JP2012196167A - Plant species identification method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the following problem: in a conventional optical remote sensing measurement, it is not easy to specify a plant species of a plant of a measurement region, and it is difficult even by a method using the Normalized Differential Vegetation Index (NDVI), for example.SOLUTION: A spectrum intensity graph of image pixels wherein a target plant species is present is obtained by hyper spectrum image photography in one season, and a plurality of (e.g. four) indexes are acquired by arithmetic processing of a plurality of (e.g. five) values each indicating the local maximum or the local minimum by use of them appearing in a specific wavelength band in common in each plant species in the graph. A DB of a plant species specification index group thereof is generated, furthermore, a plant species specification index group in a different season of an unknown plant is also obtained, and a DB thereof is generated. The plant species specification index group of the one season of the unknown plant is obtained, and is compared with the plant species specification index group inside the DB. With the according plant species specification index group, the plant species is specified, and when there is no found plant species specification index group, the plant species specification index group in another different season and the DB in the same season are compared to specify the plant species.

Description

  The present invention relates to a method for identifying and identifying plant species, particularly in vegetation situation surveys using remote sensing.

  Conventional methods of vegetation survey in areas where trees and other plants exist, including forests, mountains, satoyama, fields, urban parks, etc., as well as conventional methods such as satellites and aircraft There is a method (remote sensing) method of discriminating using reflection information from vegetation, and each is performed independently or in combination.

  Remote sensing using satellites and aircraft is roughly divided into a method of detecting the intensity of reflected waves from vegetation by radiating radio waves, and optical detection of the intensity of reflected light from vegetation such as sunlight. There is a way to do it. The latter method using reflected light is widely used for various vegetation surveys from a relatively wide observation range to a narrow range.

  Sensing methods used for remote sensing by optical detection have shifted from panchromatic (detected in black and white) to multispectral (detected in several bands, for example, 2 to 4 band color spectra) methods. Photographs of reflected reflections of sunlight are taken by specialists using the intensity difference between black and white in the panchromatic method and spectral intensity difference in the multispectral method, and vegetation identification is performed. A wide range of vegetation maps have been created from the vegetation identification using this photograph and the geographic information system GIS (Geographic Information System).

  In recent years, hyperspectral sensors have been put to practical use that can measure with the number of bands more than 10 times the multispectrum, and the amount of information obtained has increased dramatically. In addition, the global environment satellite (satellite name EQ-1; onboard sensor name Hyperion, satellite name PROBA; onboard sensor name CHRIS) equipped with this hyperspectral sensor will be launched. Has come to be done.

JP 2001-357380 A JP 2005-189099 A JP 2006-250827 A JP 2006-314215 A

  Recently, I heard reports that exotic plants are destroying local plants. If there is such a report, it is necessary to investigate the vegetation status of the land in more detail, including the type of vegetation. When investigating the types of vegetation plants by conventional remote sensing, the judgment of the type requires the judgment of a skilled expert who has relevant knowledge including sensing performance. The type determination is not always easy because there are cases where the type determination is different. In addition, it is necessary to confirm by site survey each time it is difficult to determine the type of vegetation from remote sensing. As described above, it takes a lot of time to create a vegetation map in the observation area based on the vegetation survey. As a result, when the vegetation map was completed, it was possible that there was already a discrepancy from the current situation.

  In the vegetation survey, a normalized vegetation index (NDVI) is widely used as an index representing the degree of coverage of vegetation in a wide range of survey areas. NDVI utilizes the property that the green leaves of plants absorb the wavelengths of the blue region and the red region and strongly reflect the wavelengths of the near infrared region, and are expressed by the following formula (Formula 1).

NDVI = (IR-R) / (IR + R) (Equation 1)
Here, R is the reflectance in the visible region red, and IR is the reflectance in the near infrared region.

  NDVI indicates a normalized value between -1 and +1, and a positive positive number indicates, for example, that the vegetation is darker.

  As can be seen from Equation 1, this NDVI index can be applied to the sensing results of the multispectral method, which uses the reflectance in the two spectral bands of the visible region red and the near infrared region, respectively. It is also a possible index.

  This NDVI is effective in identifying plants and non-plants such as rivers, roads, and buildings, but currently the plant type cannot be clearly identified. In addition to the above-described examples of determining the vegetation status of foreign plants, the identification of plant species in the vegetation region and the creation of vegetation maps based thereon are very important regardless of the size of the target area.

  Therefore, an object of the present invention is to provide a method capable of specifying a plant species in a vegetation region with relatively high accuracy even in remote sensing.

The plant species identification method of the present invention comprises:
Spectra in other time periods including a predetermined time period using an optical sensor device that can observe the spectral intensity of the imaging region where the plant species to be measured are vegetated in a predetermined observation wavelength range with a wavelength band of at least several tens. Taking a picture, and
In the reflection image at the pixel position where the measurement target plant species is photographed in the spectrum image, a plurality of spectral intensity extreme bands indicating a spectral intensity extreme value that is a maximum value or a minimum value of the spectral intensity is acquired. Steps,
Extracting each singular spectral intensity extreme value in a plurality of singular spectral intensity extreme bands out of the spectral intensity extreme bands;
Obtain a plant species specific index group consisting of a plurality of plant species indexes of the measurement target plant species, which is calculated by a calculation process using a plurality of predetermined unique spectrum intensity extreme values among the specific spectrum intensity extreme values. Steps,
The plant species identification index group of the predetermined time of the unknown measurement target plant species is acquired, and compared with the plant species identification index group acquired from the known plant species, the plant species of the unknown measurement target plant species Identifying steps;
If the plant species identification is impossible, the step of identifying the plant species of the unknown measurement target plant species by the respective steps after the step of capturing the spectrum image of the other time period different from the predetermined time period; It is characterized by having.

  Since the method of the present invention performs hyperspectral image capturing on the measurement range, performs spectral intensity measurement for each pixel, and further uses the captured results at multiple times, it is relatively easy and high. It has an effect that the plant species can be specified by the photographing pixel with accuracy.

Diagram explaining remote sensing The figure explaining the example of the spectrum intensity graph for every plant The figure explaining the example of the singular point in the spectrum intensity graph Diagram explaining examples of multiple index values for each plant species Graph of each index example Graph of the index of 5 plant species Graph of the index of five plant species at different times Mapping example Mapping example Diagram explaining plant species specific flow

  Embodiments of the present invention will be described below with reference to the accompanying drawings.

  FIG. 1 is a schematic diagram showing an outline of photographing in remote sensing.

  As shown in FIG. 1, remote sensing is performed by photographing an object 3 that is a region such as a forest where a plant to be measured is vegetated from an aircraft 1 or an artificial satellite 2 equipped with a sensor (not shown), and the electromagnetic wave ( Light, etc.) is observed. In remote sensing by optical detection, the sensor detects reflection, scattering, or radiation from the ground surface or trees received from sunlight 4. The above example is remote sensing from the sky. For example, a sensor-equipped camera 5 installed far away from the ground is used to photograph a ground object 3 such as a tree in a forest or a park, and observe its vegetation. In some cases.

  As a sensor used for such remote sensing, a spectral image sensor that can measure the spectral characteristics of each pixel unit of a captured image of an object with high spatial and wavelength resolution is often used. For example, in the measurement wavelength range of 350 nm to 1050 nm, a device capable of performing hyperspectral measurement with multiple bands such as 141 bands in increments of 5 nm is used.

  Fig. 2 shows the reflection spectrum intensities of various types of plant species measured at a certain daytime time at a certain time using a sensor capable of hyperspectral measurement (hyperspectral sensor). It is a graph of. The horizontal axis is the wavelength (nm) and the vertical axis is the spectral intensity (arbitrary unit). In the case of FIG. 2, each of eight kinds of onigurumi, harienju, shirakashi, ginkgo, cedar, zelkova, mokoku, and someiyoshino The measured values for the trees are shown superimposed.

  The measurement result of FIG. 2 shows that the wavelength range where the singular point having the extreme value such as the maximum value and the minimum value appears in the same wavelength band in the reflection spectrum intensity of any kind of tree. I understand that.

  FIG. 3 is a schematic spectral intensity graph showing such singular point positions. In FIG. 3, the maximum value singular point A is at a position of 550 ± 5 nm in the visible light region, the minimum value singular point B is at 680 ± 10 nm near the red edge, and the maximum value singular point C is near red. At a position of 750 ± 5 nm in the outer region, a singular point D having a minimum value is at a position of 760 ± 5 nm in the near infrared region, and a singular point E having a maximum value is at a position of 770 ± 5 nm in the near infrared region. It was understood that each appeared. The appearance of this extreme value was a characteristic commonly obtained in other types of trees.

That is,
(1) Wavelength band BW _A of appearance (intensity A) of local maximum singularity _A : 550 ± 5 nm
(2) Wavelength band BW _B of appearance (intensity B) of minimum value singularity _B : 680 ± 10 nm
(3) Wavelength band BW _C of appearance (intensity C) of local maximum singularity _C : 750 ± 5 nm
(4) Wavelength band BWD of the appearance (intensity D) of the local minimum singularity _D : 760 ± 5 nm
(5) Wavelength band BW _E of appearance (intensity E) of local maximum singularity _E : 770 ± 5 nm
It is.

Next, the graph characteristics of the reflection spectra of various trees are expressed as four types of indices by the following arithmetic processing using each (peak) intensity in each wavelength band in which the above five kinds of extreme values appear. did. That is,
(1) Index 1: Ratio of strength C and strength A
That is: C / A
(2) Index 2; ratio of [difference between intensity C and intensity B] and [difference between intensity A and intensity B]
That is: (C−B) / (A−B)
The above expression is simplified and expressed as CB / AB.

(3) Index 3; [ratio of strength A to strength B] and [ratio of strength A to strength C] That is: (A / B) / (A / C)
(4) Index 4 [ratio of [difference between intensity C and intensity B] and [difference between intensity C and intensity D]], [difference between intensity C and intensity D], and [difference between intensity C and intensity E] Ratio], ie: ((C−B) / (C−D)) / ((C−D) / (C−E))
The above expression is simplified and expressed as (CB / CD) / (CD / CE).
And there.

  As described later, each of the four indexes is referred to as a plant species index in the meaning of expressing one reflection spectrum characteristic of a certain plant species in the work of identifying the plant species. A set of all four indices is referred to as a plant species-specific index group in the sense that it is distinguished from other plant species by four indices.

  FIG. 4 shows a reflection spectrum intensity measurement of a tree at a certain time period shown in FIG. 2 and a reflection spectrum intensity of a total of 13 kinds of trees of aogiri, camphor tree, quail, cinnamon, and weeping cherry at the same time period. The values of the plant species specific index group (that is, the four plant species indexes) obtained from the measured values are summarized. Here, the vertical column indicates the plant species name, and the horizontal column indicates each plant species index, that is, C / A, CB / AB, (A / B) / (A / C), (CB / CD) / (CD / CE).

  FIG. 5 shows four graphs for each plant species index shown in FIG. In either case, the horizontal axis represents the name of each plant species, and the vertical axis represents the plant species index.

  Using FIG. 4 and FIG. 5, each plant species index of each plant species is compared.

  First, when comparing the 4 index of Onigurumi and the other, it can be seen that, among the 4 indices of cedar, 2 indices have relatively close numbers. However, it can be seen that the walnut has a characteristic that both the C / A index and the CB / AB index have large numerical values. Based on this, a database (DB) that has been acquired in advance at the same time and accumulated plant species specific index groups such as various kinds of trees including oysters is created, and plant species specific index groups of unknown plants are acquired. Thus, it is easy to identify that it is a walnut by comparing it with the data in this DB.

  Similarly, comparing the Hariendi 4 index with other ones, it can be seen that, in particular, the 3 indices among the 4 indices of Cedar have relatively close numbers. However, the value of the (CB / CD) / (CD / CE) index is greatly different, and it can be said that identification from the Hariend DB is easy from this index tendency. From the above two types of identification, it is possible to differentiate between the plant species identification index group of cedar and other plants, in particular, the plant species identification index group of Onigurumi Harienju, and it is also possible to identify cedar.

  Next, when comparing the plant species-specific index group of birch and the plant species-specific index group of other plants, it can be said that there is a comparatively similar tendency to that of kinmokusei and mockoku. However, among them, the birch (A / B) / (A / C) has the remarkable feature that it is the largest in all other plants. On the other hand, Mockoku has both C / A and CB / AB values larger than the other two types. In addition, (A / B) / (A / C) of cinnamon is the smallest among all plants. For these two similar species, the comparison of these plant species-specific index groups shows that separation from other plant species is possible, including identification by separating Shirakashi, Kinmokusei, and Mokkuk.

  Examination of the Ginkgo plant species specific index group 4 index shows that there are many similar parts to that of the weeping cherry, but Ginkgo is particularly large in (A / B) / (A / C) index. Since the weeping cherry has a significantly larger (CB / CD) / (CD / CE) index than the ginkgo, both species of plants The species are clearly distinguishable.

  Thus, by comparing the examples of these plant species-specific index groups obtained in a specific time period, the eight species of Onigurumi, Harienju, Shirakashi, Ginkgo, Cedar, Kinmokusei, Weeping Cherry, and Mokkok are each Separately, it was possible to identify the type from at least all 13 types measured.

  FIG. 6 shows C / A, CB / AB, (A / B) / (A / C), (CB / CD) / (CD / CE) for the remaining five species of Aogiri, Kusunoki, Yuzuriha, zelkova, and Yoshino cherry. The comparative graph of the plant species specific index group 4 index is extracted and shown again. From this figure, even if each index value of each plant species is compared with each other and trying to identify the plant species as described above, the numerical trends are similar to each other and it is difficult to clearly distinguish the plant species from each other. .

  FIG. 7 is an example of plant species-specific index groups of five kinds of plants of Aogiri, camphor, quail, zelkova, and Yoshino cherry obtained at a different time from the time of obtaining FIG. By acquiring plant species-specific index groups at different times, for example, in the C / A index and CB / AB index, Aogiri and zelkova are significantly different from others and show a small value (A / B) In the / (A / C) index, the values of camphor and Yoshino are small, but Aogiri shows a large value, but Yoshino shows the smallest value, and in the (CB / CD) / (CD / CE) index, Yuzuriha, Yoshino However, zelkova has the next largest value, and Aogiri and camphor tree have the smaller value. From these combination comparisons, it is possible to specify five plant species.

  By using the plant species identification index group at different time periods, first, the unknown plant species that can be identified there are classified by comparison at the specific time period (comparison with the data in the DB). It can be seen that the accuracy of specifying the plant species is greatly improved by taking measures such as specifying the remaining unknown plant species by comparison at different times.

  Next, an example of vegetation mapping of the imaging region will be shown. This is an example of mapping a plant species to be measured using a planting site such as a tree as an observation target site at a certain time. In advance, a reflection spectrum for each plant species as shown in FIG. 2 and a plant species specific index group as shown in FIG. 4 (and FIG. 5) derived therefrom are obtained and converted into a DB.

  In order to create a vegetation mapping, according to the method described above, the spectrum intensity is photographed with a camera having a hyperspectral sensor, a plant species identification index group is obtained for each pixel in the captured image, and the value and the plant species identification index are obtained. Referring to the group DB, the plant species of the plant photographed at the pixel is specified. For example, by changing the display color for each plant species and displaying the color of the specified plant species for each pixel, the vegetation status of the plant species in the photographing range can be mapped.

  FIG. 8 is a diagram schematically showing a simple example of mapping for each plant species. In this case, two types of trees are overlapped in a standing tree state, and plant species A and plant species B is a color difference (or black-and-white difference), which can be displayed separately as shown in FIG. 8A or FIG. 8B, and it is possible to map the identification of each tree and the prosperity of each.

  FIG. 9A is an image of an ordinary color photograph taken in an actual park and converted into a monochrome image. It can be seen that a plurality of trees are densely grown. 9 (b) and 9 (c) show the same observation region using an image obtained by a camera having a hyperspectral sensor, specify the plant species for each pixel in the same manner as described above, and use the result as the plant. The color display for each species is shown. In this example, it can be seen that the white and ginkgo vegetation are adjacent to each other, and FIG. 9B shows an example in which the brightness of the pixel determined to be white is large and the ginkgo portion is small. On the contrary, FIG. 9C is an example in which the brightness of the pixel determined to be a ginkgo is increased and the white portion is decreased. As described above, the plant species identification method of the present invention can be easily applied to the mapping for each plant species in the imaging region.

  The specific examples of plant species as described above indicate that the plant species can be identified by remote sensing as follows.

  That is, for a known plant species, the reflection spectrum intensity of the plant species is measured with high wavelength resolution by remote sensing using a sensor capable of hyperspectral measurement such as a hyperspectral camera, and this is performed at every time. From the spectrum thus obtained, the maximum value and the minimum value of the reflection intensity appearing in the same wavelength band band common to each plant species are obtained in the measurement wavelength range, and the calculation is derived using each extreme value. Create a database for each plant species and time period, using multiple types of indices as plant species specific index groups, and measure the reflection spectrum intensity of unknown plant species in a certain time period to obtain plant species specific index groups. This is compared with the plant species identification index group database of the same time, and the one that matches it is defined as the plant species. At this time, when it is difficult to identify clearly and there is a possibility of specifying multiple plant species, measure the reflection spectrum intensity of the unknown plant species in another time period, A group is obtained, a difference between plant species is obtained by comparing with a plant species identification index group database of different time periods of a plurality of plant species that can be identified, and an unknown plant species can be identified. Furthermore, using these results, it is possible to apply to the mapping of plant species in the measurement region (imaging region).

  In the identification of plant species by remote sensing as described above, when measuring the reflection spectrum intensity of various types of plants, as in the examples, in the measurement wavelength range of 350 nm to 1050 nm, the wavelength band is in increments of 5 nm, such as 141 bands. It is possible to apply not only a sensor capable of measuring a band but also an optical sensor capable of measuring a spectral intensity with hyperspectral measurement, that is, a predetermined observation wavelength region with at least several tens of wavelength bands.

  Further, in this embodiment, the wavelength position to be indexed out of the wavelength positions (spectral intensity extreme value bands) where the extreme values appear as the maximum value and the minimum value that exist in common in the measured spectrum intensity, that is, the singularity. The aforementioned five band wavelength bands are selected as the spectral intensity extreme value bands. However, it is possible to obtain multiple characteristic extreme values that appear in the classification range of a specific plant species, for example, by refining the measurement wavelength bandwidth, etc., and spectral intensity extreme values including them It is also possible to increase six or more singular spectrum intensity extreme bands in the band as necessary. On the contrary, if necessary, it is possible to select four or less singular spectrum intensity extreme value bands.

  From the extreme values in the singular spectrum intensity extreme value band, in the above embodiment, four kinds of indices are selected, and these are used as plant species identification index groups, and are used as plant species identification key indices, and the comparison is performed. is doing. However, the plant species-specific index group need not be limited to these four indices, and further, create another new index, add this new index, or replace it with another index, or It is also possible to simplify the process by excluding certain indexes, and to set such a set of indexes as a new plant species specific index group.

  Further, in the above-described embodiment, the description is made using the example of the result of the measurement at a predetermined time period or a different time period. However, this time period includes one season of four seasons, one month of 12 months per year, Or the season, season, etc. according to the changing state of the plant species, especially the state of leaf change throughout the year (for example, young leaf state, grown leaf state, autumnal leaf state, deciduous state, etc.) It can be timing.

  In the embodiment, an example in which a plant species such as a forest photographed on the ground in a relatively narrow area, such as a park area, is specified. However, this method is applicable to a wide range of aerial images, for example, from an aircraft. Needless to say, the measurement can be applied to the above. Even in the technology that is currently in practical use, for example, when taking an aerial image from an altitude of 1,500 m, the spatial resolution per pixel of a 640 × 480 pixel hyperspectral image is approximately 66 cm □, which enables the creation of highly accurate vegetation maps. It becomes. Of course, if there is a situation where a hyperspectral image can be captured so that there are multiple plant species in one pixel, it is virtually impossible to specify the plant species, so it is important to set the altitude in consideration of them. Needless to say.

  FIG. 10 is a flow chart showing the plant species identification process described in detail so far.

  In FIG. 10, first, a hyperspectral image of the measurement region including the target plant species is photographed (step 1, S1). Next, the reflection spectrum intensity at the image pixel where the target plant species is photographed is acquired as data (ie, wavelength vs. spectrum intensity distribution diagram) (step 2, S2).

  Using this data, a plurality of wavelength bands indicating a plurality of spectral intensity maximum values and minimum values, that is, spectral intensity extreme bands are specified and acquired (step 3, S3). A plurality of singular spectrum intensity extrema bands necessary for an index and singular spectrum intensity extrema in each band are extracted (step 4, S4).

  A plurality of extracted singular spectrum intensity extremums are calculated by devising a plurality of types of plant species indexes that can be calculated by four arithmetic operations, and the plurality of indexes are obtained as a plant species specific index group (step 5, S5). This plant species specific index group is stored in a database (DB) and converted into a DB.

  Or, further, a plant species identification index group of a time period different from that obtained in step 5 of the target plant species is acquired (steps 6 and S6), and this is also stored in the database (DB) and converted into a DB.

  Here, in order to newly specify an unknown plant species (START), the same plant species identification index group as that of the plant species that has been converted to DB is acquired (steps 7 and S7). At this time, the time period to be measured is the same as the time period previously obtained in Steps 1 to 5, and the unknown plant species specific index group is compared with the DB-known known plant species specific index group (Step 8, S8). If the same plant species specific index group is not found, a plant species specific index group of a different time is obtained and compared with the plant species specific index group converted to DB of that time. In this way, an unknown plant species is identified (step 9, S9).

  Furthermore, using this result, the plant species at that location is specified for each captured pixel, and this result is reflected in the entire imaging region to map the identified plant species (step 10). , S10), the plant type map in the entire photographing region is completed (END).

  As described above, the plant species identification method of the present invention performs hyperspectral imaging for a measurement range, performs spectral intensity measurement for each pixel, and further uses imaging results at multiple times. Therefore, it is understood that the plant species can be identified with the captured pixels relatively easily with high accuracy, and further, the vegetation mapping can be performed with high accuracy by performing this result on all the captured pixels. .

  The following additional notes are disclosed regarding the embodiments including the above examples.

(Appendix 1)
Spectra in other time periods including a predetermined time period using an optical sensor device that can observe the spectral intensity of the imaging region where the plant species to be measured are vegetated in a predetermined observation wavelength range with a wavelength band of at least several tens. Taking a picture, and
In the reflection image at the pixel position where the measurement target plant species is photographed in the spectrum image, a plurality of spectral intensity extreme bands indicating a spectral intensity extreme value that is a maximum value or a minimum value of the spectral intensity is acquired. Steps,
Extracting each singular spectral intensity extreme value in a plurality of singular spectral intensity extreme bands out of the spectral intensity extreme bands;
Obtain a plant species specific index group consisting of a plurality of plant species indexes of the measurement target plant species, which is calculated by a calculation process using a plurality of predetermined unique spectrum intensity extreme values among the specific spectrum intensity extreme values. Steps,
The plant species identification index group of the predetermined time of the unknown measurement target plant species is acquired, and compared with the plant species identification index group acquired from the known plant species, the plant species of the unknown measurement target plant species Identifying steps;
If the plant species identification is impossible, the step of identifying the plant species of the unknown measurement target plant species by the respective steps after the step of capturing the spectrum image of the other time period different from the predetermined time period; A plant species identification method comprising:
(Appendix 2)
The plurality of singular spectral intensity extreme bands include at least the following five band wavelength bands of band A, band B, band C, band D, and band E, and the spectral intensity extremes of band A, band C, and band E. The plant species identification method according to appendix 1, wherein the value indicates a maximum value, and the spectrum intensity extreme values of bands B and D indicate minimum values.
(1) A band: 550 ± 5 nm
(2) B band: 680 ± 10 nm
(3) C band: 750 ± 5 nm
(4) D band: 760 ± 5 nm
(5) E band: 770 ± 5 nm
(Appendix 3)
The plant species specific index group is:
When the specific spectrum intensity extreme values of the band A, band B, band C, band D, and band E are designated as intensity A, intensity B, intensity C, intensity D, and intensity E, respectively. The plant species identification method according to supplementary note 2, comprising at least the following index 1, index 2, index 3, and index 4.
(1) Index 1: Ratio of strength C and strength A
That is: C / A
(2) Index 2; ratio of [difference between intensity C and intensity B] and [difference between intensity A and intensity B]
That is: (C−B) / (A−B)
(3) Index 3; [ratio of strength A to strength B] and [ratio of strength A to strength C] That is: (A / B) / (A / C)
(4) Index 4 [ratio of [difference between intensity C and intensity B] and [difference between intensity C and intensity D]], [difference between intensity C and intensity D], and [difference between intensity C and intensity E] Ratio], ie: ((C−B) / (C−D)) / ((C−D) / (C−E))
(Appendix 4)
The plant species identification method according to any one of appendices 1 to 3, wherein the predetermined observation wavelength range is 350 nm to 1050 nm, and the number of wavelength bands is 141.
(Appendix 5)
The plant species identification method according to any one of appendices 1 to 4, wherein the time is the four seasons or a post-transitional stage of the prosperous state of the plant species to be measured.
(Appendix 6)
The method for creating a vegetation map of the imaging region, wherein the plant species in all pixels of the imaging region is specified by the plant species identification method according to attachment 1.

1 Aircraft 2 Artificial Satellite 3 Object 4 Sunlight 5 Camera

Claims (5)

Spectra in other time periods including a predetermined time period using an optical sensor device that can observe the spectral intensity of the imaging region where the plant species to be measured are vegetated in a predetermined observation wavelength range with a wavelength band of at least several tens. Taking a picture, and
In the reflection image at the pixel position where the measurement target plant species is photographed in the spectrum image, a plurality of spectral intensity extreme bands indicating a spectral intensity extreme value that is a maximum value or a minimum value of the spectral intensity is acquired. Steps,
Extracting each singular spectral intensity extreme value in a plurality of singular spectral intensity extreme bands out of the spectral intensity extreme bands;
Obtain a plant species specific index group consisting of a plurality of plant species indexes of the measurement target plant species, which is calculated by a calculation process using a plurality of predetermined unique spectrum intensity extreme values among the specific spectrum intensity extreme values. Steps,
The plant species identification index group of the predetermined time of the unknown measurement target plant species is acquired, and compared with the plant species identification index group acquired from the known plant species, the plant species of the unknown measurement target plant species Identifying steps;
If the plant species identification is impossible, the step of identifying the plant species of the unknown measurement target plant species by the respective steps after the step of capturing the spectrum image of the other time period different from the predetermined time period; ,
A plant species identification method comprising: The plurality of singular spectral intensity extreme bands include at least the following five band wavelength bands of band A, band B, band C, band D, and band E, and the spectral intensity extremes of band A, band C, and band E. The plant species identification method according to claim 1, wherein the value indicates a maximum value, and the spectrum intensity extreme values of band B and band D indicate minimum values.
(1) A band: 550 ± 5 nm
(2) B band: 680 ± 10 nm
(3) C band: 750 ± 5 nm
(4) D band: 760 ± 5 nm
(5) E band: 770 ± 5 nm The plant species specific index group is:
When the specific spectrum intensity extreme values of the band A, band B, band C, band D, and band E are designated as intensity A, intensity B, intensity C, intensity D, and intensity E, respectively. The plant species identification method according to claim 2, comprising at least the following index 1, index 2, index 3, and index 4.
(1) Index 1: Ratio of strength C and strength A
That is: C / A
(2) Index 2; ratio of [difference between intensity C and intensity B] and [difference between intensity A and intensity B]
That is: (C−B) / (A−B)
(3) Index 3; [ratio of strength A to strength B] and [ratio of strength A to strength C] That is: (A / B) / (A / C)
(4) Index 4 [ratio of [difference between intensity C and intensity B] and [difference between intensity C and intensity D]], [difference between intensity C and intensity D], and [difference between intensity C and intensity E] Ratio], ie: ((C−B) / (C−D)) / ((C−D) / (C−E))   The plant species identification method according to any one of claims 1 to 3, wherein the predetermined observation wavelength range is 350 nm to 1050 nm, and the number of wavelength bands is 141.   The plant species identification method according to any one of claims 1 to 4, wherein the time period is a four-season or a post-transitional stage time of a prosperous state of the plant species to be measured.