Surface analysis device

Abstract

A surface analysis device is provided. In the case of phase analysis or the like, the analyst is facilitated to perform a transformation, division, integration, or the like of a cluster region, thereby reducing the burden on the analyst and improving the work efficiency. One embodiment of the surface analysis device according to the present application includes: a measurement unit (1, 2, 4, 5, 7, 90) that acquires signals that reflect amounts of a plurality of components or elements that are analysis targets at a plurality of positions on a sample; a scatter diagram generation unit (92) that generates a scatter diagram based on a measurement result of the measurement unit; a cluster analysis unit (93) that clusters points in the scatter diagram; and a cluster region detection unit (94) that, based on cluster information given to each point in the scatter diagram by the cluster analysis unit, calculates cluster region boundary information of a polygonal shape in which the number of vertices is equal to or less than a predetermined number for each cluster.

Description

Technical Field

[0001] This invention relates to a surface analysis device for investigating the distribution of components and elements in one-dimensional or two-dimensional measurement regions on a sample. The surface analysis device includes an electron probe microanalyzer (EPMA), a scanning electron microscope (SEM), and a fluorescence X-ray analysis device, among others. Background Technology

[0002] In elemental mapping analysis using EPMA, it is possible to investigate the types and amounts of elements contained in numerous tiny regions within a two-dimensional area of ​​the sample. A common method is phase analysis (or phase resolution): when analyzing the results obtained through such elemental mapping analysis, a scatter plot of characteristic X-ray intensities of two or three elements, or elemental concentrations calculated from those intensities, is generated. The types and proportions of compounds contained in the sample are then identified based on the distribution of markers on the plot (see Patent Documents 1 and 2). For example, Figure 10 of Patent Document 2 shows an example of a binary scatter plot, and Figure 11 of the same document shows an example of a ternary scatter plot.

[0003] A point on a scatter plot corresponds to a point (tiny region) on the sample. Therefore, a densely packed area of ​​points on the scatter plot is estimated to correspond to a portion of the sample containing the element in the same proportion. Thus, generally, in phase analysis, the analyst identifies a densely packed area of ​​points on the scatter plot as a cluster, i.e., a set of associated points, and uses a pointing device such as a mouse to circle this area with an appropriate graphic such as a polygon, assigning a different display color to each area. Once this is completed, the phase diagram obtained by coloring the positions on the sample corresponding to the points contained in one or more cluster regions with the specified colors is displayed on the EPMA display screen.

[0004] In recent years, with the rapid development of AI (Artificial Intelligence) technology, attempts have been made to utilize this technology to automatically assign numerous points on a scatter plot to multiple sets. For such processing, clustering analysis, as a representative method of unsupervised machine learning, is preferred. Various clustering analysis algorithms are known, but density-based clustering, as disclosed in non-patent documents 1 and 2, is useful as a method for dividing points on a scatter plot into multiple clusters based on their density. Figure 6The example shown illustrates the automatic extraction of clusters using density-based clustering analysis from a binary scatter plot obtained through measurements. In this example, six clusters were known to have been extracted.

[0005] Existing technical documents

[0006] Patent documents

[0007] Patent Document 1: Japanese Patent Application Publication No. 2006-125952

[0008] Patent Document 2: Japanese Patent Application Publication No. 2011-153858

[0009] Non-patent literature

[0010] Non-patent literature 1: Ester M. et al. (3 others), “A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise”, Proceedings of the 2nd International Conference on Knowledge Discovery and Data Mining (KDD-96), pp. 226-231, 1996.

[0011] Non-patent literature 2: Ricardo JGB, Campello et al., “Density-Based Clustering Based on Hierarchical Density Estimates”, Springer, pp. 160-172, 2013.

[0012] Non-patent document 3: "Ramer-Douglas-Peucker algorithm", Wikipedia, [online], [searched on November 11, 2020], Internet<URL:https: / / en.wikipedia.org / wiki / Ramer%E2%80%93Douglas%E2%80%93Peucker_algorithm>

[0013] Non-patent literature 4: Wu S. et al., “A non-self-intersection Douglas-Peuckeralgorithm”, Computer Science, Mathematics, 16th Brazilian Symposium on Computer Graphics and Image Processing, pp. 60-66, 2003.

[0014] Non-patent document 5: "Convex hull algorithms", Wikipedia, [online], [searched on November 11, 2020], Internet<URL:https: / / en.wikipedia.org / wiki / Convex_hull_algorithms> Summary of the Invention

[0015] The problem the invention aims to solve

[0016] When performing automatic clustering of scatter plots as described above, discrepancies often arise between the results and the analyst's judgment of the cluster regions. One reason for this is that the analyst possesses prior knowledge of the proportions of various elements present in the sample, and makes judgments based on this knowledge. In contrast, automatic clustering struggles to reflect information based on such knowledge. Furthermore, in phase analysis, the following operation is also crucial: while the analyst performs operations such as excluding some points from clusters or merging multiple clusters into one on the automatically clustered scatter plot, the analyst simultaneously confirms the corresponding phase diagram.

[0017] Therefore, the manual correction or modification of the results of automatic clustering by the analyst is indispensable for phase analysis. Previously, when the analyst defined cluster regions by enclosing multiple points on the scatter plot using appropriate shapes such as polygons, modifying these regions was very simple. However, the clustering information obtained through automatic clustering only provides identification information such as which cluster each point belongs to (or whether it belongs to any cluster), and does not provide location information of the cluster regions. Therefore, it is difficult for the analyst to perform operations such as moving, deforming, segmenting, or merging multiple clusters regarding the automatically derived clusters. In such cases, the analyst needs to temporarily reset the automatic clustering results and manually re-assign cluster regions while referring to those results. Therefore, operations such as deforming cluster regions are very cumbersome and inefficient.

[0018] This invention was made to solve such problems, and its main objective is to provide a surface analysis apparatus that, when performing phase analysis based on scatter plots to determine cluster regions, enables the analyst to easily perform operations such as modifying, altering, dividing, and integrating cluster regions, thereby reducing the analyst's burden and improving operational efficiency.

[0019] Solution for solving the problem

[0020] One embodiment of the surface analysis apparatus according to the present invention, which is a solution to the above-mentioned problems, comprises:

[0021] The measuring unit acquires signals reflecting the amounts of multiple components or elements being analyzed at multiple locations on the sample.

[0022] The scatter plot generation unit generates a scatter plot based on the measurement results of the measurement unit;

[0023] Cluster analysis unit, which clusters the points in the scatter plot; and

[0024] The cluster region detection unit, based on the clustering information assigned to each point in the scatter map by the clustering analysis unit, calculates the boundary information of the cluster region, which is a polygon with a number of vertices of less than a specified number, for each cluster.

[0025] The surface analysis apparatus involved in this invention includes, for example, EPMA, SEM, and fluorescence X-ray analysis apparatus. In such an analysis apparatus, by repeatedly measuring while changing the position of the excitation line (electron beam, X-ray, etc.) irradiating the sample, signals reflecting the presence of multiple elements at various locations within a two-dimensional or one-dimensional region of the sample can be obtained.

[0026] The effects of the invention

[0027] In the above-described manner of the surface analysis apparatus according to the present invention, the clustering analysis unit uses, for example, a density-based clustering algorithm to cluster all points on the scatter map. Consequently, each point on the scatter map is labeled with the cluster it belongs to (or whether it does not belong to any cluster). This label is clustering information. The cluster region detection unit, based at least on the clustering information of each point, calculates a polygonal shape representing the inner and outer boundaries of each cluster region as cluster region boundary information. This cluster region boundary information may not necessarily include all points labeled with that cluster.

[0028] In the above-described manner of the surface analysis apparatus according to the present invention, it is possible to determine the regions occupied by multiple clusters after automatic clustering in a scatter plot. The determined cluster regions have size and position information in the scatter plot. Therefore, according to the above-described manner of the surface analysis apparatus according to the present invention, processing corresponding to graphical operations such as deformation, movement, segmentation, and integration of cluster regions can be easily performed. This reduces the workload of the analyst during phase analysis and improves work efficiency. Furthermore, in the above-described manner of the surface analysis apparatus according to the present invention, it is also easy to calculate the area value, perimeter, and other values ​​of each cluster region and compare these values ​​between different clusters. Attached Figure Description

[0029] Figure 1 This is a structural diagram of the main part of the EPMA as one embodiment of the present invention.

[0030] Figure 2 This is a flowchart illustrating an example of the processing procedure for detecting cluster regions in the EPMA of this embodiment.

[0031] Figure 3 This is a flowchart illustrating another example of the processing procedure for detecting cluster regions in the EPMA of this embodiment.

[0032] Figure 4 It shows the use of Figure 2 The diagram shows the cluster region boundary lines generated by the processing procedure.

[0033] Figure 5 It shows the use of Figure 3 The diagram shows the cluster region boundary lines generated by the processing procedure.

[0034] Figure 6 This is a diagram illustrating an example of automatic clustering of a binary scatter plot.

[0035] Explanation of reference numerals in the attached figures

[0036] 1: Electron beam irradiation unit; 100: Electron gun; 2: Sample stage; 3: Sample; 4: Spectroscopic crystal; 5: X-ray detector; 6: Electron detector; 7: Sample stage drive unit; 8: Analysis and control unit; 9: Data processing unit; 90: Element intensity calculation unit; 91: Data storage unit; 92: Scatter plot generation unit; 93: Cluster analysis unit; 94: Cluster region detection unit; 95: Observation image generation unit; 96: Display processing unit; 10: Central control unit; 11: Operation unit; 12: Display unit. Detailed Implementation

[0037] The EPMA, which is one embodiment of the display analysis device according to the present invention, will be described with reference to the accompanying drawings. Figure 1This is a structural diagram of the main parts of the EPMA in this embodiment.

[0038] like Figure 1 As shown, the electron beam irradiation unit 1 includes an electron gun 100, deflection coils (not shown), etc., for irradiating a sample 3 placed on a sample stage 2 with a small diameter electron beam. After receiving the electron beam, characteristic X-rays with wavelengths specific to the element are emitted from the surface of the sample 3. In addition, secondary electrons are also emitted from the surface of the sample 3.

[0039] The characteristic X-rays emitted from sample 3 are wavelength-dispersed by spectrophotometer 4, and the diffracted X-rays at specific wavelengths are detected by X-ray detector 5. The electron beam irradiation position on sample 3, spectrophotometer 4, and X-ray detector 5 are always located on a Rowland circle. A drive mechanism (not shown) moves the spectrophotometer 4 linearly while tilting it, and the X-ray detector 5 rotates in conjunction with this movement. Thus, wavelength scanning of the X-rays being analyzed is achieved while satisfying the Bragg diffraction condition—that is, maintaining the incident angle of the characteristic X-rays to the spectrophotometer 4 equal to the exit angle of the diffracted X-rays to the spectrophotometer 4. The detection signal of the X-ray intensity obtained by X-ray detector 5 is input to data processing unit 9. Additionally, electron detector 6 detects secondary electrons emitted from sample 3 and inputs the detection signal of its electron intensity to data processing unit 9.

[0040] The sample stage 2 can be moved along the mutually orthogonal X and Y axes by the sample stage drive unit 7, and the irradiation position of the electron beam on the sample 3 can be scanned two-dimensionally by this movement. Alternatively, the emission direction of the electron beam can be deflected in the electron beam irradiation unit 1 without moving the sample stage 2, thereby scanning the irradiation position of the electron beam on the sample 3.

[0041] The data processing unit 9 includes functional modules such as an elemental intensity calculation unit 90, a data storage unit 91, a scatter plot generation unit 92, a cluster analysis unit 93, a cluster region detection unit 94, an observation image generation unit 95, and a display processing unit 96. The analysis control unit 8 controls the operation of the sample stage drive unit 7, the drive mechanism that moves the spectrophotometer crystal 4 and the X-ray detector 5, etc., to perform analysis on the sample 3. The central control unit 10 handles the overall control and input / output processing of the device, and is connected to the display unit 12 and the operation unit 11, which includes a keyboard, mouse (or other indicating devices).

[0042] Furthermore, for example, all or part of the central control unit 10, the analysis and control unit 8, and the data processing unit 9 may be composed of a personal computer, and each function may be achieved by executing dedicated control / processing software installed on the computer.

[0043] In the case of elemental mapping analysis in the EPMA of this embodiment, the analysis control unit 8 fixes the position of the spectroscopic crystal 4 in a manner corresponding to the characteristic X-ray wavelength of the target element, changes the irradiation position (micro-region) of the electron beam in a predetermined order within a specified (usually specified by the analyst) two-dimensional region on the sample 3, and operates the sample stage drive unit 7, etc., to repeatedly detect characteristic X-rays and secondary electrons. Then, if the acquisition of the intensity distribution for one element is completed, the same measurement is performed on other target elements. The element intensity calculation unit 90 acquires the intensity (concentration) of the target element for each micro-region on the sample 3, and the intensity data is stored in the data storage unit 91. Furthermore, when using an energy-dispersive X-ray spectrometer, the element intensity calculation unit 90 generates an X-ray spectrum for each micro-region within the two-dimensional region, detects the peak of a specific wavelength corresponding to the target element on the X-ray spectrum, and calculates the intensity (concentration) of the target element by determining the intensity of the peak.

[0044] When the analyst completes the measurement of all micro-regions within the two-dimensional region on sample 3 and performs the prescribed operation from operation unit 12, scatter plot generation unit 92 reads the prescribed data from data storage unit 91 and generates a scatter plot (binary scatter plot) representing the relationship between the intensities of the two elements. Each point on the scatter plot corresponds to a micro-region on sample 3. Therefore, for example, if the measurement is performed on 1000 micro-regions on sample 3, the number of points marked on the scatter plot is 1000.

[0045] The clustering analysis unit 93 performs clustering on the points on the generated scatter plot according to the prescribed algorithm, and adds a label to each point indicating which of one or more clusters it belongs to or whether it does not belong to any cluster.

[0046] Various methods of cluster analysis are known. Generally, in the clustering of points on such a scatter plot, clustering is performed using the distance between points. In scatter plots obtained through surface analysis such as EPMA, regions with extremely high point density and regions with low point density are frequently generated. In such cases, in regions with high point density, even if the distance between points is relatively short, clusters with a large number of points will form if they are not separated from other clusters. Conversely, in regions with low point density, even if the distance between points is relatively long, clusters with a very small number of points will form if they are not included in the same cluster. To address this situation, the density-based cluster analysis method based on hierarchical density estimation disclosed in Non-Patent Document 2 is adopted here. This method is an improvement on the general density-based cluster analysis disclosed in Non-Patent Document 1, and according to the present invention, it can cluster points in the scatter plot quite well.

[0047] exist Figure 4 and Figure 5 The image shows the results of clustering the scatter plot obtained through actual measurements using the method described above (where the cluster region boundaries described later are removed). In this example, nine clusters were extracted from the densely packed points.

[0048] The clustering result described above is the state of each point on the scatter map after a label representing a cluster has been attached to it; each cluster is simply a collection of points. Therefore, the cluster region detection unit 94 then calculates the region occupied by each cluster on the scatter map. The region corresponding to a cluster is a polygonal region with the number of vertices limited to a predetermined number.

[0049] Limiting the number of vertices simplifies subsequent operations and processing, such as deforming the shape of the cluster region or dividing or integrating the region. However, if the number of vertices is too small, the shape of the cluster region becomes too simple, resulting in a larger deviation from the region where the extracted points reside. Conversely, if the number of vertices is too large, the shape of the cluster region becomes too complex, making it impossible to achieve the simplification of the aforementioned deformation and other operations and processing. Therefore, in this device, an upper limit range for the number of vertices is predetermined, allowing the user (analyst) to select the upper limit within this range. Specifically, a slider is provided in the parameter setting window displayed on the screen, allowing the upper limit of the number of vertices to be changed by operating the slider.

[0050] Cluster regions in a scatter plot can be detected using either contour detection or convex hull detection. These two methods will be explained separately.

[0051] (1) Contour detection method

[0052] In this contour detection method, a set of points belonging to a cluster on a scatter map is treated as an object. Image processing is used to extract contours from the image of this object, and the extracted contours are approximated using polygons. Figure 2 Describe the processing procedure in detail.

[0053] First, the cluster region detection unit 94 generates a colored scatter map by assigning a different display color to each point extracted as a cluster on the scatter map (step S1). Next, the cluster region detection unit 94 extracts only the points of the display color of the cluster to be processed from the colored scatter map to obtain an extracted image (step S2).

[0054] The extracted image is a so-called point drawing. Therefore, the cluster region detection unit 94 assigns a predetermined size to each point on the extracted image and performs binarization on this basis to obtain a binarized image (step S3). However, assigning a predetermined size to each point on the extracted image is to connect the pixels corresponding to two points that are extremely close to each other on the extracted image. Therefore, if such pixels are originally connected in the extracted image, the above process can simply be binarization.

[0055] The aforementioned binarized image is an image where the object corresponding to the cluster of the processed object is white against a black background. For example, the cluster region detection unit 94 scans pixel by pixel, starting from the top left corner of the image, identifying the initially detected white pixel as the starting point of an object's outline. Then, starting from that starting point, it traces counter-clockwise the white pixels that connect to the black pixels. If it returns to the starting point pixel, the search for the object's outline ends. In this way, the cluster region detection unit 94 extracts the outline pixels that constitute the outline of an object (step S4). However, Figure 4 , Figure 5 The same applies to other objects; an object may not be entirely composed of white pixels, and some pixels on its inner side may sometimes be black. Therefore, conditions can be set during the search to exclude white pixels that are adjacent to regions of such isolated black pixels on the inner side of the object.

[0056] Through image processing in step S4, the contour line of the object corresponding to a cluster region is obtained. However, in most cases, this contour line is essentially close to a curve, and is a polygonal shape with a very large number of vertices. Therefore, the cluster region detection unit 94 performs approximate processing on the obtained contour line by using polygons with a predetermined number of vertices or less, thereby obtaining an approximate contour line. This processing can utilize the well-known Douglas-Peucker algorithm (see Non-Patent Documents 3 and 4) (step S5). Generally, the Douglas-Peucker algorithm is an algorithm that simplifies polyline segments with multiple points, but it is well known that in the case of polygons, the Douglas-Peucker algorithm can be extended to polygons by appropriately selecting two points from the vertices and setting them as the initial polyline segment.

[0057] Using the processing in step S5, an approximate contour line defining a cluster region can be obtained. This approximate contour line is represented by the coordinates of the vertices of the polygon representing the contour. In this case, the coordinates of the vertices are the positions of pixels on the scatter map image. Next, the cluster region detection unit 94 determines whether the processing of all clusters extracted from the scatter map is complete (step S6). If there are unprocessed clusters, it returns to step 2 and performs the processing of steps S2 to S6 on the next cluster to be processed.

[0058] By repeating the processing steps S2 to S6 a number of times corresponding to the number of extracted clusters, an approximate contour line can be obtained for all clusters on the scatter map. Then, the display processing unit 96 displays the approximate contour line as the cluster region boundary line on the scatter map (step S7).

[0059] exist Figure 4 The image shows the cluster region boundary line obtained using this contour detection method. From... Figure 4 It can be seen that, on the boundary line of the cluster region generated by this contour detection method, although there may be cases where points belonging to the cluster are located outside the boundary line, the range of the set of points belonging to the cluster is well defined.

[0060] (2)convex hull method

[0061] A convex hull is the smallest convex polygon that includes all the given points. As disclosed in Non-Patent Document 5, various algorithms for the convex hull method are known, but any algorithm can produce a polygon that includes all specified points. Figure 3 Describe the processing procedure in detail.

[0062] First, the cluster region detection unit 94 extracts all points belonging to a cluster as the processing target from all points on the scatter map, and obtains the position information (x, y coordinates on the scatter map image) of each point in the point set (step S11). Then, the cluster region detection unit 94 arranges the number of vertices to a predetermined number or less, and uses the convex hull method to calculate the convex hull polygon containing all points in the aforementioned point set (step S12).

[0063] Next, the cluster region detection unit 94 determines whether the processing of all clusters extracted from the scatter map is complete (step S13). If there are unprocessed clusters, it returns to step 11 and performs the processing of steps S11 to S13 on the next cluster to be processed. By repeating the processing of steps S11 to S13 a number of times corresponding to the number of extracted clusters, the convex hull polygon can be calculated for all clusters on the scatter map. In this case, the convex hull polygon is also represented by the coordinates of the vertices of the polygon. Then, the display processing unit 96 displays the convex hull polygon as the cluster region boundary line on the scatter map (step S14).

[0064] exist Figure 5 The diagram shows the cluster region boundary obtained using this convex hull method. Figure 5 It can be seen that, on the boundary line of the cluster region generated by the convex hull method, compared with the contour detection method, although it is easy to include points that do not belong to the cluster, it can delineate the cluster region without omitting points that belong to the cluster.

[0065] In the EPMA of this embodiment, polygonal cluster regions can be determined based on the results obtained by automatically clustering points in the scatter plot as described above. For example, when performing phase analysis, if the analyst deforms the shape of the cluster regions or integrates multiple cluster regions, the following processing is performed.

[0066] When the analyst performs the prescribed operation via the operation unit 11 for phase analysis, the display processing unit 96 displays a phase analysis screen on the display unit 12, showing a scatter plot and a distribution plot. As a distribution plot, it displays an SEM image of the measurement area on the sample 3 generated by the observation image generation unit 95 based on secondary electron intensity data. As a scatter plot, such as... Figure 4 or Figure 5 The diagram shows a scatter plot with the cluster region boundaries superimposed.

[0067] When an analyst wants to deform the shape of a cluster region, for example, the analyst uses operation unit 11 to specify the vertices of the cluster region boundary lines to be deformed on the scatter plot. Then, the analyst drags these vertices to move them to any position on the scatter plot. This deforms the shape of the cluster region boundary lines. Furthermore, when merging multiple cluster regions into one, for example, the analyst deletes the lines connecting two vertices on the boundary lines of each cluster region and instead adds new lines between the vertices of the different cluster region boundary lines. This allows for the merging of multiple cluster regions. Thus, operations such as moving, deforming, dividing, and merging cluster regions are very simple. Moreover, the processing corresponding to such operations essentially only involves changing the coordinates of the vertices of the polygonal cluster region boundary lines, therefore the computer processing is very simple and has a low load.

[0068] Furthermore, while the above-described embodiment is EPMA, this invention can be comprehensively applied to various analytical devices such as SEM and fluorescence X-ray analysis equipment, which can acquire signals reflecting the amount of elements or components (compounds, etc.) in many tiny regions within a one-dimensional or two-dimensional area on a sample. That is, this invention is not particularly limited to the measurement method or the analytical method itself, but can be applied to any analytical device capable of performing mapping analysis.

[0069] Furthermore, the above-described embodiments are merely one example of the present invention, and appropriate modifications, alterations, additions, etc., made within the scope of the spirit of the present invention are naturally included within the scope of the claims of this application.

[0070] [Various methods]

[0071] Those skilled in the art should recognize that the above-described embodiments are specific examples of the following methods.

[0072] (First item) One aspect of the surface analysis apparatus according to the present invention comprises:

[0073] The measuring unit acquires signals reflecting the amounts of multiple components or elements being analyzed at multiple locations on the sample.

[0074] The scatter plot generation unit generates a scatter plot based on the measurement results of the measurement unit;

[0075] Cluster analysis unit, which clusters the points in the scatter plot; and

[0076] The cluster region detection unit, based on the clustering information assigned to each point in the scatter map by the clustering analysis unit, calculates the boundary information of the cluster region, which is a polygon with a number of vertices of less than a specified number, for each cluster.

[0077] In the surface analysis apparatus described in the first item, it is possible to determine the regions occupied by each of multiple clusters automatically clustered in a scatter plot. The determined cluster regions have size and positional information in the scatter plot. Therefore, according to the apparatus described in the first item, graphical operations such as deformation, movement, segmentation, and integration of cluster regions, as well as corresponding processing, can be easily performed. This reduces the workload of the analyst during phase analysis and improves work efficiency. Furthermore, according to the apparatus described in the first item, it is also possible to calculate the area value, perimeter, and other values ​​of each cluster region and compare these values ​​between different clusters.

[0078] (Second item) In the surface analysis apparatus described in the first item, it can be configured such that: the cluster region detection unit extracts all points belonging to a cluster from the scatter map, obtains the outline of an object corresponding to the set by image processing of a binary image displaying the set of points, and approximates the outline using a polygon with a number of vertices less than a specified number, thereby obtaining the cluster region boundary information.

[0079] (Third item) In the surface analysis apparatus described in the second item, the cluster region detection unit is able to approximate polygons using the Douglas-Puk algorithm.

[0080] According to the surface analysis apparatus described in the second and third items, it is possible to obtain cluster region boundary information that includes most points belonging to a cluster and almost none points not belonging to the cluster.

[0081] (Fourth item) In addition, in the surface analysis apparatus described in the first item, it can be configured such that: the cluster region detection unit extracts all points belonging to a cluster in the scatter map, and uses the position information of the point in the scatter map to calculate the convex hull polygon containing all the points, thereby obtaining the cluster region boundary information.

[0082] According to the surface analysis apparatus described in the fourth item, it is possible to obtain cluster region boundary information that includes all points belonging to a cluster without omission.

[0083] (Fifth item) In addition, the surface analysis apparatus described in any one of the first to fourth items can be configured to further include a display processing unit that overlays cluster region boundary information obtained for each cluster on the scatter map onto the scatter map, and accepts operations of selecting vertices of cluster region boundary information on the scatter map and moving the selected vertices to deform the cluster region boundary information.

[0084] As described above, the boundary information of a polygonal cluster region is essentially composed of the position information of the vertices of the polygon, and the movement of the vertices in the cluster region boundary information is merely a change in position information. Therefore, according to the surface analysis apparatus described in item five, not only are operations such as deformation, movement, segmentation, and integration of the cluster region easy, but the corresponding processing within the apparatus is also very simple.

Claims

1. A surface analysis device, comprising: The measuring unit acquires signals reflecting the amounts of multiple components or elements being analyzed at multiple locations on the sample. The scatter plot generation unit generates a scatter plot based on the measurement results of the measurement unit; The clustering analysis unit clusters the points in the scatter plot; The cluster region detection unit, based on the clustering information assigned to each point in the scatter map by the clustering analysis unit, calculates the boundary information of the cluster region in polygonal shape with the number of vertices in each cluster being less than a specified number. as well as The display processing unit overlays the cluster region boundary information obtained for each cluster on the scatter map onto the scatter map, and accepts operations to select vertices of the cluster region boundary information on the scatter map and move the selected vertices to arbitrary positions to deform the cluster region boundary information.

2. The surface analysis apparatus according to claim 1, wherein, The cluster region detection unit extracts all points belonging to a cluster from the scatter map, obtains the outline of an object corresponding to the set from the binary image displaying the set of points, and approximates the outline with a polygon with a number of vertices less than a specified number, thereby obtaining the cluster region boundary information.

3. The surface analysis apparatus according to claim 2, wherein, The cluster region detection unit uses the Douglas-Puk algorithm to approximate polygons.

4. The surface analysis apparatus according to claim 1, wherein, The cluster region detection unit extracts all points belonging to a cluster from the scatter map and uses the position information of the points in the scatter map to calculate the convex hull polygon containing all the points, thereby obtaining the cluster region boundary information.

Images