JPH11185009A - Image data storing and identifying device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To sort/identify an object out of satellite images acquired on a single wavelength band by finding the uniformity data of pixels corresponding to the digital data of images, sorting and identifying the object from the relative relation of uniformity data and digital image data. SOLUTION: An image 24 acquired by a satellite is inputted to an image data memory 25. A uniformity data extracting means 23 prepares uniformity data 26 from digital data 25 of images acquired by the satellite outputted from the memory 22 and outputs them to a correlative arithmetic means 6. A linearity data extracting means 31 prepares linearity data 32 from the digital data 25 of images and outputs them to the correlative arithmetic means 6. The correlative arithmetic means 6 performs sorting and identifying for each pixel from the correlative relation of the digital image data 25 and the uniformity data 26. Image display data 12 showing the sorted and identified result are displayed on a display device 7.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for classifying and identifying an object such as vegetation using an image obtained by a flying object such as a satellite. Here, for convenience of explanation, an apparatus for classifying and identifying an object such as vegetation using an image acquired by a satellite will be described below.

[0002]

2. Description of the Related Art Wide-area surface observation using satellite images is important for estimating crop yields, monitoring the destruction of the environment such as forests, and understanding the damage status of disasters such as floods. In this case, it is necessary to acquire a large number of images in different wavelength bands in order to perform classification and identification of a target area with high accuracy. In addition, since an optical image is used as a satellite image, weather conditions such as cloud cover in a target area are very important.

[0005] FIG. 5 is a diagram for explaining the positional relationship when observing the earth's surface with a satellite. The satellite 1 usually orbits a polar orbit passing near the north and south poles to observe almost the entire surface of the earth. In order to classify objects such as vegetation included in the image data from the acquired images, a plurality of images are usually acquired simultaneously in different wavelength bands. However, here, for the sake of explanation, two images are acquired using two types of wavelength bands. The case in which is obtained will be described below.

FIG. 6 shows the configuration of this type of apparatus which is generally tried for the purpose of inputting two satellite images acquired in two wavelength bands and classifying and identifying objects such as vegetation in the acquired area. 4, reference numeral 4 denotes a first wavelength band image data memory, 5 denotes a second wavelength band image data memory, 6 denotes a correlation calculating means, 7 denotes a display device, 8 denotes an image acquired in the first wavelength band, 9 Are the images acquired in the second wavelength band, 1
0 is digital data of an image acquired in the first wavelength band, 11 is digital data of an image acquired in the second wavelength band, and 12 is image display data indicating a classification identification result. In FIG. 6, an image 8 acquired in the first wavelength band
And the image 9 acquired by the second wavelength band is input to the first wavelength band image data memory 4 and the second wavelength band image data memory 5, respectively. Classification and identification for each pixel are performed based on the correlation between the digital data 10 and 11 described above. The classification and identification results thus obtained are output to the display device 7 as image display data 12 indicating the classification and identification results.

FIG. 7 is a diagram showing an example of a correlation calculating means in a conventional image data classification / identification apparatus. The horizontal axis 13 is the image brightness in the first wavelength band, and the vertical axis is 14.
Indicates the luminance of the image in the second wavelength band, 15 indicates the first classification result, 16 indicates the second classification result, and 17 indicates the third classification result. When the luminance of the image in the first wavelength band is set on the horizontal axis 13 and the luminance of the image in the second wavelength band is set on the vertical axis 14 and the distribution is examined for each pixel, a correlation diagram as shown in FIG. 7 is obtained. Objects of the same kind, such as vegetation, are distributed on substantially the same coordinate position, and are classified as objects of the same kind, such as vegetation. Further, from the distribution positions on the coordinates shown in FIG. 7, for example, the first classification result 15 is identified as sea, the second classification result 16 is urban area, and the third classification result 17 is identified as vegetation. Here, the correlation calculation of two types of images has been described for the sake of convenience, but the correlation calculation is performed in the same manner in the case of more images.

FIG. 8 is a view showing an example of a display in a conventional image data classification / identification apparatus.
9 is the identification result indicating the sea, 20 is the identification result indicating the city area,
21 is an identification result indicating vegetation. In general, identification results can be clearly shown by displaying each identification result in a different color.

[0007]

The conventional classification of objects such as vegetation using satellite images has a function of performing classification and identification based on the correlation between a plurality of images acquired in different wavelength bands as described above. The main thing. However, the amount of image data required at this time increases in proportion to the number of wavelength bands to be used, and the number of pixels per pixel tends to rapidly increase due to the recent increase in resolution of observation sensors. . This increase in the amount of image data also affects the increase in communication time when transmitting image data from the satellite to the ground and the processing speed of the data recording equipment on the ground. In addition, there is a drawback that a satellite for acquiring an image must have a plurality of sensors corresponding to those wavelength bands, and in particular, when performing classification and identification based on radar images, providing a plurality of sensor wavelength bands is difficult. Since it is necessary to install a plurality of transmission / reception devices and mounting on a satellite becomes physically difficult, a satellite having only a single-wavelength radar is mainly used at present. For this reason, there has been a drawback that the conventional technique is not suitable for classification and identification. However, radar images are expected to be used in recent years because they can be obtained continuously without depending on weather conditions such as cloud cover in the target area at the time of acquisition.

The present invention has been made to solve such a problem, and an object of the present invention is to provide an image data classification and identification device capable of classifying and identifying an object from a satellite image acquired in a single wavelength band.

[0009]

According to a first aspect of the present invention, there is provided an image data classification / identification apparatus for generating uniformity data for each pixel by texture analysis processing from an image data memory holding an image acquired by a satellite. Means for classifying and identifying objects such as vegetation included in the image from the correlation of the uniformity data with the image obtained by the satellite,
A means for displaying the classification and identification results of these objects such as vegetation is provided.

According to a second aspect of the present invention, there is provided an image data classification and identification apparatus.
Means for creating linearity data for each pixel by a texture analysis process from an image data memory holding an image acquired by a satellite, and vegetation included in the image based on a correlation between the linearity data and the image acquired by the satellite. And means for classifying and identifying objects such as vegetation and displaying the results of classification and identification of these objects such as vegetation.

According to a third aspect of the present invention, there is provided an image data classification and identification apparatus.
Means for creating the uniformity data shown in the first invention from an image data memory holding an image acquired by a satellite, and creating the linearity data shown in the second invention from the first image memory Means for classifying and identifying objects such as vegetation included in the image based on the correlation between the uniformity data, the linearity data and the image acquired by the satellite, and an object such as vegetation Means for displaying the classification and identification results.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 FIG. 1 shows the configuration of a first embodiment of an image data classification / identification apparatus according to the present invention. Reference numeral 22 denotes an image data memory, 23 denotes uniformity data extracting means, 24 denotes an image acquired by a satellite, and 25 denotes a satellite. Digital data 26 of the acquired image is uniformity data. Image 2 acquired by satellite in FIG.
4 is input to the image data memory 22. Then, the uniformity data extracting means 23 outputs the digital data 25 of the image acquired by the satellite output from the image data memory 22.
More uniformity data 26 is created and output to the correlation calculator 6. The correlation calculating means 6 performs classification and identification for each pixel from the correlation between the digital data 25 and the uniformity data 26 of the image acquired by the satellite. The classification and identification results thus obtained are output to the display device 7 as image display data 12 indicating the classification and identification results, and displayed on the display device 7.

An example of the uniformity data extracting means 23 will be described below. FIG. 2A is a diagram for explaining an arrangement of pixels of digital data 25 of an image acquired by a satellite, in which 27 is an arbitrary pixel, 28 is a direction indicating west → east direction, and 29 is north →
This is the azimuth indicating the south direction. For the sake of simplicity, the number of pixels is N ×
As shown in FIG. 2A, the upper left (northwest) corner is coordinates (0, 0), the upper right (northeast) corner is coordinates (N, 0), and the lower left (southwest) corner is coordinates ( 0, N), and the lower right (southeast) corner is represented by coordinates (N, N). Here, the coordinates of an arbitrary pixel 27 are set to (x, y), and an n × n region centered on the coordinates is considered. However, n is an odd number, and here, n = 3 for simplicity. FIG. 2 (b)
Is a diagram showing a 3 × 3 area centered on an arbitrary pixel 27, and each pixel in the target area 30 is given k-level luminance corresponding to the pixel. Here, for the sake of simplicity, the description will be made assuming that the luminance is four gradations of 0 to 3. In the target area 30, the probability P that the luminance of a point at a distance d in an arbitrary direction from the coordinates of the luminance i of the image becomes j is P
A co-occurrence matrix having (i, j) as an element is obtained. Here, for simplicity, it is assumed that an arbitrary direction is the horizontal direction in the target region 30 and the distance d = 1. For example, target area 3
At 0, there are two combinations in which the luminance of a point separated by a distance d = 1 coordinate in the horizontal direction from the coordinate of luminance 0 is 1; therefore, the probability P (0,1) is 2. When this is obtained for all combinations, the co-occurrence matrix P (i, j) is as shown in Equation 1. The co-occurrence matrix P found here
F obtained by substituting (i, j) into Equation 2 becomes the uniformity data at an arbitrary pixel 27 shown in FIG. The above is a method of obtaining uniformity data at an arbitrary pixel 27 having coordinates (x, y). This calculation is performed on all the pixels of the digital data 25 of the image shown in FIG. Thus, uniformity data of the number of pixels N × N with respect to the digital data 25 of the image shown in FIG. 2A is obtained.

[0014]

(Equation 1)

[0015]

(Equation 2)

Embodiment 2 FIG. 3 shows the configuration of an embodiment 2 of the image data classification and identification device according to the present invention, wherein 31 is linearity data extracting means, and 32 is linearity data. In FIG. 3, the image 24 acquired by the satellite is
The data is input to the image data memory 22. Then, the linearity data extracting means 31 creates linearity data 32 from the digital data 25 of the image obtained by the satellite output from the image data memory 22 and outputs it to the correlation calculating means 6. The correlation calculation means 6 performs classification and identification for each pixel based on the correlation between the digital data 25 and the linearity data 32 of the image acquired by the satellite. The classification and identification results thus obtained are output to the display device 7 as image display data 12 indicating the classification and identification results, and displayed on the display device 7.

An example of the linearity data extracting means 31 will be described below. The number of pixels of the digital data 25 of the image acquired by the satellite is defined as an N × N square matrix, and arbitrary pixels that are coordinates (x, y) in the digital data 25 of the image are described in the first embodiment. The co-occurrence matrix P
(I, j) is obtained. The co-occurrence matrix P found here
By substituting (i, j) into Equations 3 and 4, respectively, Px
(I), Py (j) are obtained, and then Px (i), Py
By substituting (j) into Equations 5 and 6, μx and μ
y is required. Further, Px (i) and μx are converted into Equation 7,
By substituting Py (j) and μy into Equation 8, σx and σy are obtained. P (i, j) obtained above,
g obtained by substituting μx, μy, σx, and σy into Equation 9 is the coordinates (x,
y) is linearity data at an arbitrary pixel. The above is a method of obtaining linearity data at an arbitrary pixel having coordinates (x, y). By performing this calculation for all the pixels of the digital data 25 of the image, the digital data 25 of the image is obtained. , The linearity data of the number of pixels N × N is obtained.

[0018]

(Equation 3)

[0019]

(Equation 4)

[0020]

(Equation 5)

[0021]

(Equation 6)

[0022]

(Equation 7)

[0023]

(Equation 8)

[0024]

(Equation 9)

Embodiment 3 FIG. 4 shows the configuration of the third embodiment of the image data classification and identification device according to the present invention. In FIG. 4, an image 24 obtained by a satellite is input to an image data memory 22. Then, the uniformity data extracting means 23 creates uniformity data 26 from the digital data 25 of the image obtained by the satellite output from the image data memory 22 and outputs it to the correlation calculating means 6. Further, the linearity data extracting means 31 includes the image data memory 22
The linearity data 32 is created from the digital data 25 of the image obtained by the satellite output from the satellite and output to the correlation calculating means 6. The correlation calculation means 6 performs classification and identification for each pixel based on the correlation between the digital data 25, the uniformity data 26, and the linearity data 32 of the image acquired by the satellite. The classification and identification results thus obtained are output to the display device 7 as image display data 12 indicating the classification and identification results, and displayed on the display device 7.

In the first to third embodiments, an apparatus for classifying and identifying an object such as vegetation using an image acquired by a satellite has been described. However, the present invention relates to a flying object such as a satellite, an aircraft, an airship, and the like. It is needless to say that the present invention can also be applied to an apparatus for classifying and identifying an object such as vegetation using the image acquired in step (1).

[0027]

According to the first aspect of the present invention, from a satellite image including a natural object such as the sea and vegetation obtained in a single wavelength band,
The natural object can be easily detected. Especially,
Since an image acquired in a single wavelength band is used as a satellite image, the required amount of image data is significantly reduced as compared with the conventional method. In addition, conventional methods make it easier to apply radar images that were not suitable for classification and identification due to physical reasons, and have the effect of enabling continuous classification and identification independent of weather conditions such as cloudiness. is there.

According to the second aspect, an area including the artificial building can be easily detected from a satellite image including many artificial buildings such as an urban area and a bridge acquired in a single wavelength band. In particular, since an image acquired in a single wavelength band is used as a satellite image, there is an effect that the required data amount of the image is significantly reduced as compared with the conventional method. In addition, conventional methods make it easier to apply radar images that were not suitable for classification and identification due to physical reasons, and have the effect of enabling continuous classification and identification independent of weather conditions such as cloudiness. is there.

According to the third aspect of the present invention, the natural object and the artificial building are obtained from a satellite image including a large number of artificial objects such as the sea and vegetation and an artificial building such as an urban area and a bridge acquired in a single wavelength band. An area containing many things can be easily detected. In particular, since an image acquired in a single wavelength band is used as a satellite image, there is an effect that the required data amount of the image is significantly reduced as compared with the conventional method. In addition, conventional methods make it easier to apply radar images that were not suitable for classification and identification due to physical reasons, and have the effect of enabling continuous classification and identification independent of weather conditions such as cloudiness. is there.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a first embodiment of an image data classification and identification device according to the present invention.

FIG. 2 is a diagram illustrating a pixel arrangement of digital data of an image acquired by a satellite according to the first embodiment of the image data classification and identification device according to the present invention, and a 3 × 3 pixel having an arbitrary coordinate centered on the image. It is a figure showing an area.

FIG. 3 is a diagram showing a configuration of an image data classification and identification device according to a second embodiment of the present invention;

FIG. 4 is a diagram showing a configuration of an image data classification and identification device according to a third embodiment of the present invention;

FIG. 5 is a diagram illustrating an arrangement relationship at the time of ground surface observation by a satellite.

FIG. 6 is a diagram showing a configuration of a conventional image data classification and identification device.

FIG. 7 is a diagram illustrating an example of a correlation operation unit in a conventional image data classification and identification device.

FIG. 8 is a diagram showing an example of display in a conventional image data classification and identification device.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 satellite, 2 earth's surface, 3 field of view, 4 first wavelength band image data memory, 5 second wavelength band image data memory, 6 correlation operation means, 7 display device, 8 images acquired by 1st wavelength band , 9 an image acquired in the second wavelength band, 10 digital data of an image acquired in the first wavelength band, 11 digital data of an image acquired in the second wavelength band, 12 image display data indicating a classification identification result, 13 image brightness by first wavelength band, 14 second
Brightness of the image according to the wavelength band of 15, 15 first classification result, 1
6 Second classification result, 17 Third classification result, 18 display screen, 19 Sea identification result, 20 City identification result, 21 Vegetation identification result, 22 Image data memory, 23 Uniformity data extraction Means, image acquired by 24 satellites, digital data of image acquired by 25 satellites, 26 uniformity data, 27 arbitrary pixels, 2
8 Direction indicating west → east direction, 29 direction indicating north → south direction, 30 target area, 31 straight line data extracting means, 32 straight line data.

Claims (3)

[Claims] 1. An apparatus for classifying and identifying an object such as a vegetation included in an image using digital data of an image obtained by a flying object such as a satellite. An image data classification and identification apparatus, comprising: means for obtaining state data; and means for classifying and identifying the object based on a correlation between the uniformity data and the digital data of the image. 2. An apparatus for classifying and identifying an object such as vegetation included in an image using digital data of an image obtained by a flying object such as a satellite, wherein a straight line of pixels corresponding to the digital data of the image is used. An image data classification and identification apparatus, comprising: means for determining sex data; and means for classifying and identifying the object based on a correlation between the linearity data and the digital data of the image. 3. An apparatus for classifying and identifying an object such as a vegetation included in an image using digital data of an image obtained by a flying object such as a satellite. Means for obtaining uniformity data; means for obtaining linearity data of pixels with respect to the digital data of the image; and classification of the object based on a correlation between the uniformity data and the linearity data and the digital data of the image. And an identification means.