JP5222314B2 - Surface measurement method and apparatus - Google Patents

Description

  The present description relates to a method and apparatus for measuring the surface of an object.

  There is known a surface measurement method and apparatus used for measuring a surface dimension after processing in order to determine whether or not a processed object satisfies a processing instruction value.

  For example, in Patent Document 1, when a large processing error occurs in the concave measurement result of the processed object, a measurement point that deviates from the function by a predetermined value or more is removed using an error distribution function of the molding surface. A technique for estimating an error amount and outputting a tool path for correction machining is described.

  Thus, in the case of the target object in which a recessed part is formed in the surface, it is necessary to remove a recessed part and to confirm the processing precision of other plane parts.

  Conventionally, techniques described in Non-Patent Document 1 and Non-Patent Document 2 are known as methods for separating a flat portion and a concave portion.

JP 2009-136937 A Otsu, Nobuyuki: A Threshold Selection Method From Gray-Level Histograms, IEEE Transactions on Systems, Man, and Cybernetics, Vol. SMC-9, No. 1, pp. 62-66, 1979 Nobuyuki Otsu: Automatic threshold selection method based on discrimination and least squares criterion, IEICE Transactions D, Vol.63, No.4, pp.349-356, 1980

  The discriminant analysis binarization method described in the above-mentioned non-patent document is one of image processing methods, and has been proposed as a method for automatically determining a threshold for separating a background from an image photographed by a camera.

  Consider the case where the discriminant analysis binarization method is applied to a method for separating a flat portion and a concave portion of a measurement object.

  First, an approximate straight line is created by a least square method from a plurality of measurement data obtained from an object. Next, a difference between the approximate line and the measurement data is created. Thereafter, the minimum value and the maximum value of the difference are created, and the interval between the minimum value and the maximum value is divided, and the number of the differences that fall within the interval is determined. A histogram with the interval number as the horizontal axis and the number as the vertical axis is created, and the inter-class variance in the discriminant analysis binarization method is created from the histogram.

  Next, the maximum value of interclass variance is set as a threshold value. Since this threshold value is a section number, a difference value corresponding to the section number is determined, it is determined whether the difference between the approximate line and the measurement data is greater than or equal to the difference determined from the threshold value, and the measurement data is separated. In this way, the plane portion and the recess can be separated.

  However, when the above method is applied to an object in the case where a small concave portion (hole portion) is formed in the vicinity of the flat portion, the maximum value of the interclass variance with respect to the histogram becomes a threshold value. The difference from the measurement data is smaller than the threshold value, and the hole portion cannot be removed.

  In addition, when the discriminant analysis binarization method is applied to measurement points where the variance between classes is generally a normal distribution, a threshold value is set in the central portion of the normal distribution. There exists a problem that the above measurement data will be isolate | separated.

  An object of the present invention is to realize a surface measurement method and apparatus capable of accurately separating a flat portion and a concave portion of a measurement object.

  In order to achieve the above object, the present invention is configured as follows.

  A surface measurement method and apparatus for separating a planar portion and a concave portion of a surface of a measurement object, calculating an approximate line by a least square method according to position data of the surface of the measurement object, and calculating the approximate line and the measurement data Divide into multiple classes having class numbers corresponding to the distances, calculate the number of distance values between the calculated approximate line and the measured data that enter the divided classes, and determine the class number of the class with the largest interclass variance. The average value of all classes is calculated from the class number and the number of members in the class, and the measurement is less than the first threshold until the difference between the first threshold and the average value is within a certain value. For the measurement data obtained by excluding the data, the approximation line, the class, and the inter-class variance are calculated again, the first threshold value and the average value are calculated, and the difference from the average value is a constant value. To the first threshold within Use response to measurement data to the measuring object plane of the data.

  ADVANTAGE OF THE INVENTION According to this invention, the surface measuring method and apparatus which can isolate | separate the plane part and recessed part of a measurement target object correctly are realizable.

It is a schematic block diagram of the surface measuring apparatus which is Example 1 of this invention. It is an internal block diagram of the computer which performs the measurement point separation process of the surface measuring apparatus shown in FIG. It is a process flowchart of the measurement point separation method in Example 1 of this invention. It is a figure which shows an example for demonstrating an approximate line, a histogram, and a threshold value of the measurement data to which this invention is applied. It is a figure which shows the other example for demonstrating an approximate line, a histogram, and a threshold value of the measurement data to which this invention is applied. It is a figure which shows the further another example for demonstrating an approximate line, a histogram, and a threshold value of the measurement data to which this invention is applied. It is a figure which shows the further another example for demonstrating an approximate line, a histogram, and a threshold value of the measurement data to which this invention is applied. It is a data structure figure which shows the measurement point data in this invention, frequency data, straight line data, and threshold value 1 data.

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

  FIG. 1 is a schematic configuration diagram of a surface measuring apparatus according to an embodiment of the present invention, FIG. 2 is a configuration diagram of a computer 200 that performs the measurement point classification process 108 of FIG. 1, and FIG. 3 is a measurement point classification process of FIG. FIG. 4 to FIG. 7 are explanatory diagrams showing the relationship between measurement points, approximate curves and histograms, and FIG. 8 is an explanatory diagram showing the data structure used in the measurement point classification processing.

  In the embodiment of the present invention, the measurement object is an object that reflects light, such as a metal workpiece. Then, in order to determine whether or not the processed object satisfies the processing instruction value, the surface measuring device used for measuring the surface position dimension (position data (coordinate data)) after processing. Examples will be described.

  In FIG. 1, 101 is a sensor, 102 is line light (measurement light) emitted from the sensor, 103 is a measurement object, 104 is diffuse reflection light generated when the line light 102 is irradiated on the measurement object 103, and 105 is The field of view of the camera (reflected light imaging means) built in the sensor 101, 106 a controller (measuring point conversion means) for converting an image signal from the camera built in the sensor 101 into measurement points, and 107 for storing the measurement points. A storage device 108 is a measurement point classification processing function block executed by the computer.

  In FIG. 1, the sensor 101 is disposed above the measurement target 103 and is in a state where the upper surface of the measurement target 103 can be photographed. As shown in FIG. 2, the computer 200 includes a communication device 201 for transmitting and receiving data to the controller 106 in FIG. 1 and other computers (not shown), an input device 202 such as a mouse and a keyboard, a display device 203 such as a graphic display, A storage device 204 such as a hard disk or an external memory, a memory 205 for storing command data, a data transfer to the communication device 201, the storage device 204, and the memory 205 in accordance with input data from the input device 202, and a display on the display device 203 A central processing unit 206 is provided for executing instructions for sending data. The slit light 102 is irradiated from the sensor 101 onto the measurement object 103, the diffuse reflected light 104 of the slit light 102 is read by a camera built in the sensor 101, and the image is processed by the controller 106. It is created by converting to a point in a three-dimensional space.

  The operation of the measurement point classification process 108 executed by the computer 200 will be described with reference to FIGS.

  When the computer 200 inputs a process execution command from the input device 202, the measurement point data stored in the storage device 107 is read by the measurement point reading process 110 (step 301 in FIG. 3). The measurement point data read from the measurement point reading process 110 is excluded by the out-of-range point exclusion process 111 (step 302 in FIG. 3).

  Here, the data outside the measurement range is non-numeric data “NaN”, measurement range upper limit data “1000.0” and measurement range lower limit data “−1000” determined by the specifications of the sensor 101 and stored in the storage device 107 in advance. .0 ".

  The exclusion process is to set exclusion to the identifier of the measurement point data shown in FIG.

  More specifically, the identifier is composed of logical data, and “True” indicating adoption and “False” indicating exclusion are set. The identifier is not limited to logical data, and may be an integer, real number, character string, or the like that can be distinguished from adoption and exclusion.

  Next, the read mode measurement points are separated into necessary measurement points by the most frequent point extraction process (measurement data extraction processing means) 112 (step 303 to step 308 in FIG. 3).

  Next, the measurement points are classified by the threshold value 2 set in advance and stored in the storage device 107 by the accuracy improvement processing (accuracy improvement processing means) 113 (steps 310 to 312 in FIG. 3).

  The threshold value 2 is a maximum allowable dispersion value determined by, for example, the specification of the measurement object, specifications, and the like. For this reason, the threshold value 2 is predetermined for each measurement object and stored in the storage device 107.

  Finally, the measurement point storage processing 114 stores the separated measurement points in the storage device 107 (step 313 in FIG. 3).

  Next, operations of the most frequent point extraction process 112 and the system improvement process 113, which are the features of the present invention, will be described in detail with reference to the flowchart of FIG.

  First, the most frequent point extraction process 112 will be described using steps 303 to 308 in the flowchart of FIG.

  The most frequent point extraction process 112 receives, as an input, measurement point data in which an identification code indicating adoption / rejection is set as an identifier for identifying adoption / rejection of a measurement point by the out-of-range point removal process 111.

  First, in step 303, measurement point data to which an identification code indicating adoption is set is extracted from the measurement point data, and a curve is created from the extracted measurement point data by the least square method.

  In one embodiment of the present invention, the curve is a primary curve (straight line) for the sake of simplicity. The curve is based on the premise that a normal equation in the least square method can be determined. However, an Nth order polynomial curve, a trigonometric curve, a spline curve, or the like may be used.

  The formula for defining the least squares curve is shown in (Formula 1).

In the above (Formula 1), P _i is the i-th adopted measurement point. If the measurement point is two-dimensional, P _i = [X _i , Y _i ] is a vector represented by P _i = [X _i = [X _i]. , Y _i , Z _i ], Q (t _i ) is a point at the parameter t _i on the curve.

Here, the curve Q (t _i ) is, for example, an expression shown in (Expression 2) if it is a first-order polynomial, and Q _j in (Expression 2) is a two-dimensional or three-dimensional coefficient vector in the first-order polynomial.

  The curve is expressed by (Expression 3) by the coefficient vector created by Expression (2) and the parameter t.

  For example, in the case of a two-dimensional straight line (Equation 3), L is set to 2 and t is changed to determine a point on the straight line.

  Next, in step 304, a difference between the measurement point and the curve is created. This difference has a plus or minus sign for the curve. A positive distance represents the upper part of the curve, and a negative represents the lower part of the curve. For example, in FIG. 4, the distance from the upper measurement point 501 with respect to the least square curve 502 is positive, and the lower side is negative.

The calculation method of the length (difference (distance L _i ) between the measurement point and the curve) is shown in (Formula 4) and (Formula 5).

The length L _i is obtained by first obtaining the curve parameter t of the shortest point of the curve and the measurement point by (Formula 4) and substituting the curve parameter t into (Formula 5). Here, (Formula 5) is a case where the curve is on a two-dimensional plane, and n is a normal line of the plane.

When the curve is in a three-dimensional space, the distance L _i is obtained by (Formula 6).

In (Expression 6), C ₀ is a point (measurement data point) below the curve.

  Next, in step 305, a histogram is created.

  First, the maximum value and the minimum value of the difference between the distance between the measurement point and the curve obtained in step 304 are extracted. For the maximum value and the minimum value, for example, the distances obtained by (Equation 5) or (Equation 6) are rearranged in ascending order, and the first and last values of the rearranged distances may be extracted.

  As another method, the initial value of the distance obtained by (Equation 5) or (Equation 6) is set as the initial value of the maximum value and the minimum value, and when the distance smaller than the minimum value appears, the minimum value is set. The process of replacing the maximum value when a distance greater than the maximum value appears may be repeated until the end.

The maximum value L _Max and the minimum value L _Min of the distance are shown in (Formula 7).

  Next, the minimum value and the maximum value for each section are determined by (Formula 8).

  In (Equation 8), W is a predetermined number of sections (the number of sections on the horizontal axis of the histogram) and is stored in the storage device 107. k is a coefficient that varies from 0 to W-1.

As shown in (Formula 8), a section width is created by dividing the difference between the maximum value L _Max of the distance and the minimum value L _Min by the number of sections W, and the section width is multiplied by a coefficient k to obtain the minimum By adding the value L _Min , the minimum value L _{k, Min} and the maximum value L _{k, Max} for each section are determined.

  Next, the frequency for each section is created.

Since the frequency is the number of measurement data points having a distance entering the section, the number of measurement points that are equal to or greater than the minimum value L _{k, Min} and equal to or less than the maximum value L _{k, Max} may be determined.

  503 and 513 in FIG. 4 are examples of histograms created in this way.

  Next, in step 306, a first threshold value (threshold value 1) is created.

The frequency in the interval I that was created in step 305 and _{n I.} Here, I varies from 1 to W.

  First, the frequency is normalized using (Formula 9).

The reason for normalizing the frequency n _I is to simplify the following (Equation 10) to (Equation 13), and the normalized frequency P _I of (Equation 10) to (Equation 13) is expressed as (Equation 9). ), It is not necessary to normalize the frequency.

The overall average level mu _T in normalized frequency and (Equation 10).

The interclass variance σ _B ² (I), which is the variance in class I, is given by (Equation 11).

  In (Expression 11), ω (I) and μ (I) are functions expressed by (Expression 12) and (Expression 13).

  The first threshold 1 is determined by a section I in which the variance between the sections of (Formula 11) is maximum.

  In (Formula 11), when the denominator is 0, the variance between sections is also 0. This needs to be taken into consideration in order to avoid 0 percent when the processing is implemented in the computer.

  The threshold value 1 is calculated by executing the processes described in (Equation 9) to (Equation 13).

  Next, in Step 307, it is determined whether or not the threshold 1 obtained in Step 306 and the average value can be regarded as the same. The range in which the threshold value 1 can be regarded as the same as the average value described later is an amount that depends on the total number of sections and the distribution degree of measurement points, and is preset and stored in the storage device 107.

  In addition, the range which can be considered to be the same in the first embodiment is set as a case where 5% of the total number of sections is obtained as a result of executing a large number of measurement point samples by the surface measurement apparatus capable of executing the processing according to the present invention.

  If it is determined in step 307 that the threshold value and the average value can be regarded as the same, the accuracy improving process 113 is executed.

  By providing step 307, when a measurement point 601 having a frequency close to a normal distribution as shown in FIG. 5 described later is input, it is determined that the threshold interval I and the average value are the same. Measurement point elimination is not performed. For this reason, the problem that more than half of measurement points are separated can be prevented.

  When the normal distribution frequency is input, the threshold value interval I and the average value are the same, which means that a large number of measurement point samples are executed by the surface measurement apparatus capable of executing the processing according to the present invention. It is a matter obtained from the result.

  If it is determined in step 307 that the threshold value 1 is different from the average value, the process of step 308 is performed.

  In step 308, the measurement data below the threshold value 1 is excluded and separated from the measurement data exceeding the threshold value 1. Then, returning to step 303, steps 303 to 308 are executed.

In step 306, threshold value 1 is given by section number I as shown in (Formula 11). Therefore, the section number I which is the threshold value 1 is set to the coefficient k of (Formula 8), and the minimum value L _{I, Min} and the maximum value L _{I, Max} of the section are created.

Then, an average value is obtained by (LI _{, Min} + LI _{, Max} ) / 2, and threshold 1 is set as a distance threshold.

  Next, in step 307, it is determined whether or not the threshold value 1 and the average value are the same (the difference between them is within a predetermined range). If not, in step 308, the distance between the measurement points and the distance threshold value 1 are compared. Then, with respect to the measurement points having a distance equal to or less than the distance threshold value 1, exclusion is set as the identifier of the measurement point data shown in FIG. After the setting, the processes in steps 303 to 308 are repeated in the same manner as described above.

  As described above, by performing the processing from step 303 to step 308, the histograms 503 and 513 of the measurement points 501 and 511 shown in FIG. 4 are separated by the threshold value 1, and finally, as shown in the histogram 603 of FIG. A histogram close to a normal distribution can be created. For this reason, a hole part can be excluded from the measurement point 501 and the measurement point 511 shown in FIG.

  In step 307, the relationship between whether the threshold value 1 and the average value match is not an amount that changes depending on the distribution state of the measurement points as in the class separation degree. Can be terminated. For this reason, automatic separation processing can be realized.

  FIG. 6 is a diagram illustrating a result of executing the processing from step 303 to step 309 on the measurement point 701 and the approximate straight line 702. The interclass variance 704 is created from the histogram 703 in FIG. 6, and the measurement value 701 is separated using the average value 706 with the maximum value 705 as the threshold value 1. As a result, it is possible to separate the measurement points having a normally distributed histogram, and to eliminate the V-shaped hole data (measurement data 701 on the right side in FIG. 6) generated at the measurement point 701. It becomes possible.

  Next, the accuracy improvement processing 113 shown in FIG. 1 will be described using steps 310 to 312 shown in the flowchart of FIG.

  In step 310, the length from the average value 605 shown in FIG. 5 to both ends of the histogram 603 is calculated for the measurement data extracted in the most frequent point extraction process 112 (steps 303 to 308).

The average value 605 is (L _{I, Min} + L _{I, Max} ) / 2, and the maximum and minimum distances are L _Min and L _Max. Therefore, the average value 605 is represented by the histogram 603 from the average value 605 as shown in (Formula 14). The lengths L _Left and L _{Right up} to both ends are created.

Next, in step 311, it is determined whether or not the lengths L _Left and L _Right are smaller than the second threshold value 2 set in advance and set in the storage device 107.

If the lengths L _Left and L _Right are smaller than the threshold value 2, the measurement points are stored in the storage device 107 in step 313, and the process is terminated.

If the lengths L _Left and L _Right are larger than the threshold value 2 in step 311, step 312 is executed and measurement points equal to or less than the threshold value 2 are excluded.

That is, first, the smaller one of the distances L _Left and L _Light is set as L _thred , then L _thred is compared with the threshold 2 and the smaller one is redefined as L _thred .

Subsequently, as shown in (Formula 15), a threshold value is created by _{subtracting the} average value plus L _thred .

  Finally, a measurement point having a distance that does not fall between the threshold values created in (Equation 15) (between the threshold values 808 and 809 in FIG. 8) is determined, and the identifier of the measurement point data shown in FIG. Then, the measurement point storage in step 313 is executed.

  As described above, by executing the processing from step 310 to step 312, the measurement point having the width indicated by the measurement point 803 in FIG. 7 can be separated into the measurement points having the width according to the threshold value 2. Therefore, the vibration width that increases depending on the camera and the measurement object can be reduced.

  The section, that is, the class, can be defined as a plurality of classes having a class number corresponding to the distance with a predetermined number of divisions between the maximum value and the minimum value of the calculated distance.

  The average value can be defined as a value obtained by multiplying the class number by the number of classes and adding the number of all classes and dividing by the total number.

  Next, an example when the present invention is not applied will be described with reference to FIGS.

  The lower left of FIG. 4 is an example in which the discriminant analysis binarization method is applied to the measurement point 501 at the upper left of FIG. In the example shown in FIG. 4, first, an approximate straight line 502 is created from the measurement point 501 by the least square method. Next, a difference between the straight line 502 and the measurement point 501 is created.

  Thereafter, a minimum value and a maximum value of the difference between the straight line 502 and the measurement point 501 are created, and the interval between the minimum value and the maximum value is divided into a plurality of sections, and the number of the differences that enter each section is determined. A number is assigned to the divided sections, and a histogram 503 having the horizontal axis and the number of differences as the vertical axis is created, and an interclass variance 504 in the discriminant analysis binarization method is created from the histogram 503.

  Next, the maximum value 505 of the interclass variance 504 is extracted and set as a threshold value. Since this threshold value is a section number, a difference value corresponding to the section number is determined, it is determined whether or not the difference between the measurement point 501 and the approximate straight line 502 is greater than or equal to the difference determined from the threshold value, The measurement points 501 that are less than the difference are separated. In this way, only the upper measurement point can be separated.

  Here, when the discriminant analysis binarization method is simply applied to the measurement data 511 having the hole portion 517 as shown in the upper right of FIG. 4, the maximum value 515 of the interclass variance 514 with respect to the histogram 513 becomes the threshold value. In this case, since the measurement data 517 is less than the difference determined from the threshold value, the hole portion indicated by the measurement point 517 remains at the measurement point 511.

  That is, when specifying a planar portion of the measurement object and confirming the processing accuracy, if a minute hole portion within the design range exists in the vicinity of the planar portion, this hole portion is also a part of the measurement data 511. It will be judged as a part. For this reason, there is a possibility that the accuracy of the measurement value of the flat surface portion is lowered.

  Further, when the discriminant analysis binarization method is applied to measurement points having a globally normal distribution as shown in FIG. 5, a threshold value 605 is set in the center portion of the interclass variance 604, and therefore an approximate straight line More than half of the measurement points below 602 are removed, which may reduce the accuracy of the measurement values on the flat surface.

  Class separation is not an absolute quantity, but varies depending on the distribution of measurement points. For example, in the case of the distribution shown in FIG. 4, the class separation degree is 0.978 at the beginning, and after separation with the threshold value determined by this class separation degree, the class separation degree becomes 0.764, and with the next determined threshold value. After separation, the value becomes 0.547, and at this time, a measurement point from which the hole 517 is removed may be obtained.

  On the other hand, in the case of the distribution of measurement data shown in FIG. 6, the class separation degree is 0.633 for the first and 0.621 for the second. The removed measurement data may be obtained.

  As described above, the degree of class separation varies depending on the distribution state of measurement points, and in general, it is impossible to predict the distribution state of measurement points. For this reason, a clear index for ending the repetition for removing the hole is not determined, and automatic separation processing is difficult.

  On the other hand, in the present invention, as shown in steps 303 to 308 in FIG. 3, the threshold value and the average value are calculated by the discrimination binarization method, and the threshold value and the average value are within a certain difference. The measurement data below the threshold value is removed, the removed measurement data is used to calculate the threshold value and the average value, and the process is repeated until the threshold value and the average value are within a certain difference.

  As a result, it is possible to clearly determine the end of repeated processing, and automatic separation processing is possible.

  Furthermore, in the present invention, as shown in steps 310 to 312 of FIG. 3, the measurement data is extracted from the measurement data extracted in steps 303 to 308 using a threshold 2 determined by the allowable maximum dispersion value. Since the sorting is performed, the surface measurement accuracy can be further improved.

  The graphs shown in FIGS. 4 to 7 are displayed on the display device 203. Further, data obtained as a result of executing the processing shown in FIG. 3 can be displayed on the display device 203.

  DESCRIPTION OF SYMBOLS 101 ... Measurement sensor, 102 ... Line light, 103 ... Measurement object, 104 ... Scattered light, 105 ... Camera visual field, 106 ... Measurement sensor controller, 107 ... Memory | storage 108, measurement point classification processing unit, 110, measurement point reading means, 111, out-of-range point exclusion means, 112, most frequent point extraction means, 113, accuracy improvement processing means, 114 ... Measurement point storage processing means, 200 ... computer, 201 ... communication device, 202 ... input device, 203 ... display device, 204 ... storage device, 205 ... memory, 206 ... Central processing unit

Claims (4)

In the surface measurement method for separating the flat portion and the concave portion of the surface of the measurement object,
Measure the position of the surface part of the measurement object, calculate the approximate line by the least square method according to the measured data,
Calculate the distance between the calculated approximate line and the measurement data,
Dividing into a plurality of classes having a class number corresponding to the distance by a predetermined number of divisions between the maximum value and the minimum value of the calculated distance, and the calculated approximate line and measurement data entering the divided class Calculate the number of distance values of
By calculating the inter-class variance of the plurality of classes, the class number of the class having the maximum inter-class variance is set as a first threshold, and the average value of all classes is calculated from the class number and the number of classes included in the class,
Determining whether the difference between the first threshold and the average value is within a certain value;
If the difference between the first threshold value and the average value is not within a certain value, the approximate line, the class, and the measurement data obtained by excluding the measurement data less than the first threshold value are again used. Calculating the inter-class variance, calculating the first threshold value and the average value, determining whether the difference between the first threshold value and the average value is within a certain value,
Calculating the first threshold value and the average value until the difference between the first threshold value and the average value is within a certain value;
A surface measurement method characterized in that measurement data corresponding to a first threshold value whose difference from the average value is within a certain value is used as measurement object plane data. The surface measurement method according to claim 1,
For the measurement data corresponding to the first threshold whose difference from the average value is within a certain value, the measurement data outside the predetermined maximum allowable dispersion value that is the second threshold is excluded, and the second A surface measurement method, wherein measurement data within a threshold is used as data of a measurement object plane part. In the surface measuring device that separates the flat part and the concave part of the surface of the measurement object,
A sensor that irradiates the measurement object with measurement light and images reflected light from the measurement object;
Measurement point conversion means for converting an image captured by the sensor into a measurement point;
For the measurement data converted by the measurement point conversion means, an approximate line is calculated by the least square method, the distance between the calculated approximate line and the measurement data is calculated, and the distance between the maximum value and the minimum value of the calculated distance is calculated. Dividing into a plurality of classes having a class number corresponding to the distance by a predetermined number of divisions, calculating the number of distance values between the calculated approximate line and the measurement data entering the divided class, The inter-class variance is calculated, the class number of the class having the maximum inter-class variance is set as the first threshold value, the average value of all classes is calculated from the class number and the number of classes, and the first threshold value is It is determined whether or not the difference from the average value is within a certain value. If the difference between the first threshold value and the average value is not within a certain value, measurement data equal to or less than the first threshold value is excluded. About the measurement data obtained The approximate line, the class, and the inter-class variance are calculated again, the first threshold value and the average value are calculated, and whether or not the difference between the first threshold value and the average value is within a certain value. The first threshold value and the average value are calculated until the difference between the first threshold value and the average value is within a certain value, and the difference between the first value and the average value is within a certain value. Measurement data extraction processing means using the data corresponding to the threshold value of 1 as measurement data;
A surface measuring device comprising: The surface measurement apparatus according to claim 1,
For the measurement data corresponding to the first threshold whose difference from the average value is within a certain value, the measurement data outside the predetermined maximum allowable dispersion value that is the second threshold is excluded, and the second An apparatus for measuring a surface, further comprising an accuracy improvement processing unit that uses measurement data within a threshold to obtain data of a measurement object plane part.

Images