique is not limited thereto. For example, the vegetation map creating apparatus 100 may divide a region to be investigated into a plurality of regions and use a plurality of HSDs 111 photographed simultaneously for each region as a processing target.

  The output device 102 outputs various information. For example, the output device 102 outputs information on the tree species specified by the specifying unit 126 described later. The output device 102 corresponds to a display, a monitor, or the like.

  The storage unit 110 has an HSD 111. The storage unit 110 corresponds to, for example, a semiconductor memory device such as a random access memory (RAM), a read only memory (ROM), and a flash memory, and a storage device such as a hard disk or an optical disk.

  The HSD 111 is data having a three-dimensional configuration that includes two-dimensional elements as an image and elements as spectrum data. For example, the HSD 111 has spectral data including wavelength information and light intensity or reflectance information for each pixel of the image. For example, the HSD 111 is captured by a hyperspectral sensor. The HSD 111 is also referred to as hyperspectral data.

  FIG. 2 is a diagram illustrating an example of the HSD. FIG. 2 shows an example of an image showing the HSD 111 in pseudo RGB (Red-Green-Blue color model). The HSD 111 is photographed by, for example, a hyperspectral sensor in which an area to be investigated is mounted on an aircraft. Each pixel of the HSD 111 has spectral data corresponding to a tree species covering each area. In addition, although the case where HSD111 image | photographed with the hyperspectral sensor mounted in the aircraft was demonstrated here, this invention is not limited to this. For example, the HSD 111 may be taken by a hyperspectral sensor mounted on an artificial satellite or a vehicle traveling on the ground, or may be taken by a human hand. The HSD 111 may be acquired from an external storage device that stores the HSD 111 taken in the past.

  With reference to FIG. 3, the spectral data of the HSD will be described. FIG. 3 is a diagram illustrating an example of spectrum data included in the HSD. In FIG. 3, the horizontal axis indicates the wavelength, and the vertical axis indicates the spectral intensity. In FIG. 3, the case where it image | photographs with the hyper spectrum sensor whose measurement wavelength range is 400 nm to 900 nm is shown. In FIG. 3, spectrum data 3a shows an example of cypress spectrum data. The spectrum data 3b is an example of larch spectrum data. Moreover, the spectrum data 3c shows an example of spectrum data of deciduous broad-leaved trees. Note that the measurement wavelength range of the hyperspectral sensor used for imaging is not limited to 400 nm to 900 nm. For example, the measurement wavelength range of the hyperspectral sensor may be appropriately selected according to the tree species in the area to be investigated. In addition to spectral intensity, for example, relative reflectance may be used as spectral data. The relative reflectance is a value obtained by removing the spectral component of the light source from the spectral intensity.

  The control unit 120 includes a reception unit 121, an extraction unit 122, a collation unit 123, a first classification unit 124, a second classification unit 125, and a specification unit 126. The function of the control unit 120 can be realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA). Further, the function of the control unit 120 can be realized, for example, by a CPU (Central Processing Unit) executing a predetermined program.

  The receiving unit 121 receives various types of information. For example, the reception unit 121 receives the HSD 111 in which the area to be investigated is captured, and stores the received HSD 111 in the storage unit 110. In addition, the reception unit 121 receives position information indicating a point where each tree species to be investigated is vegetated, and outputs the received position information to the extraction unit 122. In addition, the reception unit 121 receives an instruction to start processing by the user from the input device 101 and sends the received instruction to the extraction unit 122.

  The extraction unit 122 extracts a reference spectrum. For example, the extraction unit 122 extracts the reference spectrum of each tree type based on the HSD 111 in which the area to be surveyed is photographed and position information indicating a point where each tree type to be surveyed is vegetated. The extraction unit 122 outputs the reference spectrum of each tree species to the first classification unit 124 and the second classification unit 125 described later.

  As an example, the extraction unit 122 acquires from the storage unit 110 the HSD 111 in which the area to be investigated is captured. The extraction unit 122 receives, from the reception unit 121, position information indicating a point where each tree species to be investigated is vegetated. The extraction unit 122 extracts the spectrum data of the pixel corresponding to the point where a certain tree species is vegetated in the HSD 111 as a reference spectrum of a certain tree species.

  As an aspect, the extraction unit 122 receives position information indicating a point where cypress is vegetated from the reception unit 121. The extraction unit 122 extracts spectrum data included in pixels corresponding to a point where cypress is vegetated from the HSD 111 illustrated in FIG. 2 as a cypress reference spectrum. The cypress reference spectrum extracted here is, for example, the spectrum data 3a shown in FIG. In addition, the positional information which shows the point where each tree type used as investigation object vegetates is acquired when the investigator who performs vegetation investigation visits the field directly, and distinguishes the tree type of each area visually.

  The collation unit 123 collates the HSD 111 and the reference spectrum. For example, when the collation unit 123 receives an instruction to start processing from the reception unit 121, the spectral data of each pixel included in the HSD 111 in which the area to be investigated is captured, and the reference spectrum extracted by the extraction unit 122 Is matched. When the collation unit 123 collates all the pixels included in the HSD 111, the collation unit 123 outputs a collation result to a first classification unit 124 and a second classification unit 125 described later. In addition, the collation part 123 collates HSD111 and a reference spectrum, for example using the Spectral Angle Mapper method.

  The process by the collation part 123 is demonstrated using FIG. 4A-FIG. 4D. 4A to 4D are diagrams for explaining processing by the matching unit 123. FIG. FIG. 4A shows pixels determined to have spectrum data similar to the cypress reference spectrum in the HSD 111 of FIG. FIG. 4B shows pixels determined to have spectrum data similar to the larch reference spectrum in the HSD 111 of FIG. FIG. 4C shows pixels determined to have spectrum data similar to the reference spectrum of the deciduous broad-leaved tree in the HSD 111 of FIG. FIG. 4D is a superposition of FIGS. 4A to 4C. In addition, the white area | region of FIG. 4D is an area | region whose tree species is unknown.

  As one aspect, the collation unit 123 collates the spectrum data of each pixel included in the HSD 111 in FIG. 2 with the spectrum data 3a in FIG. 3 which is a cypress reference spectrum. As shown in FIG. 4A, the matching unit 123 identifies pixels having spectrum data similar to the cypress reference spectrum in the HSD 111. Further, the collation unit 123 collates the reference spectra of larch and deciduous broad-leaved trees with the spectrum data of each pixel included in the HSD 111 in FIG. 2 and identifies each pixel in FIG. 4B and FIG. Then, as illustrated in FIG. 4D, the collation unit 123 outputs the collation result obtained by superimposing FIGS. 4A to 4C to the first classification unit 124 and the second classification unit 125.

  The first classifying unit 124 classifies the spectrum data of each tree type into a pattern based on the spectrum data whose tree type is known and the spectrum data determined to be the same as the tree type. For example, the first classification unit 124 normalizes spectrum data having a known tree type and spectrum data determined to be the same type as the tree type among images captured by a hyperspectral camera. The first classification unit 124 classifies the spectrum data of the tree species into patterns by using the normalized correlation of each spectrum data. The first classification unit 124 outputs the classified result to the specifying unit 126 described later.

  Hereinafter, an example of processing of the first classification unit 124 will be described. The first classifying unit 124 acquires, from the HSD 111, the reference spectrum of each tree type and the spectrum data of pixels determined to be the same type as each tree type. As an example, the first classification unit 124 obtains the reference spectrum of cypress and the spectrum data of the pixel determined as the cypress in FIG. 4A from the HSD 111. The spectrum data acquired here is a plurality of spectrum data illustrated in FIG. 5A. In other words, each waveform illustrated in FIG. 5A represents the spectral data of each pixel included in the area indicated by the oblique lines in FIG. 4A. Moreover, the 1st classification | category part 124 acquires each reference | standard spectrum and the spectrum data determined to be the same kind as each tree kind from HSD111 also about a larch, deciduous broad-leaved tree, and a cedar like a cypress. The acquired larch spectrum data is, for example, the spectrum data of FIG. 5B, the deciduous broadleaf tree spectrum data is the spectrum data of FIG. 5C, and the cedar spectrum data is the spectrum data of FIG. 5D. 5A to 5D are diagrams for explaining processing by the first classification unit. 5A to 5D, the horizontal axis indicates the wavelength, and the vertical axis indicates the spectral intensity.

  As one aspect, the first classification unit 124 normalizes the acquired spectrum data of each tree species to the maximum value. For example, the first classification unit 124 normalizes the acquired hinoki spectrum data with a maximum value with reference to the peak value near 570 nm which is the maximum in FIG. 5A. In other words, the first classification unit 124 normalizes the spectrum intensity of each spectrum data so that the peak values near 570 nm of all the spectrum data included in FIG. 5A are the same as the reference peak value. Here, the spectrum data normalized to the maximum value is, for example, a plurality of spectrum data illustrated in FIG. 6A. Moreover, the 1st classification | category part 124 normalizes the acquired spectrum data by maximum value also about a larch, a deciduous broad-leaved tree, and a cedar like a cypress. Here, the larch spectrum data normalized to the maximum value is, for example, the spectrum data of FIG. 6B, the deciduous broad-leaved tree spectrum data is the spectrum data of FIG. 6C, and the cedar spectrum data is the spectrum data of FIG. 6D. It is. 6A to 6D are diagrams for explaining processing by the first classification unit. 6A to 6D, the horizontal axis indicates the wavelength, and the vertical axis indicates the maximum value normalized spectrum intensity. Although the case where the first classification unit 124 performs normalization using maximum value normalization has been described here, the present invention is not limited to this. For example, the first classification unit 124 may normalize using the sunlight spectrum on the same day as the day when the HSD 111 was captured.

  The first classifying unit 124 calculates Pearson's product-moment correlation coefficient between arbitrarily selected spectrum data and other spectrum data among the spectrum data normalized for the maximum value for each tree species. For example, the first classification unit 124 selects one arbitrary spectrum data from the spectrum data normalized for the maximum value for each tree type. The first classifying unit 124 calculates the Pearson product moment correlation coefficient between the selected spectrum data and the other spectrum data using the following equation (1).

In equation (1), r corresponds to the Pearson product moment correlation coefficient. x corresponds to the wavelength. y corresponds to the maximum normalized spectral intensity. n corresponds to the number of data in the spectrum data. x (-) corresponds to an arithmetic average of data x = {x _i } (i = 1, 2, 3,... n). y (−) corresponds to the arithmetic mean of the data y = {y _i } (i = 1, 2, 3... n).

  The first classification unit 124 classifies spectrum data in which r is equal to or greater than a threshold value into the same pattern as the selected spectrum data. As illustrated in FIG. 7, the first classification unit 124 calculates an average value 7a and a standard deviation of r calculated for cypress. The first classification unit 124 calculates “average value−k × standard deviation” as the threshold 7b. Here, k is an arbitrary coefficient. The first classification unit 124 classifies the spectrum data of r that is equal to or greater than the threshold 7b into the same pattern as the selected spectrum data. In addition, FIG. 7 is a figure for demonstrating the process by a 1st classification | category part. The vertical axis in FIG. 7 indicates the value of the correlation coefficient.

  The first classification unit 124 repeatedly performs the process of classifying the spectrum data that has not been classified as the same pattern into the above-described pattern until all the spectrum data is classified into any pattern. For example, the first classification unit 124 classifies the cypress spectrum data of FIG. 6A into three patterns as shown in FIGS. 8A to 8C. 8A to 8C are diagrams for explaining processing by the first classification unit. 8A to 8C, the horizontal axis indicates the wavelength, and the vertical axis indicates the maximum value normalized spectrum intensity.

  The first classification unit 124 calculates an average value of the spectrum data classified into the same pattern and classifies it into the pattern. For example, the first classification unit 124 calculates the average value of the cypress spectrum data in FIG. 8A and acquires the spectrum data 9a in FIG. 9A. Also, the first classification unit 124 calculates the average value of the cypress spectrum data in FIG. 8B and acquires the spectrum data 9b in FIG. 9A. The first classifying unit 124 calculates the average value of the cypress spectrum data in FIG. 8C and obtains the spectrum data 9c in FIG. 9A. In this way, the same kind of spectral data is classified into multiple patterns in this way, even for the same tree species, depending on the amount of water at each point and the nature of the soil, This is because there is a difference in hue. 9A to 9C are diagrams for explaining processing by the first classification unit. 9A to 9C, the horizontal axis indicates the wavelength, and the vertical axis indicates the maximum value normalized spectrum intensity.

  Moreover, the 1st classification | category part 124 classify | categorizes the spectrum data of a larch, deciduous broad-leaved tree, and a cedar into a pattern similarly to a cypress. For example, the first classification unit 124 classifies the larch spectrum data of FIG. 6B into the spectrum data 9d, 9e, and 9f of FIG. 9B, respectively. For example, the first classification unit 124 classifies the spectrum data of the deciduous broad-leaved tree in FIG. 6C into the spectrum data 9g and 9h in FIG. 9C, respectively. For example, the first classification unit 124 classifies the cedar spectrum data in FIG. 6D into the spectrum data 9i and 9j in FIG. 9D. Thus, the 1st classification | category part 124 classify | categorizes the spectrum data of each tree species into a pattern.

  Returning to the description of FIG. The second classifying unit 125 classifies regions with unknown tree species into groups using the correlation of spectral data of regions with unknown tree types. For example, the second classifying unit 125 may calculate the Pearson product moment correlation coefficient between the spectral data arbitrarily selected from the spectral data of the pixels included in the white area in FIG. 4D and the other spectral data. Each r is calculated. The second classification unit 125 classifies the spectrum data in which r is equal to or greater than the threshold into the same group as the arbitrarily selected spectrum data. For spectrum data whose r is less than the threshold value, the second classifying unit 125 selects again any spectrum data from those spectrum data, and repeats the process until it is classified into any group. In addition, the threshold value set here may be arbitrarily set by a person using the vegetation map creating apparatus 100.

  As an aspect, the second classification unit 125 classifies pixels included in the white area in FIG. 4D into four groups as illustrated in FIG. The second classification unit 125 calculates the average value of the spectrum data of each classified group, and acquires the spectrum data 11a, 11b, 11c, and 11d as shown in FIG. 11A. The second classification unit 125 normalizes the acquired spectrum data 11a, 11b, 11c, and 11d, respectively, to obtain the spectrum data 11e, 11f, 11g, and 11h in FIG. 11B. The second classification unit 125 outputs the spectrum data normalized to the maximum value to the specifying unit 126.

  Here, a case has been described in which the second classification unit 125 classifies a region of unknown tree species into a group using the Pearson product-moment correlation coefficient, but the disclosed technique is not limited to this. For example, the second classifying unit 125 may classify regions with unknown tree species into groups using, for example, the Spectral Angle Mapper method. For example, the second classification unit 125 extracts arbitrary spectrum data from a region where the tree species is unknown instead of the reference spectrum, and collates the spectrum data with the unknown tree species with other spectrum data, so that the tree species is unknown. Classify areas into groups.

  For example, the specifying unit 126 compares the pattern classified by the first classifying unit 124 with the spectrum data of each group classified by the second classifying unit 125, and specifies the tree type of the region where the tree type is unknown. As an example, the specifying unit 126 calculates a correlation coefficient with the pattern classified by the first classification unit 124 for each group classified by the second classification unit 125. The specifying unit 126 specifies the tree species corresponding to the pattern having the highest correlation coefficient as the tree species to be vegetated in the group area.

  Processing performed by the specifying unit 126 will be described with reference to FIG. FIG. 12 is a diagram for explaining processing by the specifying unit. FIG. 12 is a table showing correlation coefficients between tree species. 12 corresponds to the spectrum data 9i in FIG. 9D. The cedar Type B corresponds to the spectrum data 9j in FIG. 9D. Cypress Type A corresponds to the spectrum data 9a in FIG. 9A. Cypress Type B corresponds to the spectrum data 9b in FIG. 9A. Cypress TypeC corresponds to the spectrum data 9c in FIG. 9A. Larch Type A corresponds to the spectral data 9d in FIG. 9B. Larch Type B corresponds to the spectrum data 9e in FIG. 9B. Larch TypeC corresponds to the spectrum data 9f in FIG. 9B. Deciduous broad-leaved tree Type A corresponds to the spectral data 9g of FIG. 9C. Deciduous broad-leaved tree Type B corresponds to the spectrum data 9h in FIG. 9C. Area 1 corresponds to the spectrum data 11e of FIG. 11B. Area 2 corresponds to the spectrum data 11f of FIG. 11B. Area 3 corresponds to the spectrum data 11g in FIG. 11B. Area 4 corresponds to the spectrum data 11h in FIG. 11B.

  As illustrated in FIG. 12, the specifying unit 126 calculates correlation coefficients between the area 1 and cedar, cypress, larch, and deciduous broad-leaved trees. As an example, the identifying unit 126 calculates the Pearson product moment correlation coefficient “0.96993” between the area 1 and the cedar Type A, and the Pearson product moment correlation coefficient “0.93654” between the area 1 and the cedar Type B. To do. Since the pattern having the highest correlation coefficient among the correlation coefficients calculated for area 1 is “Hinoki Type A”, the specifying unit 126 specifies the tree species to be vegetated in the area 1 as “Hinoki”. Similarly to area 1, the specifying unit 126 calculates correlation coefficients with patterns classified for the areas 2, 3, and 4, respectively, and specifies the tree species to be vegetated in each region. For example, the identifying unit 126 identifies a tree species to be vegetated in the area 2 area as “cedar”, identifies a tree species to be vegetated in the area 3 area as “cedar”, and designates a tree species to be vegetated in the area 4 area as “cypress”. ". The specifying unit 126 applies the specified tree species to the image of FIG. 10 to generate the image of FIG. The specifying unit 126 outputs the generated image of FIG. 13 to the output device 102. FIG. 13 is a diagram for explaining processing by the specifying unit.

  Next, a processing procedure of the vegetation map creating apparatus 100 according to the first embodiment will be described. FIG.14 and FIG.15 is a flowchart which shows the process sequence of the vegetation map creation apparatus 100. FIG. The process illustrated in FIG. 14 is executed, for example, when the reception unit 121 receives an instruction to start the process by the user.

  As illustrated in FIG. 14, when the processing timing comes (Yes in Step S101), the extraction unit 122 extracts a reference spectrum (Step S102). Subsequently, the collation unit 123 collates the HSD 111 and the reference spectrum (step S103). The extraction unit 122 is in a standby state until the processing timing comes (No in step S101).

  The first classification unit 124 performs a pattern classification process for classifying the spectrum data of each tree type into patterns (step S104). Here, the pattern classification process will be described with reference to FIG.

  The first classifying unit 124 acquires, from the HSD 111, the reference spectrum of each tree type and the spectrum data of the pixels determined to be the same type as each tree type (Step S201). The first classification unit 124 normalizes the acquired spectrum data of each tree type by the maximum value (step S202).

  The first classification unit 124 selects one arbitrary spectrum data from the unclassified spectrum data (step S203). The first classification unit 124 calculates a correlation coefficient between the selected spectrum data and other spectrum data (step S204). The first classification unit 124 classifies spectrum data having a correlation coefficient equal to or greater than a threshold value into the same pattern as the selected spectrum data (step S205).

  The first classification unit 124 determines whether there is unclassified spectrum data (step S206). If there is unclassified spectrum data (Yes at step S206), the first classification unit 124 returns to the process of step S203. On the other hand, when there is no unclassified spectrum data (No at Step S206), the first classification unit 124 ends the pattern classification process.

  Returning to the description of FIG. The second classifying unit 125 classifies the regions with unknown tree species into groups using the correlation of the spectrum data of the regions with unknown tree species (step S105). The specifying unit 126 compares the pattern classified by the first classifying unit 124 with the spectrum data of each group classified by the second classifying unit 125, and specifies the tree type of the region where the tree type is unknown (step S106). ).

  Note that the processing procedures described above do not necessarily have to be executed in the above order. For example, after the process of step S103 for comparing the HSD 111 and the reference spectrum is executed, the process of step S105 for classifying the area of unknown tree species into a group is executed, and then step S104, which is a pattern classification process, is executed. May be.

  Next, the effect of the vegetation map creating apparatus 100 according to the first embodiment will be described. The vegetation map creating apparatus 100 classifies the spectrum data of each tree species into patterns based on the spectrum data of which tree species are known and the spectrum data determined to be the same species as the tree species. The vegetation map creating apparatus 100 compares the classified pattern spectrum data with the spectrum data of an area where the tree species is unknown, and identifies the tree species of the area. For this reason, the vegetation map creation apparatus 100 can create a vegetation map without performing an additional field survey. According to this, the vegetation map creating apparatus 100 can reduce the man-hours required for creating the vegetation map, and can avoid the danger associated with the investigator visiting the site directly.

  Although the embodiments of the present invention have been described so far, the present invention may be implemented in other embodiments besides the above-described embodiments. Therefore, other embodiments will be described below.

  In the above embodiment, a case has been described in which the vegetation map creating apparatus 100 specifies a tree species by a single process, but the present invention is not limited to this. For example, if the vegetation map creation apparatus 100 cannot identify the tree species of all regions, it can identify the tree species again using the spectrum data of the regions identified up to that point.

  Referring to FIG. 16, when the vegetation map creating apparatus 100 cannot identify tree species in all regions, a processing procedure in the case of identifying tree species using the spectrum data of the regions identified up to that point will be described. FIG. 16 is a flowchart illustrating a processing procedure of the vegetation map creation apparatus.

  As shown in FIG. 16, when the processing timing comes (Yes at Step S301), the extraction unit 122 extracts a reference spectrum (Step S302). Subsequently, the collation unit 123 collates the HSD 111 and the reference spectrum (step S303). The extraction unit 122 is in a standby state until the processing timing is reached (No at Step S301).

  The first classification unit 124 executes a pattern classification process for classifying the spectrum data of each tree type into a pattern (step S304). The second classifying unit 125 classifies the regions with unknown tree species into groups using the correlation of the spectrum data of the regions with unknown tree species (step S305). The identification unit 126 compares the pattern classified by the first classification unit 124 with the spectrum data of each group classified by the second classification unit 125, and identifies the tree species of the region where the tree species is unknown (step S306). ).

  The specifying unit 126 determines whether or not the tree species of all the regions have been specified for the HSD 111 to be investigated (step S307). If the tree species of all the regions cannot be specified (No at Step S307), the process proceeds to Step S304. At this time, the first classification unit 124 uses the spectrum data of the pixel whose tree type is specified in the process of step S306, the reference spectrum of each tree type, and the spectrum data of the pixel determined to be the same type as each tree type, Perform pattern classification processing. On the other hand, when the tree species of all regions can be identified (Yes at Step S307), the identifying unit 126 ends the process.

  In this way, the vegetation map creating apparatus 100 classifies the spectrum data of each tree species into patterns by further using the spectrum data of the tree species identified by the process of identifying the tree species. For this reason, the vegetation map creating apparatus 100 can accurately identify the tree species.

  Further, among the processes described in the first embodiment, all or a part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed can be performed. All or a part can be automatically performed by a known method. In addition, the processing procedures, control procedures, specific names, and information including various data and parameters shown in the above-described document and drawings can be arbitrarily changed unless otherwise specified.

  Moreover, each component of the vegetation map creating apparatus 100 shown in FIG. 1 is functionally conceptual and does not necessarily need to be physically configured as illustrated. In other words, the specific form of dispersion / integration of the vegetation map creating apparatus 100 is not limited to the illustrated one, and all or a part of the vegetation drawing creating apparatus 100 may be functional or physical in arbitrary units depending on various loads and usage conditions. Can be distributed and integrated.

  FIG. 17 is a diagram illustrating an example of a computer that executes a vegetation map creation program. As illustrated in FIG. 17, the computer 300 includes a CPU 301 that executes various arithmetic processes, an input device 302 that receives data input from a user, and a monitor 303. The computer 300 also includes a medium reading device 304 that reads a program or the like from a storage medium, an interface device 305 for connecting to another device, and a wireless communication device 306 for connecting to another device wirelessly. The computer 300 also includes a RAM (Random Access Memory) 307 that temporarily stores various types of information and a hard disk device 308. Each device 301 to 308 is connected to a bus 309.

  The hard disk device 308 stores a vegetation map creation program having the same functions as the processing units of the first classification unit 124 and the identification unit 126 shown in FIG. The hard disk device 308 stores various data for realizing a vegetation map creation program.

  The CPU 301 reads out each program stored in the hard disk device 308, develops it in the RAM 307, and executes it to perform various processes. In addition, these programs can cause the computer 300 to function as the first classification unit 124 and the identification unit 126 illustrated in FIG.

  Note that the above vegetation map creation program is not necessarily stored in the hard disk device 308. For example, the computer 300 may read and execute a program stored in a recording medium readable by the computer 300. The computer-readable recording medium corresponds to, for example, a portable recording medium such as a CD-ROM, a DVD disk, a USB (Universal Serial Bus) memory, a semiconductor memory such as a flash memory, a hard disk drive, and the like. Further, the program may be stored in a device connected to a public line, the Internet, a LAN (Local Area Network), a WAN (Wide Area Network), etc., and the computer 300 may read and execute the program therefrom. good.

  The following supplementary notes are further disclosed with respect to the embodiments including the above examples.

(Supplementary note 1)
Based on the spectrum data of known tree species and the spectrum data determined to be the same species as the tree species, the spectrum data of each tree species is classified into patterns,
A vegetation map creating program characterized in that the spectrum data of the classified pattern is compared with the spectrum data of an area where the tree type is unknown, and each process for specifying the tree type of the area is executed.

(Additional remark 2) The process classified into the said pattern normalizes the maximum value of the spectrum data by which the tree type was known among the images imaged with the hyper spectrum camera, and the spectrum data determined to be the same type as the tree type, The vegetation map creating program according to appendix 1, wherein the spectral data of the tree species is classified into patterns using correlation.

(Additional remark 3) The process classified into the said pattern is the product of the Pearson between the spectrum data selected arbitrarily among spectrum data with which the tree species is known, and the spectrum data determined to be the same species as the tree species. The vegetation map creation program according to appendix 1 or 2, wherein a rate correlation coefficient is calculated, and spectrum data having the coefficient equal to or greater than a threshold is classified into the same pattern as the arbitrarily selected spectrum data.

(Additional remark 4) The process classified into the said pattern uses the value which subtracted the standard deviation of the said coefficient from the average value of the said coefficient as said threshold value, The vegetation map preparation program of Additional remark 3 characterized by the above-mentioned.

(Supplementary note 5) Further causing the computer to classify the areas into groups using the correlation of spectral data of each area,
The process of specifying the tree species compares the spectrum data of the group and the spectrum data of the pattern, and specifies the tree species of the area, according to any one of appendices 1 to 4, Vegetation map creation program.

(Additional remark 6) The process classified into the said group calculates the Pearson product-moment correlation coefficient between the spectrum data arbitrarily selected among the spectrum data whose tree species are unknown, and other spectrum data, respectively, The vegetation map creating program according to appendix 5, wherein spectrum data having a threshold value equal to or greater than a threshold value is classified into the same group as the arbitrarily selected spectrum data.

(Additional remark 7) The process classified into the said pattern further classify | categorizes the spectral data of each tree species into a pattern using the spectral data of the tree species specified by the process which specifies the said tree species further of the additional marks 1-6 characterized by the above-mentioned. The vegetation map creation program according to any one of the above.

(Appendix 8) A pattern classification unit that classifies the spectrum data of each tree species into patterns based on the spectrum data of which tree species are known and the spectrum data determined to be the same species as the tree species,
A vegetation map creating apparatus, comprising: a specifying unit that compares the spectrum data of the classified pattern and the spectrum data of an area where the tree type is unknown, and specifies the tree type of the area.

(Additional remark 9) The said pattern classification | category part normalizes the maximum value of the spectrum data by which the tree type was known among the images image | photographed with the hyper spectrum camera, and the spectrum data determined to be the same kind as the said tree type, and correlates each spectrum data. The vegetation map creating apparatus according to appendix 8, wherein the spectral data of the tree species is classified into patterns.

(Additional remark 10) The said pattern classification | category part is Pearson's product moment phase between the spectrum data arbitrarily selected among the spectrum data with which tree species are known, and the spectrum data determined to be the same species as the said tree species The vegetation map creating apparatus according to appendix 8 or 9, wherein the number of relationships is calculated, and the spectrum data having the coefficient equal to or greater than a threshold is classified into the same pattern as the arbitrarily selected spectrum data.

(Additional remark 11) The said pattern classification | category part uses the value which subtracted the standard deviation of the said coefficient from the average value of the said coefficient as said threshold value, The vegetation map preparation apparatus of Additional remark 10 characterized by the above-mentioned.

(Additional remark 12) It further has a group classification part which classifies the area into a group using correlation of spectrum data of each area,
The said specific | specification part compares the spectrum data of the said group, and the spectrum data of the said pattern, specifies the tree species of the said area, The vegetation map preparation as described in any one of the additional remarks 8-11 characterized by the above-mentioned. apparatus.

(Additional remark 13) The said group classification | category part calculates the Pearson's product-moment correlation coefficient between the spectrum data arbitrarily selected among spectrum data with unknown tree | species and other spectrum data, respectively, and the said coefficient is a threshold value The vegetation map creating apparatus according to appendix 12, wherein the spectrum data as described above is classified into the same group as the arbitrarily selected spectrum data.

(Additional remark 14) The said pattern classification | category part further classify | categorizes the spectral data of each tree species into a pattern using the spectral data of the tree species specified by the process which specifies the said tree species, The any one of Additional remarks 8-13 characterized by the above-mentioned. The vegetation map creation device according to one.

(Supplementary note 15) A vegetation map creation method executed by a computer,
Computer
Based on the spectrum data of known tree species and the spectrum data determined to be the same species as the tree species, the spectrum data of each tree species is classified into patterns,
A method for creating a vegetation map, comprising comparing the spectrum data of the classified pattern with the spectrum data of an area where the tree species is unknown, and executing each process for specifying the tree species of the area.

(Additional remark 16) The process classified into the said pattern normalizes the maximum value of the spectrum data by which the tree type is known among the images image | photographed with the hyper spectrum camera, and the spectrum data determined to be the same kind as the said tree type, The vegetation map creating method according to appendix 15, wherein the spectrum data of the tree species is classified into patterns using correlation.

(Additional remark 17) The process classified into the said pattern is the Pearson product between the spectrum data arbitrarily selected among the spectrum data with which tree species are known, and the spectrum data determined to be the same species as the tree species. The vegetation map creating method according to appendix 15 or 16, wherein a rate correlation coefficient is calculated, and spectrum data having the coefficient equal to or greater than a threshold is classified into the same pattern as the arbitrarily selected spectrum data.

(Additional remark 18) The process classified into the said pattern uses the value which subtracted the standard deviation of the said coefficient from the average value of the said coefficient as said threshold value, The vegetation map preparation method of Additional remark 17 characterized by the above-mentioned.

(Supplementary note 19) Further causing the computer to classify the areas into groups using the correlation of spectral data of each area,
The process of specifying the tree species is performed by comparing the spectrum data of the group with the spectrum data of the pattern, and specifying the tree species of the area. How to create a vegetation map.

(Additional remark 20) The process classified into the said group calculates the Pearson product-moment correlation coefficient between the spectrum data arbitrarily selected among the spectrum data whose tree species are unknown, and other spectrum data, respectively, The method for creating a vegetation map according to appendix 19, wherein spectrum data having a threshold value equal to or greater than a threshold is classified into the same group as the arbitrarily selected spectrum data.

(Additional remark 21) The process classified into the said pattern further classify | categorizes the spectral data of each tree species into a pattern further using the spectral data of the tree species specified by the process which specifies the said tree species, The additional notes 15-20 characterized by the above-mentioned The vegetation map creation method according to any one of the above.

100 Vegetation map creation device 101 Input device 102 Output device 110 Storage unit 111 HSD
120 control unit 121 reception unit 122 extraction unit 123 collation unit 124 first classification unit 125 second classification unit 126 identification unit 300 computer 301 CPU
302 Input Device 303 Monitor 304 Medium Reading Device 305 Interface Device 306 Wireless Communication Device 307 RAM
308 Hard disk device 309 Bus

Claims (9)

On the computer,
Based on the spectrum data of known tree species and the spectrum data determined to be the same species as the tree species, the spectrum data of each tree species is classified into patterns,
A vegetation map creating program characterized in that the spectrum data of the classified pattern is compared with the spectrum data of an area where the tree type is unknown, and each process for specifying the tree type of the area is executed.   The processing for classifying the pattern is performed by normalizing the spectrum data having a known tree species and the spectrum data determined to be the same species as the tree species out of the images taken by the hyperspectral camera, and using the correlation of each spectrum data. The vegetation map creation program according to claim 1, wherein the spectrum data of the tree species is classified into patterns.   The process of classifying into the pattern is a Pearson product moment correlation coefficient between spectral data selected arbitrarily among spectral data with a known tree species and spectral data determined to be the same species as the tree species. The vegetation map creating program according to claim 1 or 2, wherein the spectrum data having the coefficient equal to or greater than a threshold is classified into the same pattern as the arbitrarily selected spectrum data.   The vegetation map creating program according to claim 3, wherein the process of classifying the patterns uses a value obtained by subtracting a standard deviation of the coefficient from an average value of the coefficient as the threshold. Further causing the computer to classify the areas into groups using the correlation of the spectral data of each area;
The process for identifying the tree species identifies the tree species of the area by comparing the spectrum data of the group and the spectrum data of the pattern. Vegetation map creation program.   The process of classifying the groups is to calculate Pearson's product moment correlation coefficient between arbitrarily selected spectrum data and other spectrum data among the spectrum data with unknown tree species, and the coefficient is equal to or greater than a threshold value. The vegetation map creating program according to claim 5, wherein the spectrum data is classified into the same group as the arbitrarily selected spectrum data.   The process for classifying into the patterns further classifies the spectrum data of each tree species into patterns by further using the spectrum data of the tree species specified by the process of specifying the tree species. The vegetation map creation program described in 1. A pattern classification unit that classifies the spectrum data of each tree species into patterns based on the spectrum data of known tree species and the spectrum data determined to be the same species as the tree species;
A vegetation map creating apparatus, comprising: a specifying unit that compares the spectrum data of the classified pattern and the spectrum data of an area where the tree type is unknown, and specifies the tree type of the area. A vegetation map creation method executed by a computer,
Computer
Based on the spectrum data of known tree species and the spectrum data determined to be the same species as the tree species, the spectrum data of each tree species is classified into patterns,
A method for creating a vegetation map, comprising comparing the spectrum data of the classified pattern with the spectrum data of an area where the tree species is unknown, and executing each process for specifying the tree species of the area.