JP2011175387A - Ground surface observation method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a ground surface observation method for determining a land cover only by measurement by a laser, and obtain a land surface shape as the need arises. <P>SOLUTION: A decision units are classified into a proper number of units, on the basis of a wavelength-reflection intensity distribution in each decision unit, by use of a proper number of reflection pulses having different wave components reflected from the ground by a multiwavelength laser sweep from a flying body 1 to the ground as the decision unit. A narrow-area attribute associated with a physical property is added to each decision unit, to observe a land cover. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a ground surface observation method, and more particularly to a ground surface observation method for observing the ground surface by laser measurement from above.

  As a method for observing the state of the ground surface, the method described in Patent Document 1 is known. In this conventional example, as the measurement data for specifying the ground surface state, a color orthophoto image generated from a photographed image obtained by photographing the target area from the sky and a laser reflection pulse swept from the sky to the ground were collected. Point cloud data is used.

  The surface condition of the surface is determined by dividing the color orthophoto image by the color intensity of the RGB band into the sun and shade areas. Is done.

JP 2006-252529 A

  However, in the conventional example described above, since the laser sensing portion determines the attribute of the ground surface based on the reflection intensity of the laser light having a single wavelength, the classification accuracy is inferior and can only be used as an auxiliary means for a color image. There is a disadvantage of not.

  The present invention has been made to eliminate the above drawbacks, and provides a ground surface observation method capable of determining land cover only by measurement with a laser and obtaining the land surface shape as required. Objective.

According to the present invention, the object is
Judgment unit group based on the wavelength-reflection intensity distribution in each judgment unit with an appropriate number of reflected pulses (P) having different wavelength components reflected from the ground by multi-wavelength laser sweep from the flying object 1 to the ground as a judgment unit Is classified into an appropriate number,
This is achieved by providing a ground surface observation method for observing the land cover state by assigning the narrow area attribute 2 associated with the physical property to each determination unit.

Covering state of the ground surface, i.e., the presence of vegetation to ground surface, or the estimation of the presence or the like of the artificial feature such as buildings, reflected pulse of laser irradiated from flying body 1 such as an aircraft to the ground (P) is This is done by analyzing. Laser irradiation from the flying object 1 to the ground is performed by sweeping the ground with lasers having a plurality of wavelengths, and a multi-wavelength laser sweep is performed by simultaneously emitting a plurality of single wavelength lights from the flying object 1 or an appropriate number of times. Or irradiating the ground with wavelength-multiplexed lasers such as so-called multi-laser light or hyper laser light that contains multiple wavelengths in a single laser pulse. (Hereinafter referred to as “wavelength multiplexed light sweep”).

  An appropriate number of reflected pulses (P) from the ground are set as a determination unit on condition that reflection points are close to each other and a plurality of wavelengths are provided, and a narrow attribute 2 is given to each determination unit. In the case of wavelength-multiplexed light sweep, the judgment unit is a single pulse in order to satisfy the conditions of multiple wavelengths and proximity reflection with a single pulse, and in the case of single-wavelength light sweep, the proximity is used. A plurality of laser pulses reflected at the positions and having different wavelengths are used as a determination unit.

  The combination of wavelengths used is, for example, the color of the ground surface such as “leaf”, “stem”, “soil”, “rock”, “water”, “artificial structure such as concrete”, or the material, etc. It is desirable to select one that has a large difference in reflection intensity with respect to attributes that are useful for state determination. The determination unit group is based on the relationship between wavelength and reflection intensity, for example, multi-dimensional level slice method, maximum likelihood method, etc. After being classified statistically using an unsupervised classification method such as clustering or unsupervised classification methods such as clustering, the narrow area attribute 2 is given to each determination unit.

  As a result, the narrow area attribute 2 corresponds to the material or the like of the reflection point, and attribute information such as vegetation is added when used together with position information including the height information inherent in the reflection pulse (P). Therefore, it is possible to estimate the land cover state more accurately.

  In addition, using this, for example, when a color corresponding to the narrow area attribute 2 is attached to each determination unit and stereoscopically displayed on a display or the like, the ground surface is in a stipulated state. It becomes possible to confirm more intuitively.

  Furthermore, the determination unit at the time of wavelength-multiplexed light sweep is a single laser pulse, and even in the case of single-wavelength light sweep, if the reflection point density is sufficiently high, the area of the determination unit is small. For this reason, the narrow area attribute 2 given to them indicates an attribute of a relatively small area, and when these are connected and assembled, they can be integrated into a larger group (feature group 3). .

  That is, for example, it is possible to define a feature group 3 such as “tree” or “vegetation” in an area where the ratio of the narrow area attribute 2 “leaf” and “trunk” is large. In the region where the ratio of “rock” is large, “bare ground” or the like can be defined.

  Such integration into the feature group 3 makes it possible to know in more detail the arrangement of features that are more realistic, and after integration into the feature group 3, the attributes of the region can be assumed. Therefore, it is possible to determine whether the determination unit is unclassified due to misclassification at the time of the above-described classification operation of the narrow area attribute 2, or by shadow, mixel, or the like based on the group concept.

  It is possible to further observe the ground surface shape by using the generated product. The ground surface shape observation results can be used as the final process of DTM (Digital Terrain Model) generation or as a pre-processing of the final filtering process to remove extra reflection pulses (P) from other than the ground surface. Is done.

  When observing the shape of the ground surface, the feature group 3 is classified according to whether or not there is a possibility of including a reflection pulse (P) from the ground surface. For example, when the feature group 3 is integrated as “vegetation”, “bare”, and “artificial feature” or “water”, the artificial feature and water are Therefore, even if “soil” or the like is added to the narrow area attribute 2, the remaining pulse group 3 such as “vegetation” is deleted. The pulse group is extracted as a final filtering target.

  In this state, since a considerable number of pulse groups that are clearly unrelated to the generation of the ground surface model can be screened, an effective filtering effect can be expected, but only low elevation data is extracted from the extracted pulse groups. When extracted, for example, since the reflected pulse (P) reflected from the top of the tree in the vegetation area can be removed, a higher filtering effect can be expected, and as a result, accurate and efficient ground surface shape can be obtained. become.

  For the low altitude data, the pulse with the lowest altitude value of the evaluation unit in the case of single wavelength light sweep or the pulse with the highest reflection order can be adopted, such as “soil” in the case of wavelength multiplexed laser sweep. A reflected pulse (P) to which the narrow area attribute 2 is given, or a reflected pulse (P) to which the narrow area attribute 2 is not given due to the influence of a mixel or the like can be used as necessary.

In addition, as a ground surface observation method only for the purpose of observation of the ground surface shape,
A ground surface observation method for narrowing down a reflection pulse (P) necessary for terrain extraction from a reflection pulse point group acquired by an aerial laser survey as a ground surface extraction element, and observing a ground surface shape from the ground surface extraction element,
Judgment unit group based on the wavelength-reflection intensity distribution in each judgment unit with an appropriate number of reflected pulses (P) having different wavelength components reflected from the ground by multi-wavelength laser sweep from the flying object 1 to the ground as a judgment unit After classifying a suitable number including at least ground surface components and vegetation components,
It is also possible to configure to extract the pulse data in the ground surface area mainly consisting of ground surface attributes and the vegetation area mainly based on vegetation attributes as ground surface extraction elements by integrating judgment units. Wavelength multiplexed laser light can be used as the sweep laser.

Furthermore, ground surface observation using wavelength multiplexed laser light
The ground surface is swept from the flying object 1 by wavelength multiplexed laser light that multiplexes the wavelength components that can be distinguished from the non-ground surface components such as vegetation and artificial features such as soil and rock by wavelength-reflection intensity distribution. And
This can also be achieved by a ground surface observation method that obtains the ground surface shape from the positional information of the reflected pulse (P) extracted as a ground surface component from the wavelength-reflection intensity distribution.

  According to the present invention, it is possible to determine the state of the land surface or the land surface shape only by measurement with a laser.

It is explanatory drawing which shows a laser measurement, (a) is a figure which shows the laser sweep from the flying body 1, (b) is a figure which shows a reflected pulse. It is a figure which shows wavelength-reflection intensity distribution, (a) is a figure which shows the difference of the reflection intensity with respect to the wavelength by a material, (b) is a figure which shows the wavelength-reflection intensity distribution of a plant, (c) is the wavelength-reflection of soil The figure which shows intensity distribution, (d) is a figure which shows the wavelength-reflection intensity distribution of water. It is a figure which shows the procedure of this invention. It is a figure which shows the Example of this invention, (a) is a figure which shows the state classified into the feature group 3, (b) is a figure which shows the state filtered by the feature group 3, (c) is a surface element. It is a figure which shows the state extracted.

  First, acquisition of laser data used in the present invention will be described with reference to FIG. In FIG. 1A, reference numeral 1 denotes a flying object such as an aircraft or a helicopter that flies along a previously planned flight course (C). The flying object 1 is equipped with a laser distance measuring device 4 and related devices (not shown) such as a GPS for measuring the position and photographing posture of the aircraft, and an IMU (Inertial Measurement Unit).

  The laser distance measuring device 4 includes a laser oscillation device 4a and a scan mirror 4b, and the laser light output from the laser oscillation device 4a is swung at right angles to the flight direction by the scan mirror 4b and directed toward the ground. Irradiated.

  The laser oscillation device 4a is a so-called multilaser or hyper laser device that outputs a wavelength-multiplexed laser pulse in which a plurality of different wavelength components are multiplexed in a single pulse. In addition, four types of wavelength components of blue region (B), green region (G), red region (R), and near infrared region (IR) are multiplexed.

  Since the beam diameter of the laser pulse emitted from the flying object 1 increases as it goes to the ground, there is a reflection target in the beam passage path depending on the density of the reflection target, and the reflected pulse (P) from the target is Even after being generated, the laser pulse can travel further straight from the gap where the beam path is not obstructed to generate a reflected pulse (P) from another object.

FIG. 1B illustrates this state. The laser pulse emitted from the flying object 1 is first reflected at the “leaf” portion of the tree and becomes the first reflected pulse (P ¹ ) (first pulse). Then, reflection (P ² ) (intermediate pulse: second pulse) is performed at the “stem” portion, and finally a reflected pulse (P ^L ) (last pulse) from the ground surface is generated. The laser distance measuring device can identify the reflection point of each reflected pulse (P) by capturing the plurality of reflected pulses (P).

  In addition to position-related data, wavelength-reflection intensity distribution data is given to each reflected pulse (P) (evaluation unit) received from the ground as described above. The position-related data is calculated based on the time required from irradiation to reception of the reflected pulse (P), or the position of the flying object 1 at the time of laser irradiation indicated by the GPS / IMU data, tilt information, and the angle of the scan mirror 4b. Consists of position information including the height information. The wavelength-reflection intensity distribution data is obtained by dispersing the reflection pulse (P) with an appropriate spectroscopic means, and is composed of reflection intensity for each of the four types of wavelengths included in the reflection pulse (P) as shown in FIG. Is done.

Land cover estimation and reflection pulse (P) filtering are performed using the reflection pulse data group obtained as described above. First, as shown in FIG. 3, each reflection pulse (P) data is obtained. Against
P _n (X _n , Y _n , Z _n )
Where n is the data number, X is the latitude, Y is the longitude, and Z is the height data
P _n (B _n , G _n , R _n , IR _n )
Where B is the reflection intensity in the blue region, G is the reflection intensity in the green region, R is the reflection intensity in the red region, and IR is the spectral distribution data consisting of the reflection intensity in the near infrared region (P) A data structure is assigned (step S1).

  Next, spectral distribution data is classified from the reflected pulse data group (step S2). Various known methods can be used for classification. For example, in the case of supervised classification, first, after setting a classification class, spectral distribution data that is known in advance to belong to the classification class is used as the teacher data. Then, the processing is completed by classifying the teacher data and the statistical distribution data of the acquired reflected pulse data with statistical similarity.

  In this example, “Leaf”, “Stem”, “Soil”, “Rock”, “Water”, “Artificial structure such as concrete” (not displayed), etc. are set as the classification class. For comparison, a multidimensional level slice method, a maximum likelihood method, or the like can be used.

  As shown in FIG. 2 (a), it is known that the reflection intensities of wavelengths in the RGB and IR regions differ depending on the reflection target. For example, plants strongly reflect the wavelengths in the IR region, and soil has a wavelength of The reflection intensity decreases as the length increases, and water does not return IR light to the laser distance measuring device 4 (total reflection). Accordingly, for the wavelengths in the RGB and IR regions, the global reflection intensity in the region where the plants are dense is shown in FIG. 2 (b), the soil is shown in FIG. 2 (c), and the water is shown in FIG. 2 (d). As the teacher data, it is possible to use data obtained by experimentally determining the distribution when subdividing these into “leaves”, “stems”, and the like.

  Classification items (narrow region attribute 2) such as “leaf” and “stem” are assigned to each reflection point by classification, and further, a reflected pulse (P) in which no significant similarity with teacher data was recognized during the classification operation. ) Is given an “unseparated” attribute.

  Next, the reflection pulse (P) to which the narrow area attribute 2 other than “unseparated” is added is integrated to give the feature group 3 attribute (step S3). The integration of the reflected pulse (P) is, for example, a unit cube set in a three-dimensional space composed of reflected pulse (P) groups, or a unit obtained by projecting a reflected pulse (P) group spreading in three dimensions into a two-dimensional space. After the attribute of the feature group 3 set separately based on the distribution of the narrow attribute 2 is given to the mesh, the same group attribute is integrated on condition that the feature group 3 attribute matches and the position is adjacent and close. Can be realized.

  In this example, as shown in FIG. 3, the feature group 3 is “grassland” having “leaves”, “trunks”, and “soil” as elements and “trees”, “leaves”, and “soil” as elements. , “Bare” with “soil” as an element, “rock” with “rock” as an element, “water” with “water” as an element, and “artificial features such as concrete” “Artificial features” (not shown). FIG. 4A shows a state in which the feature group 3 attribute is assigned to the area.

  In the above, the case where the integration into the feature group 3 is performed based on the distribution state of the narrow area attribute 2 in the area of an appropriate area has been shown. It is also possible to define the “tree” group from the set of “leaves” and “trunk” from the set state, and then derive the “tree” from the set of “tree” groups. .

  Each feature group 3 obtained as described above can be color-coded and displayed on a display or the like in a three-dimensional display or in a two-dimensional space, and the land cover state can be confirmed accurately. Can do.

  Further, when extracting the ground surface data using the land cover information, first, of the above feature groups 3, those having group attributes that may include the ground surface data are selected ( Step S4). As shown in Fig. 3, the selected group has gaps such as "leaves" and "trunks" at the time of integration in addition to the ground surface area consisting of "bare ground" and "rocky ground" that clearly constitute the ground surface. Also included are vegetation areas of “woody land” and “grassland” in which reflected pulses (P) due to pulses that have passed through and reached the ground surface may be included in higher-order pulses. On the other hand, since the ground surface shape cannot be extracted from the reflection pulse (P) from the “water part” or “artificial feature part”, these are excluded. FIG. 4B shows the filtering result excluding “water” and the like.

  After filtering the reflected pulse group by the feature group 3 as described above and extracting the reflected pulse (P) group in the selected group as the ground surface extraction element (step S5), from these extracted elements from the ground surface The reflected pulse (P) is extracted (step S6).

  Extraction of the reflection pulse (P) from the ground surface is performed by extracting those whose narrow area attribute 2 is “soil” or “rock” from the reflection pulse (P) extracted as the ground surface extraction element. Further, in addition to the adjacent and close positions in the two-dimensional space, the reflected pulse (P) classified as unclassified with a small difference from the neighboring pulse of the Z value (height information) can be included.

  FIG. 4C shows the distribution of the reflected pulse (P) extracted by the above operation. Since each reflection pulse (P) displayed on the XY plane has a Z value, for example, a mesh is set on the XY plane to give a representative position, and a TIN display or the like is performed. By doing so, a DTM can be generated.

In the above, the case where the laser point group by the wavelength multiplexed laser pulse is used is shown. However, in the case of the single wavelength light sweep, the reflection pulse data structure in step S1 is the same for each reflection pulse (P). After the position data and spectral distribution data for each evaluation unit are defined, and after each step is executed for the evaluation unit, in step S6, the ground surface element is determined based on the Z value of each reflected pulse (P). Extracted.
Further, since the height information of the “arboretum” can be obtained, the state of the forest can be grasped, and for example, it can be used for emissions trading such as carbon offset.

1 Aircraft 2 Narrow attribute 3 Feature group P Reflected pulse

Claims (7)

Using the appropriate number of reflected pulses with different wavelength components reflected from the ground by multi-wavelength laser sweep from the flying object to the ground as the judgment unit, the number of judgment units is set to the appropriate number based on the wavelength-reflection intensity distribution in each judgment unit. Classify and
A ground surface observation method for observing the land cover state by assigning a narrow area attribute associated with physical properties to each judgment unit. Using each reflected pulse from the ground of the wavelength multiplexed laser pulse group irradiated from the flying object to the ground as a judgment unit, classify the judgment unit group into an appropriate number based on the spectral distribution in each judgment unit,
A ground surface observation method for observing the land cover state by assigning a narrow area attribute associated with physical properties to each judgment unit.   The ground surface observation method according to claim 1 or 2, wherein the land cover state is observed by integrating an appropriate number of the determination units into a feature group predetermined based on a set state of the narrow area attribute. Among the feature groups according to claim 3, low altitude data in the feature group that may include a reflection pulse from the ground surface is extracted as a terrain extraction element,
A ground surface observation method for observing the ground surface shape from the topographic extraction element. A ground surface observation method for narrowing down reflection pulses necessary for terrain extraction from reflected pulse point groups acquired by aerial laser surveying as ground surface extraction elements, and observing the ground surface shape from ground surface extraction elements,
Using the appropriate number of reflected pulses having different wavelength components reflected from the ground as a result of multi-wavelength laser sweep from the flying object to the ground, the judgment unit group is at least the ground surface based on the wavelength-reflection intensity distribution in each judgment unit. After categorizing the components and vegetation components into appropriate numbers,
A ground surface observation method for extracting, as ground surface extraction elements, ground surface areas mainly composed of ground surface attributes integrated with judgment units and pulse data in vegetation areas mainly composed of vegetation attributes. The sweep laser from the aircraft is performed by wavelength multiplexed laser light, and the classification attributes include ground surface components such as soil and rock,
The ground surface observation method according to claim 5, wherein the extraction of the ground surface extraction element from the ground surface area and the vegetation area is performed on the ground surface constituent elements and pulse data that has not been separated at the time of classification. . The ground surface is swept from the flying object by wavelength multiplexed laser light that multiplexes the wavelength components that can be separated from the non-ground surface components such as vegetation and artificial features such as soil and rock by wavelength-reflection intensity distribution. ,
A ground surface observation method that obtains the ground surface shape from the position information of reflected pulses extracted as ground surface components from the wavelength-reflection intensity distribution.

Images