JP2013218227A - Digital elevation model evaluation device and digital elevation model evaluation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To evaluate the accuracy of an automatically generated digital elevation model without using a digital elevation model with true values or approximately true values or a satellite photograph pair captured from a site.SOLUTION: A digital elevation model evaluation device includes: digital elevation data down-sampling means 221 which down-samples digital elevation data being data of a digital elevation model; digital elevation data up-sampling means 222 which up-samples the down-sampled digital elevation data; digital elevation data gradation reduction means 223 which reduces the number of gradation of the digital elevation data; digital elevation data error calculation means 240 calculates an evaluation error from the digital elevation data being data of the digital elevation model and the up-sampled digital elevation data or the gradation-reduced digital elevation data; and appropriate resolution calculation means 260 which calculates an appropriate resolution from the calculated evaluation error.

Description

  The present invention relates to evaluation of a digital elevation model for a digital elevation model generated by photogrammetry from a photograph taken from an artificial satellite or an aircraft.

  Conventionally, when a digital elevation model at a certain point is generated by various different algorithms, in order to decide whether it is good or bad, it is evaluated using the true value of the digital elevation model at that point, or a point is photographed 2 One of the evaluations using one satellite photograph (satellite pair) is necessary.

When the true value of a digital elevation model (or a digital elevation model with similar accuracy) exists, it was evaluated by directly comparing with the true value.
Also, if there are two satellite photographs of a point taken from different directions, first use the digital elevation model you want to evaluate and orthographically project each of the two satellite photographs. The closer the model is to the true value, the more similar the characteristics of two orthographically projected satellite photographs, and human evaluation of the degree of overlap between the two orthographically projected satellite photographs. I was evaluating a digital elevation model. (See Patent Document 1).

JP 2011-221231 A

However, if neither the true value of the digital elevation model nor the satellite photograph pair that captured the point is available, these evaluations could not be made by the prior art.
Even if a pair of satellite photographs is available, the two satellite photographs are taken from different directions, so the color of the captured images differs between the two. For this reason, when the method described in Patent Document 1 is applied, even if the digital elevation model is correct, the two do not completely match, and it is difficult to automatically determine the degree of overlap. It was necessary to evaluate with.

  The present invention has been made to solve the above-described problems, and is automatically generated without using a true value or a numerical altitude model conforming to the true value or a pair of satellite photographs of the point. The purpose is to evaluate the accuracy of the digital elevation model.

  The present invention relates to a digital elevation data downsampling means for downsampling digital elevation data that is data of a digital elevation model, a digital elevation data upsampling means for upsampling the downsampled numerical elevation data, and a gradation of the digital elevation data. Digital elevation data that reduces the number of digital elevations, digital elevation data that is digital elevation model data, up-sampled digital elevation data, and digital elevation data that calculates evaluation errors An inter-data error calculating means and an appropriate resolution calculating means for calculating an appropriate resolution from the calculated evaluation error are provided.

  The present invention relates to a digital elevation data downsampling means for downsampling digital elevation data that is data of a digital elevation model, a digital elevation data upsampling means for upsampling the downsampled numerical elevation data, and a gradation of the digital elevation data. Digital elevation data that reduces the number of digital elevations, digital elevation data that is digital elevation model data, up-sampled digital elevation data, and digital elevation data that calculates evaluation errors Since the inter-data error calculating means and the appropriate resolution calculating means for calculating the appropriate resolution from the calculated evaluation error are provided, it is possible to evaluate the accuracy of the digital elevation model on the scale of appropriate resolution.

1 is a configuration diagram of a numerical elevation model evaluation apparatus in Embodiment 1. FIG. FIG. 3 is a configuration diagram of an appropriate resolution calculation unit 200 in the first embodiment. 5 is a flowchart showing the operation of the digital elevation model evaluation apparatus 10 according to the first embodiment. FIG. 4 shows an example of a digital elevation model (DEMa, DEMb) in the first embodiment. The figure which shows the change of the numerical elevation data after up-sampling after unsampling in Embodiment 1. FIG. FIG. 6 is a diagram showing DEMa1, DEMa2, and DEMa3 generated by the three-time loop of steps S25 to S35 in Embodiment 1, and values of RMSE. Explanatory drawing of the step which sets a gradation preferentially with respect to the altitude value with high appearance frequency which the numerical altitude data small gradation reduction means 223 in Embodiment 1 performs. The figure which shows the value of RMSE, each of DEMaa, DEMab, and DEMac which were produced | generated by the 3 times loop of step S45-S55 in Embodiment 1. FIG.

Embodiment 1 FIG.
In the first embodiment, a numerical elevation model evaluation apparatus that evaluates a numerical elevation model by calculating a substantial resolution and gradation number regardless of the apparent resolution and gradation number will be described.

FIG. 1 is a block diagram of a digital elevation model evaluation apparatus 10 in this embodiment.
First, the numerical elevation data storage means 100 is a storage area for storing a plurality of numerical elevation data to be compared.
The appropriate resolution calculation means 200 inputs the numerical elevation data stored in the numerical elevation data storage means 100, calculates the appropriate resolution, and outputs it.
Appropriate resolution refers to resolution that is functioning effectively. Also, there are two types of resolution, horizontal and vertical, which are defined as appropriate resolution (resolution) and appropriate resolution (number of gradations), respectively.

Among digital elevation models, even if the apparent resolution (a × a) is the same, the actual resolution is low, for example, the low resolution (b × b) numerical elevation model has an apparent resolution ( There may be a digital elevation model that is just upsampled (enlarged) in a × a).
Appropriate resolution calculation means 200 according to the first embodiment receives an apparent resolution (a × a) instead of an apparent resolution (a × a) when a digital elevation model having a substantially lower resolution than the apparent resolution is input. A resolution (b × b) lower than xa) is output as an appropriate resolution (resolution).

  Similarly, the digital elevation model is set as if the number of gradations is large, although the number of gradations (appropriate resolution (number of gradations)) that function effectively is small, depending on the generation method. (Although d is sufficient as the actual number of gradations, there are cases where the number of gradations is expressed by c gradations larger than d. The appropriate resolution calculation means 200 according to the first embodiment uses the apparent number of gradations c. When a numerical altitude model of d having a substantially smaller number of gradations is input, the effective number of gradations d is output as an appropriate resolution (number of gradations) instead of the apparent number of gradations c. How to derive the appropriate resolution will be described later.

  In the digital elevation model evaluation apparatus 10, there are as many appropriate resolution calculation means 200 as the number of digital elevation models to be evaluated for the same topography, and the appropriate resolution of each digital elevation model for the same topography is output. It has become.

  The digital altitude data superiority / inferiority comparison means 300 is based on the appropriate resolution (resolution) and the appropriate resolution (number of gradations) output from a plurality of altitude data for each numerical altitude model output by each appropriate resolution calculation means 200. And the ID of the most excellent digital elevation data is output.

Next, FIG. 2 will be described.
FIG. 2 is a configuration diagram of the appropriate resolution calculator 200 in this embodiment.
The configuration of the appropriate resolution calculation unit 200 will be described. The numerical elevation data temporary storage area 210 temporarily stores one numerical elevation data read from the storage area 100.

The numerical elevation data resolution adjusting means 220 is composed of a numerical elevation data downsampling means 221, a numerical elevation data upsampling means 222, and a numerical elevation data sub-gradation means 223.
The digital elevation data downsampling means 221 in the digital elevation data resolution adjustment means 220 downsamples (reduces the image) in the horizontal direction with respect to the digital elevation data stored in the digital elevation data temporary storage area 210, and thereafter The numerical elevation data upsampling means 222 performs resolution changing processing for upsampling (enlarging the image) the numerical elevation data downsampled in the horizontal direction.
Also, for the digital elevation data stored in the digital elevation data temporary storage area 210, the numeric elevation data subgrading means 223 performs a resolution changing process for reducing the gradation of the data in the vertical direction.

The intermediate generated numerical elevation data temporary storage means 230 temporarily stores the numerical elevation data generated by the numerical elevation data resolution adjustment means 220 and adjusted in resolution.
The numerical altitude data error calculation means 240 calculates the mean square error (RMSE) using the numerical elevation data temporary storage means 210 and the two numerical elevation data stored in the intermediate generated numerical elevation data temporary storage means 230. Compare with that.
The numerical altitude data error storage means 250 stores the mean square error (RMSE) calculated and calculated by the numerical altitude data error calculation means 240.
The appropriate resolution calculation means 260 compares the mean square error (RMSE) stored in the numerical elevation data error storage means 250 with a threshold value, and calculates an appropriate resolution (resolution / number of gradations). The appropriate resolution storage means 270 stores the calculated appropriate resolution (resolution / number of gradations).

Next, the operation will be described.
FIG. 3 is a flowchart showing the operation of the digital elevation model evaluation apparatus 10 in this embodiment.
FIG. 3A shows steps S10 to 70 which are the entire flowchart, and FIG. 3B shows steps S20 to S60 among steps S10 to 70 in FIG. 3A. Is shown in detail.
FIG. 4 is a diagram showing an example of the digital elevation model (DEMa, DEMb) in this embodiment.
First, it is assumed that there are DEMa and DEMb as digital elevation models for the same topography to be evaluated. It is assumed that both DEMa and DEMb are numerical elevation data that can be expressed by grayscale values having a resolution of 128 × 128 pixels and a gradation number of 32.

3A, first, in step 10, the digital elevation model is stored in the digital elevation data storage means 100 in the digital elevation model evaluation apparatus 10 as DEMa and DEMb, respectively.
Steps S15 and S65 show a repetitive loop of the flowchart, and k = 1, 2, and 1 described in S15 show an initial value, a closing price, and an increment value, respectively. That is, it shows that the appropriate resolution calculation from step S15 to steps S20 to S60 is repeated twice between S65, and this means that DEMa in one appropriate resolution calculation means 200 as shown in FIG. And the DEMb processing in the other appropriate resolution calculation means 200 are shown.

Next, in step S <b> 20 of FIG. 3B, the numerical elevation data temporary storage unit 210 reads DEMa stored in the numerical elevation data storage unit 100.
Next, in the first time in the three-time loop of steps S25 to S35 in FIG. 3B, the digital elevation data down-sampling means 221 uses the DEMa stored in the digital elevation data temporary storage means 210 in S30 recently. Downsample to a size of 64 × 64 pixels by the side method.

  Subsequently, in step S31, the digital elevation data upsampling means 222 generates numerical elevation data DEMa1 obtained by upsampling the digital elevation data DEMa after downsampling to 128 × 128 pixels by the nearest neighbor method, and intermediately generated numerical elevation data. Store in the temporary storage means 230.

FIG. 5 is a diagram showing changes in numerical elevation data after down-sampling and after up-sampling in this embodiment.
Next, in step 32, the error calculation means 240 between the digital elevation data 240 stores the digital elevation data DEMa1 stored in the intermediate generated digital elevation data temporary storage means 230 and the numerical elevation data DEMa stored in the numerical elevation data temporary storage means 210. The mean square error (RMSE), which is an evaluation error, is calculated from the above. The following formula (1) is a formula for calculating the mean square error (RMSE).

  pij is a pixel value of (i, j) coordinates on the numerical elevation data stored in the intermediate generated numerical elevation data temporary storage means 230, and Pij is (i) on the numerical elevation data stored in the numerical elevation data temporary storage means 210. , J) The pixel value of the coordinate, and n represents the resolution of the numerical elevation data stored in the temporary digital elevation data storage means 210. Thereafter, the calculated RMSE is stored in the numerical altitude data error storage means 250.

  Next, in the second time in the three-time loop of steps S25 to S35 in FIG. 3B, the above-described processing of S30 to S32 is performed, but the difference from the first time is stored in the numerical elevation data temporary storage means 210. Down-sampled to a size of 32 × 32 pixels by the nearest neighbor method, and generate digital elevation data DEMa2 up-sampled to 128 × 128 pixels by the nearest neighbor method, and intermediate generation numerical elevation data temporary storage means 230 only stored.

  Further, in the third time in the three-time loop of steps S25 to S35 in FIG. 3B, the above-described processing of S30 to S32 is performed, but the difference from the first time is stored in the digital elevation data temporary storage means 210. The digital elevation data DEMa3 is down-sampled to 16 × 16 pixels by the nearest neighbor method and up-sampled to 128 × 128 pixels by the nearest neighbor method, and the intermediate generated numeric elevation data temporary storage means 230 is generated. Just remember.

  FIG. 6 is a diagram illustrating DEMa1, DEMa2, and DEMa3 generated by the three-time loop of steps S25 to S35 in the present embodiment and the respective RMSE values.

Further, in step S40 of FIG. 3B, the appropriate resolution calculation means 260 uses the RMSE stored in the numerical altitude data error storage means 250 so that the appropriate resolution calculation means 260 has a minimum resolution within a range in which the RMSE does not exceed a certain threshold T. (Resolution) is calculated.
For example, when the threshold value T = 6 is set, when the RSME is as shown in FIG. 6, the appropriate resolution (resolution) is 64 × 64), which corresponds to the appropriate resolution (resolution). The value of RSME is not much different from the value of RMSE of the original digital elevation model even if the RMSE is calculated by reducing the resolution of the digital elevation model having a low effective resolution.
Therefore, in step S40 in FIG. 3B, an appropriate resolution (resolution) can be obtained by obtaining a minimum resolution in which an error from the original digital elevation model does not become larger than the threshold when downsampling and upsampling are performed. it can.

  Next, it is shown that the loop processing is performed three times in steps S45 to S55 in FIG. 3B. This is the same as obtaining the appropriate resolution (resolution), as in the case of obtaining the appropriate resolution (gradation). Is a process for obtaining.

Details will be described below.
In step S50, DEMa stored in the digital elevation data temporary storage means 210 is read.
Next, the numerical elevation data sub-gradation means 223 generates numerical elevation data DEMaa with the number of gradations reduced to 16, and stores it in the intermediate generated numerical elevation data temporary storage means 230. The numerical altitude data sub-gradation means 223 uniformly sets the gradations when reducing the number of gradations (for example, when the number of gradations is 4, the minimum value of the elevation value is 0, and the maximum value is 300, 0, 100). , 200, 300, etc.), a histogram of the altitude is drawn, and an altitude value having a high appearance frequency is preferentially set as a tone value that is a boundary between the tones.

This is shown in FIG.
FIG. 7 is an explanatory diagram of the steps for preferentially setting the gradation with respect to the altitude value having a high appearance frequency, which is performed by the numerical altitude data subgrading means 223 in the present embodiment.
For example, the upper left diagram of FIG. 7 shows the numerical elevation data stored in the numerical elevation data temporary storage means 210, where the vertical axis is the elevation and the horizontal axis is the column of elevation data in the digital elevation data.

A histogram is drawn for the elevation of the column of elevation data in the numerical elevation data. The middle left diagram of FIG. 7 shows a histogram for a column of elevation data in the digital elevation data shown in the upper left diagram of FIG.
As shown in the middle left diagram of FIG. 7, when the gradation is set for the altitude value having a high appearance frequency, the histogram gradation is set to 10, 100, 160, 270. Then, the gradation is reduced based on the set gradation.

The lower left diagram of FIG. 7 is a diagram comparing the digital elevation data of the upper left diagram of FIG. 7 with the data with reduced gradation. Further, the right side of FIG. 7 is a diagram showing a difference from the case where the same processing as in the left side of FIG. is there.
In other words, by using the histogram to set the boundary of the gradation using the dense part of the gradation, the boundary is set at random (such as setting the gradation to 0, 100, 200, 300). The gradation can be set more accurately than the gradation.

  In S51 of FIG. 3 (b), the error calculation means 240 between the numerical elevation data 240 calculates the DEMaa stored in the intermediate generated numerical elevation data temporary storage means 230 and the numerical elevation data DEMa stored in the numerical elevation data temporary storage means 210. To calculate RMSE. The numerical altitude data error calculating means 240 stores the calculated RMSE in the numerical altitude data error storing means 250.

  In the above, the digital elevation data DEMaa with the number of gradations reduced to 16 is calculated as subtraction of the digital elevation data as the first processing of the third loop processing in steps S45 to S55 in FIG. The process of calculating the RMSE with DEMa has been described. As the second process of the three-time loop process, the calculation of numerical elevation data DEMab with the number of gradations reduced to 8 and the RMSE with DEMa are calculated. As the third processing of the loop processing, calculation of numerical elevation data DEMac with the number of gradations reduced to 4 and RMSE with DEMa are calculated.

An example of the result of the processing of steps S45 to S55 in FIG. 3B is shown in FIG.
FIG. 8 is a diagram illustrating DEMaa, DEMab, DEMac and RMSE values generated by the three-time loop of steps S45 to S55 in the present embodiment.

Next, in S60 of FIG. 3B, the appropriate resolution calculation means 260 determines that the RMSE is tm from the RMSE for DEMaa, DEMab, and DEMac stored in the digital altitude data error storage means 250 and a certain threshold value t. The resolution (number of gradations) when the resolution (number of gradations) is minimized is calculated and stored in the appropriate resolution storage means 270. For example, when the threshold t = 6 and the RMSE is as shown in FIG. 8, the appropriate resolution (number of gradations) is 8.
Even if the number of gradations is lowered, the digital altitude model with low effective resolution does not change much from the original digital altitude model, and therefore tends to be difficult to rise as RMSE. Therefore, as in step S60, an appropriate resolution (the number of gradations) can be obtained by obtaining the maximum number of gradations that causes an error to be increased with the original numerical elevation model when the gradation is reduced.

  Next, in S65 in FIG. 3A, as described above, the appropriate resolution calculation unit 200 has calculated the numerical elevation model DEMa. Subsequently, as the second loop of S15 to S65, the numerical elevation model DEMb As in the case of DEMa, an appropriate resolution (resolution) and an appropriate resolution (number of gradations) are calculated and stored in the appropriate resolution storage means 270.

Finally, in S70 in FIG. 3A, the digital elevation data superiority / inferiority comparison means 300
Based on the appropriate resolution (resolution and the number of gradations) of DEMa acquired from the appropriate resolution calculation means 200 and the appropriate resolution (resolution and the number of gradations) of DEMb, it is determined which of DEMa and DEMb has the higher appropriate resolution.
In order to automatically determine superiority or inferiority in consideration of both the resolution in the horizontal direction (resolution) and the vertical direction (number of gradations), for example, the following equation is used.
Appropriate resolution (resolution) × w + appropriate resolution (number of gradations) × (1-w) (2)
Here, w is a weighting coefficient. Here, w is set to 0.5, and it is assumed that the horizontal and vertical resolutions can be determined with the same weighting.
The digital elevation data superiority / inferiority comparison means 300 determines that the digital elevation model having a large calculation result of the equation (2) is the best.

  Therefore, the digital elevation model evaluation apparatus in this embodiment includes the digital elevation data downsampling means 221 that downsamples the digital elevation data that is the data of the digital elevation model, and the digital elevation data that upsamples the downsampled digital elevation data. Upsampling means 222, numerical elevation data small gradation reduction means 223 for reducing the number of gradations of the numerical elevation data, numerical elevation data that is data of the digital elevation model, upsampled numerical elevation data and gradation are reduced. Since the digital altitude data error calculation means 240 for calculating the evaluation error from the calculated digital elevation data and the appropriate resolution calculation means 260 for calculating the appropriate resolution from the calculated evaluation error are provided, It is possible to evaluate the accuracy of the elevation model That.

  In addition, the numerical elevation model evaluation apparatus according to the present embodiment is a numerical value for evaluating which numerical elevation data has a higher appropriate resolution by comparing the respective appropriate resolutions calculated from the numerical elevation data of a plurality of numerical elevation models. Elevation data superiority / inferiority comparison means 300 is provided, and it is possible to evaluate which numerical elevation data has a higher appropriate resolution from the numerical elevation data of a plurality of digital elevation models. Furthermore, the appropriate resolution (resolution) and the appropriate resolution (number of gradations) By evaluating which resolution is higher according to the weighting coefficient, it is possible to select whether to place importance on the resolution or the number of gradations depending on the situation to be evaluated.

  In the present embodiment, the resolution and the number of gradations of the input digital elevation model are limited, but the present invention is not limited to these values. Further, the resolution, which is the resolution, is not limited to the same aspect ratio.

In the present embodiment, the nearest neighbor method is used for the downsampling in the digital elevation data downsampling means 221, but other alternative interpolation methods such as a linear interpolation method and a bicubic method may be used. Similarly, the digital elevation data upsampling means 222 uses the nearest neighbor method, but other alternative interpolation methods such as a linear interpolation method and a bicubic method may be used.
Which interpolation method is suitable for downsampling and upsampling depends on the algorithm that created the digital elevation model and the topography of the point where the digital elevation model was created. For example, in the case of mountainous areas, it is considered that linear interpolation is appropriate based on the formation process of the mountainous areas.

  In the present embodiment, the digital elevation data downsampling means changes the resolution of the original numeric elevation data to 64 × 64, 32 × 32, and 16 × 16 when downsampling. It is not limited to. Further, it is not necessary to limit the resolution between two times, and it is not necessary to limit the resolution to only three stages. For example, the changed resolution may be set to 100 × 100, 75 × 75, 50 × 50, or 25 × 25.

  Similarly, the numerical altitude data small gradation reducing unit 223 changes the original numerical altitude data to 16, 8, and 4 gradations, but this is not limited to this resolution. Further, it is not necessary to limit the number of gradations to twice, and the number of gradations need not be limited to three stages. For example, the changed resolution may be set to 20, 15, 10, and 5.

  In the present embodiment, the numerical elevation data sub-gradation means 223 determines the gradation by looking at the peak of the elevation histogram, but the gradation determination method is not limited to this. For example, the numerical elevation data may be divided into a target number of gradations by a clustering algorithm (k-means method or the like), and the cluster center may be reduced to a small gradation using the gradation as a gradation.

  In the present embodiment, when the digital elevation data error calculation means 240 compares the digital elevation data, evaluation is performed by calculating a mean square error (RMSE), but this is not limited to RMSE. For example, instead of the mean square error (RMSE), a value obtained by averaging the absolute values of errors may be used.

  In this embodiment, the numerical altitude data superiority / inferiority comparison means 300 sets the weighting factor of the appropriate resolution when determining the superiority or inferiority of DEMa and DEMb to 0.5, but this is between 0 and 1 If it is a real number, it is not limited to 0.5. In other words, when the weighting factor is 0 or 1, it is one of two of appropriate resolution (resolution) and appropriate resolution (number of gradations).

  In this embodiment, DEMa and DEMb, which are digital elevation models, are compared. However, the comparison is not limited to two digital elevation models. For example, three or more digital elevation models are added by adding DEMc or the like. You may compare.

  In this embodiment, when the superiority or inferiority judgment is performed by the numerical elevation data superiority / inferiority comparison means 300, an index that integrates two indices is created and the result of superiority / inferiority determination is returned. Instead, for example, the results may be returned such that DEMa is superior for the horizontal resolution (resolution) and DEMb is superior for the vertical resolution (number of gradations). In this case, it is determined that the one with the larger resolution value wins.

DESCRIPTION OF SYMBOLS 10 Numerical elevation model evaluation apparatus 100 Numerical elevation data storage means 200 Suitable resolution calculation means 210 Numerical elevation data temporary storage means 220 Numerical elevation data resolution adjustment means 221 Numerical elevation data downsampling means 222 Numerical elevation data upsampling means 223 Numerical elevation data small Gradation means 230 Intermediate generated numerical elevation data temporary storage means 240 Error calculation data between numerical elevation data means 250 Error storage means between numerical elevation data 260 Suitable resolution calculation means 270 Appropriate resolution storage means 300 Numerical elevation data superiority comparison means

Claims (7)

Digital elevation data downsampling means for downsampling digital elevation data that is data of the digital elevation model,
Digital elevation data upsampling means for upsampling the downsampled digital elevation data;
Numerical elevation data sub-gradation means for reducing the number of gradations of the numerical elevation data that is the data of the digital elevation model;
An error calculation means between numerical elevation data for calculating an evaluation error from the numerical elevation data that is the data of the digital elevation model, the upsampled numerical elevation data, and the gradation-converted numerical elevation data;
An appropriate resolution calculating means for calculating an appropriate resolution from the calculated evaluation error;
A digital elevation model evaluation device characterized by comprising:   The nearest neighbor method, the linear interpolation method, or the bicubic method is applied to the downsampling and upsampling used in the digital elevation data downsampling means and the digital elevation data upsampling means. The numerical elevation model evaluation device described.   2. The altitude value having a high appearance frequency is set as a tone value from a histogram of the altitude of the numerical elevation data when the numerical elevation data subgrading means reduces the number of gradations of the numerical elevation data. Or the digital elevation model evaluation apparatus according to 2;   4. The digital elevation model evaluation according to claim 1, wherein the evaluation error calculated by the error calculation means between numerical elevation data is RMSE (mean square error) or an average of absolute values of errors. apparatus.   A digital elevation data superiority / inferiority comparison means for evaluating which digital elevation data has a higher appropriate resolution by comparing each appropriate resolution calculated from the digital elevation data of a plurality of digital elevation models is provided. The numerical elevation model evaluation apparatus according to any one of claims 1 to 4.   The digital elevation data superiority / inferiority comparing means includes an appropriate resolution (resolution) calculated using the upsampled digital elevation data, and an appropriate resolution (number of gradations) calculated using the numerical elevation data reduced in gradation. The numerical elevation model evaluation apparatus according to claim 5, wherein the numerical elevation data is evaluated as to whether the appropriate resolution of the numerical elevation data is high. Digital elevation data downsampling step for downsampling digital elevation data that is data of the digital elevation model,
A digital elevation data upsampling step for upsampling the downsampled digital elevation data;
Numerical elevation data sub-gradation step for reducing the number of gradations of the numerical elevation data that is the data of the digital elevation model;
An error calculation step between numerical elevation data for calculating an evaluation error from the numerical elevation data that is data of the digital elevation model and the upsampled numerical elevation data or the subtracted numerical elevation data;
An appropriate resolution calculating step for calculating an appropriate resolution from the calculated evaluation error;
A digital elevation model evaluation method having

Images