JP2001357380A - Method for evaluating forest region by picture processing and storage medium with stored program concerning the evaluation - Google Patents

Abstract

PROBLEM TO BE SOLVED: To evaluate a forest region by computer-processing a color picture which is obtained by photographing the forest region from the sky. SOLUTION: The color picture obtained by photographing the forest region from the sky is converted into a black and white picture, the black and white picture is processed to obtain a crown shaped picture, the color picture is analyzed by spectrum and, then, the forest region is evaluated by a spectrum analysis result and the crown shaped picture.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for evaluating a forest area based on a color image taken from the sky of a forest area, and more particularly to a method for processing a computer image of the color image to form a crown of a tree constituting a forest. To evaluate the forest area by grasping the distribution state of trees and the degree of confusion.

[0002] In the present specification, "canopy" refers to a rounded portion composed of upper branches and leaves of each tree constituting a forest, and "canopy shape" refers to a tree directly above a forest tree. It refers to the shape of the canopy viewed from just above).

[0003]

2. Description of the Related Art Conventionally, evaluation of a forest area is performed by setting a relatively small area of a vast forest area as a survey area in advance, and actually going to the forest area to determine the distribution of different trees, the density of tree vegetation, Tree trunk diameters and tree heights were measured, and the entire forest area was evaluated based on field surveys in this survey area. It is extremely difficult to accurately evaluate the entire forest area because the entire forest area is estimated by estimation based on such partial survey results.

The conventional procedure for evaluating a forest area is as follows: (a) A part of the forest area is determined as a survey area by referring to a forest plan map or aerial photograph of a municipal government having a scale of 1/5000. . (b) Use appropriate transportation to get near the study area and walk into the study area. (C) If GPS (global positioning system) is available, the position of the survey point is obtained using GPS. GP
If S is not available, the position of the survey point is specified from the surrounding topographical conditions and the like. (d) Write the survey results on a map and survey notebook.

[0005]

As described above, since the survey area has many mountainous areas where vehicles and the like cannot be used, traveling within the survey area has to rely on walking, and the field survey requires a great deal of cost and expense. There was a problem that labor was required. Furthermore, even if the tree species and age in the survey area are assumed to be the same as those in other areas, the growth conditions of the trees will differ due to differences in the location of the trees. It has been extremely difficult to accurately assess the entire forest area based on this.

Further, in areas where trees are overgrown, GPS
In many cases, it is difficult to receive position information data from artificial satellites, and therefore, it has often been difficult to accurately determine the position of a survey point.

[0007]

SUMMARY OF THE INVENTION Accordingly, the present invention provides an aerial color photograph or a satellite color image obtained by photographing a forest area to a computer, obtains a crown image of the forest area by computer image processing, and obtains the crown shape image and the original crown image. The purpose of this study is to investigate and evaluate tree vegetation in the entire forest area from the color and brightness of trees obtained from color images.

[0008]

According to the present invention, a color image taken from above in a forest area is converted into a black and white image,
Image processing is performed on the black-and-white image to obtain a crown shape image, spectrum analysis of the color image is performed, and a forest area is evaluated using a computer based on the spectrum analysis result and the crown shape image.

[0009] The color image is taken using an airplane or an artificial satellite, and the evaluation of the forest area includes specifying a type of a tree in the forest.

Further, before performing the black-and-white image conversion, there is a case where a process of increasing the resolution of the color image by performing a sub-pixel conversion process on the color image is performed.

Further, the image processing for the black and white image includes:
It includes processing of image smoothing and emphasis on the border of the canopy.

Further, the present invention converts a color image obtained by photographing a forest area from the sky into a black-and-white image, performs image processing on the black-and-white image to obtain a canopy-shaped image, and performs spectral analysis of the color image. The present invention relates to a recording medium recording a program for evaluating a forest area from a spectral analysis result and the canopy shape image.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment according to the present invention will be described below with reference to FIGS.

FIG. 1 is a diagram for explaining a crown and a crown shape. As described above, the "canopy" refers to a rounded portion composed of the upper branches and leaves of each tree constituting the forest, and the "canopy shape" refers to the forest tree directly above (or approximately true). Refers to the shape of the canopy viewed from above. FIG. 1A shows a state in which a forest area is photographed from an aircraft or an artificial satellite, and which part is a crown portion. FIG. 1B shows a state in which the crown is viewed from directly above (or almost directly above).

FIG. 2 is a flowchart for explaining an image processing step (process) according to the present embodiment.
In FIG. 2, an analog color photograph of a forest area taken from an aircraft or an artificial satellite is input to a computer and converted into a digital image (step 10). However, if a color image taken by a digital camera from an aircraft or artificial satellite is input to a computer, step 1
Digitization of an analog color image at 0 can be omitted.

FIG. 3 is a view showing an example of a color image taken from an aircraft and input to a computer as a black and white image. This is because the drawings attached to the specification cannot be displayed in color.

Next, if the digital color image obtained in step 10 does not have a sufficient resolution for extracting the crown shape, a sub-pixel conversion process is performed in step 12 in order to accurately determine the crown shape. Go to add pixels. However, step 12 can be omitted if the digital color image has a sufficient resolution.

At step 14, the color digital image is converted into a multi-valued black and white digital image (monochrome digital image). At the next step 16, the image is smoothed to prepare for the subsequent image processing. The purpose of this smoothing process is to eliminate the problem that a single crown shape is divided into a plurality of parts when determining the crown shape in step 20 described later when there is a fine pixel value variation in the image. It is. This image smoothing process is known as a method using a Gaussian filter.

Next, at step 18, the boundaries between adjacent tree crowns are emphasized. The purpose of this process is to make it easier to determine the crown shape in subsequent image processing. This boundary line enhancement can be performed by, for example, high-pass filter processing. Of course, this processing step can be omitted in an image in which a portion considered to be a boundary of the crown shape is sufficiently clearly shown. However, it is usually considered that the boundary line emphasis processing in step 18 is necessary.

Next, at step 20, the crown shape is determined. A method of obtaining the crown shape will be described with reference to FIG. When the crown is photographed from an aircraft or an artificial satellite, the brightness increases according to the swelling of the crown. That is, the luminance is highest at the vertex of the tree crown, and the luminance change is the smallest at this vertex. The brightness decreases toward the periphery of the canopy, and becomes lowest at the boundary between adjacent canopies. In the embodiment of the present invention, a change in luminance due to the roundness of the canopy is used, and a so-called watershed / watershed is used.
The crown shape is determined by an algorithm (watershed algorithm).

In the watershed algorithm,
A marker is set at the center of the portion where the change in luminance (gradient) is the smallest. That is, the marker is set at or near the vertex of each tree crown (in FIG. 4, a mark * is added to the initial setting portion of the marker). Next, markers initially set at the vertices of the respective crowns are grown in four directions along the change in brightness (gradient) (indicated by a plurality of arrows in FIG. 4). Subsequently, a location where the marker of the adjacent crown hits the marker is determined as a border of the crown. As described above, in the watershed algorithm, the crown shape is extracted by growing a region along the brightness change with the crown top having a small brightness change (gradient) as a starting point.

FIG. 5 shows the above-mentioned watershed image from the color image shown in FIG.
It is a figure showing the crown shape image calculated | required using the algorithm.

Subsequently, in step 22 of FIG. 2, a tree species is determined using the crown shape of the forest area and the wavelength spectrum (spectral analysis result) of the original color image input to the computer in step 10. Specifically, for example, the tree type may be obtained by performing spectral analysis by superimposing the crown shape image of FIG. 5 and the original color image.

An example of this tree species determination will be described with reference to FIG. Fig. 6 shows the light wavelengths (horizontal axis) of four types of trees, Japanese cedar (cedar), hinoki (cypress), konara (konara), and oak (oak), when a certain forest area was photographed in color in November. FIG. 4 is a diagram showing a relationship (spectrum) of luminance (vertical axis) with respect to the luminance.

As shown in FIG. 6, the brightness of the crown portion of the cedar is generally lower than that of other tree species over the entire spectrum from blue to near-infrared rays.
Blue, green, and red visible light have low brightness like cedar,
The luminance is high in the near infrared region. On the other hand, oak has higher overall luminance than cedar and cypress, and oak has overall higher luminance similar to oak, but has lower green and red spectra than oak.

Evergreen trees, deciduous trees, and the like are mixed in a forest area, and the spectrum of each tree in a captured color image varies depending on the season. However, if “seasonal variation data” appearing in color images of such forest areas are accumulated and verified over an appropriate period, an accurate tree can be obtained based on the crown shape and the spectral analysis of the color image obtained from the crown shape. Identification is possible. Furthermore, data fluctuation due to a difference in sunlight at the time of photographing is a problem that can be solved by storing and verifying photographed data.

FIG. 7 shows an example in which the image shown in FIG. 3 is input to a computer, the image processing shown in FIG. 2 is performed to obtain a crown shape image shown in FIG. 5, and the spectrum of each tree shown in FIG. 6 is referred to. It is the figure which evaluated the distribution of four types of trees of the obtained cedar, hinoki, oak, oak.

The evaluation of forest areas described above has been described in terms of tree distribution. However, for trees of the same type, it is also possible to estimate the diameter, height, and age of the trunk based on the size of the crown. Further, the present invention is not limited to trees and can be applied to evaluation of distribution of shrubs.

[0029]

As described above, according to the present invention,
An aerial color photograph or satellite color image of a forest area is input to a computer, and a canopy shape image of the forest area is obtained by computer image processing. The color and brightness of the tree obtained from the canopy shape image and the original color image are obtained. Since the survey / evaluation of tree vegetation in the entire forest area is being conducted, there is a remarkable effect that conventional problems associated with forest area evaluation based on field surveys can be overcome.

[Brief description of the drawings]

FIG. 1 is a view for explaining the crown of a tree in a forest area.

FIG. 2 is a flowchart illustrating computer processing of an image according to the embodiment of the present invention.

FIG. 3 is a diagram showing a color image of a forest area, which is input to a computer in the computer processing of FIG. 2, as a black and white image.

FIG. 4 is an exemplary view for explaining computer processing for obtaining the crown shape of FIG. 2;

FIG. 5 is a diagram showing a crown shape image obtained in the computer processing of FIG. 2;

FIG. 6 is a view showing a spectrum of a tree used for specifying the tree in the computer processing of FIG. 2;

FIG. 7 is a view showing a tree distribution obtained by the computer processing of FIG. 2;

[Explanation of symbols]

10: Computer processing step of inputting a color image (analog) of a forest area from the sky and converting it to a digital image 12: Computer processing step of performing sub-pixel processing 16: Computer processing step of smoothing an image 18: Computer processing step for emphasizing the canopy boundary line 20: Computer processing step for obtaining the canopy shape 22: Computer processing step for obtaining a model based on the canopy shape and spectral analysis

Claims (6)

[Claims] 1. A color image obtained by photographing a forest area from the sky is converted into a black and white image, image processing is performed on the black and white image to obtain a crown shape image, spectral analysis of the color image is performed, and the spectral analysis result A method for evaluating a forest area by image processing, wherein the forest area is evaluated from the canopy shape image using a computer. 2. The method according to claim 1, wherein the color image is taken using an aircraft or a satellite. 3. The method according to claim 1, wherein the evaluation of the forest area includes specifying a type of a tree in the forest. 4. The forest according to claim 1, wherein before the black-and-white image conversion is performed, the color image is subjected to sub-pixel processing to increase the resolution of the color image. Regional evaluation method. 5. The image processing for the black-and-white image includes a process of smoothing the image and emphasizing a border of a tree canopy.
The method for evaluating a forest area by the image processing according to any one of the above. 6. A color image obtained by photographing a forest area from the sky is converted into a black-and-white image, image processing is performed on the black-and-white image to obtain a crown shape image, spectral analysis of the color image is performed, and the spectral analysis result is obtained. A recording medium recording a program for evaluating a forest area from the crown shape image.