CN112116581A - Method and apparatus for acquiring atomic position in atomic imaging - Google Patents
Method and apparatus for acquiring atomic position in atomic imaging Download PDFInfo
- Publication number
- CN112116581A CN112116581A CN202011008851.2A CN202011008851A CN112116581A CN 112116581 A CN112116581 A CN 112116581A CN 202011008851 A CN202011008851 A CN 202011008851A CN 112116581 A CN112116581 A CN 112116581A
- Authority
- CN
- China
- Prior art keywords
- image
- bright
- function
- obtaining
- fitting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
- G06T2207/10061—Microscopic image from scanning electron microscope
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
- Image Processing (AREA)
Abstract
The invention provides a transmission electron microscope image analysis method, which comprises the following steps: an input step of receiving an original image of a material having image pixel values; a bright atom bit shape rough extraction step, which is used for obtaining the outline data of bright spots in the original image according to the pixel value of the image; a bright atom position and shape fine calculation step, wherein pixel values of a region containing bright spots in the original image are extracted according to the contour data, and the accurate position of the central point of the bright atom is obtained through fitting; a step of fine calculation of the position and shape of other atoms, which obtains the central positions of other atoms according to the central positions of a plurality of adjacent bright atoms and the lattice type of the material; and an output step of outputting the atom arrangement information in the original image.
Description
Technical Field
The present invention relates to the field of image recognition of microscopic particles, and more particularly, to a method and apparatus for obtaining atomic positions in atomic imaging.
Background
The electron scanning transmission electron microscope STEM can distinguish atoms with the size of less than 1nm, and is an important means for modern material characterization. The high-angle annular dark field image HAADF image of the electronic scanning transmission electron microscope has the characteristics of high magnification and capability of directly recording the real morphology of atoms with certain resolution. However, generally, the recorded atomic morphology is not the entire atomic morphology, and only part of the atomic profiles corresponding to the bright spots with higher brightness values in the HAADF image can be resolved. In addition, a large amount of HAADF images taken of the target material require determination of the precise configuration of atoms therein, which results in a large amount of work.
In contrast, the prior art "precise automatic determination of atomic peak position in high resolution atomic image and rapid visualization of electric polarization domain" (south tiger et al, report on electron microscopy, vol. 35, pp. 191-200, 2016, 03) provides a method for rapidly locating the position of an atomic peak in a high resolution atomic image, which comprises estimating a rough angular point distribution configuration by angular point detection, then measuring a vector value corresponding to the position of a nearest neighboring angular point as a basis vector, constructing an evaluation function to calculate an optimal basis vector, thereby estimating a rough bragg configuration, then fitting and correcting each possible atomic point, and estimating other possible neighboring atoms by using a corrected atomic position. The method has the defects that due to algorithm limitation, the atomic resolution precision is not high, and the application range is narrow. Specifically, the lattice type, size and direction of a target material are determined to calculate a basis vector, and an evaluation function is used to obtain an optimal basis vector and calculate a predicted atomic position; secondly, the fitting function used for fitting and correcting assumes the atoms as ellipses, so that it is difficult to improve the accuracy. This algorithm does not provide a usable solution for defective or irregularly arranged bravae lattices. Meanwhile, the parameters related to the crystal lattice need to be preset, the operation is complex, and manual operation cannot be separated.
Disclosure of Invention
In view of the above, the present application proposes an analysis method for determining atomic positions in scanning transmission electron microscope imaging, comprising:
acquiring a transmission electron microscope original image of the material;
obtaining contour data of the bright spots in the original image according to the image pixel value of each image pixel of the original image, wherein the contour data defines the outer boundary and the central point position of the contour data;
extracting the pixel values of the area containing the bright spots in the original image according to the positions of the outer boundary and the central point in the contour data,
and fitting to obtain the accurate position of the central point of the bright spot, and taking the accurate position as the central position of the bright atom corresponding to the bright spot.
Preferably, the method further comprises a preprocessing step of the original image for obtaining a sharp outline of the bright spot, wherein the preprocessing step comprises:
according to an equalization function, carrying out contrast operation on the original image and obtaining an equalized image;
performing convolution operation on the equalized image according to a fuzzy function to obtain a smooth image;
according to a self-adaptive threshold segmentation method, performing threshold segmentation operation on the smooth image and obtaining a threshold segmentation image; and
and according to an open operation function, carrying out corrosion operation and expansion operation on the threshold segmentation image and obtaining a preprocessed original image.
Preferably, wherein
The equalization function adopts a self-adaptive equalization function capable of limiting contrast, wherein the pixel value of a target pixel point is calculated and updated by the self-adaptive equalization function, and the updating is determined by the pixel values of the target pixel point and the pixel points adjacent to the target pixel point.
Preferably, wherein
The equalization function employs one of: gamma transform function, linear transform function, histogram normalization function.
Preferably, the fuzzy function uses one of the following: gaussian blur function, mean filter function, median filter function, custom filter function.
Preferably, wherein
The process of obtaining the contour data of the bright spots in the original image further comprises: and setting a screening threshold value according to the area distribution of the contour so as to reject contour data with larger deviation from the center of the area distribution of the contour.
Preferably, wherein
The fitting includes: setting a fitting model as a circular Gaussian function, taking the pixel values in the area containing the bright spots as data, and obtaining undetermined parameters of the fitting model by using the fitting function, wherein the circular Gaussian function is as follows:
wherein x0And y0For the accurate position of the central point of the bright spot to be fitted, Amp is the amplitude of a Gaussian function, sigma is the area parameter of the bright spot, and offset is the translation parameter of the central point of the bright spot in the original image.
Preferably, the fitting function uses a least squares curve fitting method.
Preferably, the method further comprises:
and obtaining the central positions of other atoms according to the central positions of a plurality of adjacent bright atoms and the lattice type of the material.
Preferably, the other atomic bit shape fine calculation step includes:
establishing a kd-tree data structure and establishing mapping of each bright atom and the bright atoms adjacent to the bright atom;
calculating estimated central point positions of other atoms among the plurality of neighboring bright atoms according to the lattice structure relationship of the material and the accurate positions of the central points of the plurality of neighboring bright atoms;
extracting pixel values of corresponding areas in the original image according to the estimated central point position, and fitting to obtain the accurate central point positions of the other atoms, wherein the fitting comprises the following steps:
setting a fitting model as a circular Gaussian function, taking the pixel values in the region as data, and obtaining undetermined parameters of the fitting model by using the fitting function, wherein the circular Gaussian function is as follows:
wherein x0And y0For the accurate position of the central point of the bright spot to be fitted, Amp is the amplitude of a Gaussian function, sigma is the area parameter of the bright spot, and offset is the translation parameter of the central point of the bright spot in the original image.
Another aspect of the present invention provides a transmission electron microscope image analysis apparatus, including a processing circuit, configured to obtain atom arrangement information of a material according to an original image of the material in a transmission electron microscope, where the image analysis process includes:
acquiring a transmission electron microscope original image of the material;
obtaining contour data of the bright spots in the original image according to the image pixel value of each image pixel of the original image, wherein the contour data defines the outer boundary and the central point position of the contour data;
extracting the pixel values of the area containing the bright spots in the original image according to the positions of the outer boundary and the central point in the contour data,
and fitting to obtain the accurate position of the central point of the bright spot, and taking the accurate position as the central position of the bright atom corresponding to the bright spot.
In summary, according to some embodiments of the present application, the method and apparatus for obtaining the atomic position in atomic imaging can determine the central position and the outline of each atom according to each image pixel in the original image, the calculation accuracy is higher, the calculation efficiency is increased due to a better algorithm, and the accurate atomic position in atomic imaging can be obtained in batch without manual intervention.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention. Obviously, the figures in the following description are only some embodiments of the invention, in which:
FIG. 1 illustrates a flow diagram of a method of determining atomic positions in scanning transmission electron microscope imaging according to one embodiment of the present application;
FIG. 2 illustrates a high angle annular dark field image HAADF map of a scanning transmission electron microscope in accordance with one embodiment of the present application;
FIG. 3 shows a flow diagram of sub-steps of the image pre-processing step (step S120) according to one embodiment of the present application;
fig. 4 is a schematic diagram of an HAADF image after being subjected to the local contrast equalization process of step S121 according to an embodiment of the present application;
FIG. 5 is a schematic diagram of an HAADF image after local contrast equalization processing and blurring/smoothing processing according to an embodiment of the present application;
FIG. 6 is a diagram illustrating a blurred/smoothed HAADF image after a thresholding process according to an embodiment of the present application;
fig. 7 illustrates an exemplary graph of HAADF pre-processed graph data output after being subjected to a topographical operation according to one embodiment of the present application. (ii) a
FIG. 8 is a flowchart illustrating the bright atom bit shape rough extraction step (step S130) according to an embodiment of the present invention;
FIG. 9 is a schematic diagram illustrating a bit pattern after extracting light atom contours from preprocessed HAADF map data that has been topographically manipulated according to one embodiment of the present application;
FIG. 10 is a flowchart illustrating the bright atom bit shape refinement calculation step (step S140) according to an embodiment of the present application;
FIG. 11 is a schematic diagram illustrating regions of light atoms and surrounding pixel values extracted as a fitting target in an original HAADF map according to one embodiment of the present application;
FIG. 12 is a diagram illustrating the precise configuration of bright atoms obtained in the fine calculation step (step S140) according to some embodiments of the present application;
FIG. 13 is a flowchart showing the other atomic bit pattern refinement calculation step (step S150) according to an embodiment of the present invention;
FIG. 14 is a diagram illustrating the exact configuration of the dark atom, which is the other atom, obtained in the fine configuration calculation step (step S150) according to some embodiments of the present application;
FIG. 15 illustrates domain polarization diagrams obtained in accordance with further embodiments of the present application;
FIG. 16 shows a schematic representation of the cell configuration in a crystal lattice obtained according to further embodiments of the present application.
Detailed Description
The application relates to a method and a device for judging the atom position in the imaging of a scanning transmission electron microscope. While several preferred modes of carrying out the application have been described in the specification, it is to be understood that the application can be carried out in numerous ways, and should not be limited to the specific embodiments described below or to the specific ways in which the features described below can be carried out. In other instances, well-known details will not be set forth or discussed in order to avoid obscuring the present application.
FIG. 1 shows a flow diagram of a method of determining atomic positions in scanning transmission electron microscope imaging according to one embodiment of the present application. In summary, it comprises the following steps: an input step of obtaining an original image (step S110); an image preprocessing step (step S120); a bright atom bit shape rough extraction step (step S130); a bright atom bit shape fine calculation step (step S140); a step of fine calculation of the other atomic configuration (step S150); and an output step (step S160) of finally merging and outputting all the atom precision configuration parameters.
FIG. 2 shows an exemplary view of a high angle annular dark field image STEM HAADF of a scanning transmission electron microscope.
The above steps of the present invention will be described in detail with reference to fig. 2 as an example.
In the input step (step S110), the surface of the target material is photographed by using the electron scanning transmission electron microscope STEM, and a high-angle annular dark field image STEM HAADF of the scanning transmission electron microscope shown in fig. 2, which is hereinafter referred to as HAADF image, is obtained. As shown in fig. 2, the acquired HAADF image is a grayscale image, and the pixel information includes only the black and white luminance information of the pixel point, without distinguishing the ratio of other color information such as RGB components. But color maps may also be converted to grayscale maps by the built-in function cvtColor in OpenCV to implement the steps of the invention, according to other embodiments of the invention. As can be seen from fig. 2, the outline of the atoms in the original data cannot be clearly distinguished, and the light and shade change occurs at intervals, so that the contrast of each atom in the whole image is not uniform relative to the background, and it is difficult to quickly distinguish and identify.
An image preprocessing step (step S120) for improving the contrast of the bright spot profile to be measured in the HAADF image shown in fig. 2 with respect to the surrounding background, separating the bright spot profile of the HAADF picture with brightness change, eliminating noise points in the picture, and simultaneously making the profile of the bright atomic bright spots more regular.
The process of preprocessing the HAADF map shown in fig. 2 according to the first embodiment of the present invention will be described below with reference to fig. 3 to 7. Fig. 3 shows a flowchart of the image preprocessing step S120 according to an embodiment of the present invention, including:
a step S121 of performing local contrast equalization processing on each pixel value of the HAADF map; according to one embodiment of the invention, the data processing process is implemented based on the python environment and the called scipy library and OpenCV library, and the HAADF graph is used as the data to be processed. And calling a contrast-limiting adaptive histogram equalization function createCLAHE in the OpenCV library to adjust the contrast of the HAADF image data and output the data of the equalization image, wherein the adaptive equalization function obtains the equalization pixel value of the target pixel point according to the pixel value calculation of the target pixel point and the pixel values of the adjacent pixel points of the target pixel point and updates the equalization pixel value to the target pixel point. In one embodiment, the parameters that need to be set in the createclhe function mainly include clipLimit ═ 2.0 and tileGridSize ═ 8, 8. Fig. 4 shows a schematic diagram of the HAADF image after the equalization process of the local contrast in step S121.
Step S122, carrying out fuzzy/smoothing processing on the HAADF image after the equalization processing of the local contrast; according to an embodiment of the invention, a gaussian fuzzy function gaussian in an OpenCV library is called, a convolution kernel with a certain size and normal distribution characteristics is set, a convolution operation is performed on a data file of an HAADF graph and the gaussian kernel to complete gaussian smoothing, and the HAADF graph data after the fuzzy/smoothing processing is output. The parameters to be set are convolution kernel and sigma value, where the convolution kernel should be smaller than the size of atom but larger than the size of noise, and the sigma should match the convolution kernel, and in one embodiment, the convolution kernel size is 25 × 25, and the sigma is 9.
Fig. 5 shows a schematic diagram of the HAADF image after the equalization processing of the local contrast after the blurring/smoothing processing of step S122.
In step S123, the blurred/smoothed HAADF image data is subjected to binarization division processing. And calling an adaptive thresholding function, wherein the adaptive thresholding function preferably sets the parameter C to 0.8, to calculate the equalized HAADF image data and output the data of the binarized HAADF image. The self-adaptive threshold segmentation method comprises the steps of selecting adjacent areas with a certain size for coordinate points to be calculated of an equilibrium graph, calculating gray value distribution of all the coordinate points in the areas to determine a binarization threshold value at the position of the coordinate points to be calculated, judging a comparison result of the gray value of the point and the binarization threshold value, updating the gray value of the point to be 0 if the gray value of the point is smaller than the binarization threshold value, and otherwise updating the gray value of the point to be 255.
Fig. 6 shows a schematic diagram of the HAADF image after the blurring/smoothing process of step S123 after the binarization dividing process.
And step S124, performing topographic operation processing on the binary image. Wherein the topography operation adopts an opening operation, the opening operation firstly executes a corrosion algorithm and then executes an expansion algorithm. Wherein the erosion is a process of eliminating boundary points and shrinking the boundary inwards. Can be used to eliminate small and meaningless objects, the erosion algorithm is: scanning each pixel of the image by using a 5 multiplied by 5 circular structural element, and performing AND operation by using the structural element and the binary image covered by the structural element, wherein if the structural element and the binary image are both 1, the pixel of the image is 1; otherwise it is 0. The erosion results can reduce the binary image by one turn.
Dilation is the process of merging all background points in contact with an object into the object, expanding the boundary outward. Can be used to fill in voids in objects. According to one embodiment of the invention, the inflation algorithm is: scanning each pixel of the image by using a 5 multiplied by 5 circular structural element, and performing AND operation by using the structural element and the binary image covered by the structural element, wherein if the structural element and the binary image are both 0, the pixel of the image is 0; otherwise it is 1. The expansion results in a one-turn expansion of the binary image.
In the topography calculation step, on calculation is performed a plurality of times on the pixel values (gray values) of the entire binarized HAADF image. And updating the original pixel value in the binarized HAADF image into a new pixel value after multiple opening operations, and outputting HAADF preprocessed image data.
Fig. 7 shows an example of HAADF pre-processing map data output after the topographical operation of step S124.
Fig. 8 shows a flowchart of the bright atom shape rough extraction step (step S130) according to an embodiment of the present invention, which is used to quickly calculate the rough positions of the shape parameters such as the bright atom outline and the center coordinate.
According to an embodiment of the present invention, the bright atom bit shape rough extraction step S130 includes:
and S131, extracting the bright spots, namely the bright atom contours in the HAADF preprocessed image data after the topographic operation by adopting a contour extraction function. According to an embodiment of the present invention, step S131 includes calling a contour extraction function, findContours function, in the OpenCV library to operate on HAADF preprocessed map data, where the contour approximation method needs to be set to CHAIN _ APPROX _ SIMPLE. The function is used for performing traversal search on the pixel value (gray value) of each pixel point of the HAADF preprocessed graph, determining the boundary contour corresponding to each bright spot and generating bright spot contour graph data, the bright spot contour graph data generated by using the findContours function mainly records the corresponding vertical, horizontal and diagonal end point values of each contour, and the shape and position of the outline of the bright atom corresponding to each bright spot can be determined quickly and effectively.
Step S132 is to screen the generated bright spot profile data. According to one embodiment of the present invention, the specific rule is: firstly, calculating the area values of all the bright spot profiles in the bright spot profile map data, sorting according to the area values from large to small, selecting the median or the ranking order value selected according to a certain distribution proportion, for example, selecting the profile area value sorted at the 500 th position, taking the value of 1/2 as an area screening reference value, filtering the areas of all the bright spot profiles, and removing the bright spot profile data with the area smaller than the area screening reference value. And then, calculating each piece of bright spot profile data in the screened bright spot profile image, obtaining the central point position data of the screened bright spot profile image, and taking the set of the central point position data of each bright spot profile as crude extraction data of the bright atom configuration.
Fig. 9 shows a schematic diagram of the morphology of the bright spots, i.e., the bright atom contours in the preprocessed HAADF image data after the topographic operation is performed, after step S130.
Fig. 10 shows a flowchart of the fine calculation step (step S140) of the bright atom configuration according to an embodiment of the present invention, which is used to further precisely fit the bright atom configuration parameters, such as the coordinates of the center position, obtained in the previous step to obtain the accurate configuration parameters.
In some embodiments, the bright atom bit shape fine calculation step (step S140) includes step S141 and step 142.
Step S141, using the central point in the raw extraction data of the bright atomic configuration as a reference point, extracting the corresponding position point in the original unprocessed HAADF image and the pixel points in a certain area around the corresponding position point. According to an embodiment of the invention, firstly, a SciPy library is called to read each central point position in the light atom bitmap rough extraction data in the previous step, and simultaneously, a high-angle annular dark field image HAADF image as shown in fig. 2 is read; then, the central point in the light atom bitmap rough extraction data is used as a reference point, the position coordinate corresponding to the reference point is found in the original unprocessed HAADF image, the coordinate is used as the light atom central point fitting initial value (namely the assumed possible position of the light atom center), a rectangular area with 20-by-20 pixels is defined by taking the coordinate as the center, and all pixel point data in the area are extracted.
Fig. 11 shows a schematic diagram of a region of bright atoms and peripheral pixel values extracted as a fitting target in the original HAADF map.
Step 142, first define a circular (two-dimensional) gaussian function:
in the function, x0And y0Is the exact position of the center of the bright atom to be fitted; amp is the amplitude of the gaussian function, i.e. the magnitude of the peak; σ (i.e., sigma) is related to the size of the circle; offset represents the amount of up and down translation of the position of the central point value of the bright spot in the image, and e is a natural constant. Then, calling a curve fitting function under an option module in the SciPy library to perform curve fitting, wherein parameters to be transferred by the curve fitting function include: the initial value is a bright atom central point fitting initial value, the fitting target is the aforementioned circular (two-dimensional) gaussian function, the data used for fitting is all pixel values in the aforementioned rectangular region of 20 × 20 pixels, and the fitting method is a least square method. And then, according to the relative position relationship between the local graph and the whole graph (namely, the relative coordinates of the outline center is added), the accurate position of the atom center in the original graph can be obtained. Specifically, since the range to be fitted is a rectangular area of 20 × 20 pixels at a certain determined position on the HAADF map, and the relative coordinates of the target light atom in the rectangular area obtained by calculation in this step are also determined, the relative coordinates can be converted into coordinate values of the target light atom on the whole HAADF map, for example, the vector from the end point of the lower left corner of the HAADF map to the center point of the target rectangular area and the center of the target rectangular area are taken as the origin, and the vector from the end point of the lower left corner of the HAADF map to the center point of the target rectangular area and the center of the target rectangularThe sum of the vectors of the coordinates of the target bright atom is the coordinates of the target bright atom relative to the origin.
The final fitting result of step 142 is the accurate position data of the center point of the light atom in the HAADF image of the high-angle annular dark field image, which identifies and encompasses the accurate positions of the center points of all the light atoms in the HAADF image.
Fig. 12 shows a bright atom exact configuration diagram obtained through the bright atom configuration fine calculation step (step S140).
Fig. 13 shows a flowchart of the other atomic bit pattern refinement calculation step (step S150) according to an embodiment of the present invention.
As shown in fig. 13, in some embodiments, the other atomic bit shape fine calculation step (step S150) includes: step S151, establishing corresponding data indexes for any bright atom and the adjacent bright atoms in the HAADF image, so that the adjacent bright atoms can be quickly inquired for a specific bright atom; step S152, determining the atom arrangement relationship in the lattice structure corresponding to the type of material, and predicting the possible positions of atoms other than the recorded bright atoms based on the relationship.
According to an embodiment of the present invention, step S151 includes, first, in some embodiments, creating a kd-tree data structure recording the precise positions of the light atom points in the HAADF graph as an index, so that each light atom can conveniently query the data of the light atom adjacent to the light atom. Among them, a Kd-tree (short for k-dimensional tree) is a data structure for dividing a k-dimensional data space. The method is mainly applied to searching of multidimensional space key data (such as range searching and nearest neighbor searching). kd-Tree is a special case of binary spatial partitioning tree.
Step S152 includes determining the atom arrangement within the lattice structure as the presence of a dark atom at the center point of every four light atoms, for example; then, finding four neighboring bright atoms by using a kd-tree data structure, reading position data of the four neighboring bright atoms, calculating to obtain the position of a central point between the four neighboring bright atoms, and using the position as a central point fitting initial value of other atoms or dark atoms between the four neighboring bright atoms; then, a rectangular area with 20-by-20 pixels is defined for the center, and all pixel data in the area are extracted; then, the curve fitting is carried out by calling the curve _ fit function under the option module in the SciPy library, and the parameters to be transferred by the curve _ fit function are as follows: the initial value is a bright atom central point fitting initial value, the fitting target is a circular (two-dimensional) Gaussian function used in the bright atom bit shape fine calculation step, the data used for fitting is pixel values in the rectangular area of the 20 x 20 pixels, and the fitting method is a least square method. The fitting result obtained in step S152 is the atom center point accurate position data of other atoms (i.e., dark atoms) in the high-angle annular dark field image HAADF image.
Therefore, in the whole-atom precise configuration parameter obtaining step (step S160), the light-atom central point precise position data obtained in the light-atom configuration fine calculation step (step S140) and the dark-atom central point precise position data obtained in the other-atom configuration fine calculation step (step S150) are combined and output as whole-atom precise configuration data. The data is the accurate position data of all atoms in the originally obtained high-angle annular dark field image HAADF image, namely, the accurate automatic judgment of the position of the atomic peak in the high-resolution atomic image is completed.
Fig. 14 shows an accurate bit pattern diagram of the other atoms, i.e., dark atoms, obtained through the other atom bit pattern fine calculation step (step S150).
In the image preprocessing step, the pixels of the image are sorted by using the adaptive equalization function on the HAADF image, so that the pixel contrast at each coordinate in the HAADF image is more uniform, and the image is clearer. In the equalization processing step according to other embodiments of the present invention, if the overall image brightness is uniform, the following contrast enhancement method can be considered: gamma transformation, linear transformation, histogram normalization. Among them, gamma conversion is preferable.
In the image preprocessing step, blurring/smoothing is performed on each pixel value of the HAADF image to reduce the image noise point. This step is added because the original HAADF map has certain disadvantages in that the light partial areas each representing the shape of a light atom are small, irregular in shape, unable to reflect the true atom boundaries, and in the presence of periodic high-contrast/low-contrast interlacing features. In the blurring/smoothing operation according to other embodiments of the present invention, it may be also considered preferable to use methods such as mean filtering, median filtering, bilateral filtering, custom filtering, guided filtering, and the like.
In the image preprocessing step, the operation of enhancing the contrast by the adaptive equalization process is preferably placed before the operation of smoothing the image, since the contrast enhancement may introduce part of the noise, and gaussian smoothing may remove the noise.
In the image preprocessing step, the data of the HAADF image is subjected to binarization division processing, so that a target area (light atom part) and a background (non-light atom part) in the HAADF image can be separated on the data, which can effectively separate the HAADF picture with brightness change.
In the image preprocessing step, the HAADF image after the binarization processing is subjected to topographic operation, bright spots with small areas can be removed through multiple corrosion, and the remaining bright spots can be restored to the original sizes through expansion, but the boundaries can be smoother. Noise points in the picture can be effectively eliminated, and meanwhile, the outline of the bright spot corresponding to the possible bright atom position is more regular.
In the step of crude extraction of the bright atom position and shape, small-area bright spot profile data with a bright spot profile obviously lower than the range of most normal bright atom bright spot profiles needs to be removed to increase the accuracy degree of subsequent fitting, so that a screening threshold can be set for statistical data of the size distribution of the bright atom profile in the target HAADF image, and profile data with a larger deviation compared with a distribution center value is removed according to the area distribution. For example, in other embodiments, the area values covering the last third of the distribution may be sorted and removed from the distribution according to the order from the large to the small, or the profile data of the larger area with a value closer to one of the total occupied amounts [ 50% -90% ] in the distribution center may be left according to the distribution statistics.
In the step of fine calculation of the bit shape of the bright atoms, the present invention uses a circular model, which is advantageous in the problem of relative position, compared to the elliptical model of the comparison document 1. Firstly, the circular model reduces the fitting parameters, so that the fitting speed is improved; secondly, the circular model has few parameters to effectively avoid the occurrence of overfitting, and the elliptical model has more parameters to easily cause error of results due to the fact that the fitted model is too complex; and finally, the circular model is adopted, so that random errors caused by the random drift of complex atoms in the electron microscope and the aberration of the electron microscope can be partially eliminated.
In other atom bit shape fine calculation steps, by establishing a kd-tree data structure, the related data of the neighbor atoms of the target bright atom can be quickly retrieved, so that traversal search in the whole image data is avoided, and the calculation time is greatly saved.
According to the embodiment of the invention, the outline and the central position of the bright atom can be more accurately distinguished in the image preprocessing step, so that the fitting accuracy is greatly improved in the subsequent accurate fitting process, fitting divergence and large deviation caused by early-stage errors are avoided, and the time occupied by fitting calculation is remarkably reduced.
The accuracy of the fit to the precise location of the atom in embodiments according to the invention may be verified from an evaluation of the convergence of the fit. Generally speaking, free convergence means that a fitting result is irrelevant to a boundary condition, which indicates that an initial value of fitting is more reasonable, parameter values obtained by all fitting are reasonable, and no artificial interference exists, meaning that the fitting is accurate; conversely, if a certain parameter of the fitting reaches a set boundary condition and is forced to converge, the fitting result may be inaccurate. The rate of convergence of the gaussian curve in an embodiment in accordance with the invention. For most clear electron microscope pictures, the convergence rate is as high as over 99.9%, even up to 100%, which shows that the method belongs to free convergence and has very high accuracy.
In another embodiment according to the present invention, the atomic peak fine positioning method of the previous embodiment can be applied to the process of obtaining the polarized electric dipole ferroelectric domain structure of the ferroelectric material from the original HAADF image of the ferroelectric material.
Fig. 15 shows domain polarization schemes obtained according to further embodiments of the present invention. The polarized electric dipole ferroelectric domain structure of the ferroelectric material is that the polarized electric dipoles in the ferroelectric material can spontaneously rotate under certain boundary conditions to form a head-to-tail connected vortex-shaped ferroelectric domain structure. The magnetic domain size can be less than a few nanometers, while the average protocell polarization length is from tens of picometers to hundreds of picometers. The ferroelectric polarization topological domain can stably exist at room temperature, and is expected to be used as an information unit carrier of a next-generation high-density storage device to realize ultrahigh-density storage. The method obtains a ferroelectric topological domain structure with high stability and high uniformity, researches the distribution of polarization intensity in a single ferroelectric vortex domain, further realizes the control and the read-write of an external field on the ferroelectric topological domain structure, and is the key for realizing the application of a new generation of high-density ferroelectric topological device.
The actual position of the atom in the original HAADF diagram can be analyzed for error from its ideal position in the ideal lattice, so as to obtain an actual atom position shift vector diagram or a polarization diagram representing the polarization effect of the atom distribution.
Specifically, in the HAADF diagram shown in fig. 2, for example, each square lattice contains four positive ions and one negative ion, the coordinates of the four negative ions are averaged, and the difference in position from the positive ions is the electric polarization vector, i.e., the degree to which the positive center and the negative center do not coincide. And amplifying the polarization vector by dozens of times, and then drawing the polarization vector at a corresponding position of the original image to obtain a polarization distribution map. Similarly, a polarization distribution diagram can also be obtained by taking four negative ions and one positive ion as a crystal lattice.
By combining the invention with a spherical aberration electron microscope observation technology, we visually observe a ferroelectric polarization vortex domain array structure which has the size of 4nm and stably exists at room temperature in a PbTiO3 layer in a modulated PbTiO3/SrTiO3 superlattice system, and the ferroelectric polarization vortex domain array structure is the result of the coupling effect of medium elastic performance, electrostatic energy and gradient energy. And can further incorporate an Integrated Differential Phase Contrast (iDPC) image of a spherical aberration corrected transmission electron microscope to reveal the atomic structure of a single ferroelectric topological vortex domain and to accurately measure the distribution of its polarization intensity. The extensive research of the research related to the above reveals some basic information of ferroelectric topological vortex, provides structural basis for understanding the domain physical properties of ferroelectric vortex and corresponding calculation simulation, and is also helpful for understanding complex lattice-charge coupling phenomena of ferroelectric, piezoelectric, flexoelectric and the like of ferroelectric thin film on a nanometer scale.
In another embodiment according to the present invention, the cell size and configuration arrangement characteristics corresponding to the lattice can be further counted by calculating the actual position of the atom in the original HAADF graph. FIG. 16 is a schematic diagram showing the cell configuration in a lattice obtained according to further embodiments of the present invention, wherein the data shown is raw data without interpolation. Specifically, the distance between an atom and the nearest neighboring four homologous atoms in the HAADF graph is calculated for each target lattice and averaged, which is approximately equal to the size of the primitive cell. It can be seen that the method can effectively and comprehensively identify the appearance of each cell and show the configuration arrangement characteristics of the whole cell, and the cell configuration is basically consistent with the configuration disclosed in the prior art, which also proves that the method has higher accuracy.
According to another embodiment of the present invention, a system for determining the position of an atom in scanning transmission electron microscope imaging is disclosed, which comprises a processor, a storage and a storage, wherein the storage is stored on the processor, and the steps of any one of the methods for determining the position of an atom in scanning transmission electron microscope imaging in the foregoing embodiments can be implemented on the processor.
According to yet another embodiment of the present invention, a computer-readable storage medium is disclosed, on which a computer program is stored, wherein the program is executed by a processor to implement any of the steps of the method for determining the position of an atom in scanning electron microscope imaging in the foregoing embodiments.
Although the block diagrams describe steps/components in a functionally separate manner, such description is for illustrative purposes only. Those skilled in the art will be specifically aware of: the steps described above may be performed in another sequence, and components may be arbitrarily combined or separated into separate software, firmware, and/or hardware components without departing from the object and scope of the present invention. Moreover, regardless of how such components are combined or divided, they may execute on the same computing device or multiple computing devices, which may be connected by one or more networks.
Although the present invention has been described with reference to preferred embodiments, it is not intended to be limited thereto, and any simple modifications in steps that do not depart from the spirit of the present invention are intended to be included within the scope of the present invention. It is intended that all matter contained in the above description and shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.
Claims (12)
1. A transmission electron microscope image analysis method, comprising:
acquiring a transmission electron microscope original image of the material;
obtaining contour data of the bright spots in the original image according to the image pixel value of each image pixel of the original image, wherein the contour data defines the outer boundary and the central point position of the contour data;
acquiring the pixel value of the area containing the bright spots in the original image according to the position of the outer boundary and the central point in the contour data,
and fitting to obtain the accurate position of the central point of the bright spot, and obtaining the central position of the bright atom corresponding to the bright spot.
2. The method according to claim 1, further comprising a pre-processing step of the original image for obtaining sharp outlines of the bright spots, wherein the pre-processing step comprises:
carrying out contrast operation on the original image and obtaining a balanced image;
performing convolution operation on the equalized image and obtaining a smooth image;
performing threshold segmentation operation on the smooth image and obtaining a threshold segmentation image; and
and carrying out corrosion operation and expansion operation on the threshold segmentation image and obtaining a preprocessed original image.
3. A method as claimed in claim 2, wherein
In the step of carrying out contrast operation on the original image and obtaining a balanced image, a self-adaptive balancing function capable of limiting the contrast is adopted for balancing, wherein the pixel value of a target pixel point is operated and updated by the self-adaptive balancing function, and the updating is determined by the pixel values of the target pixel point and the adjacent pixel points.
4. The method of claim 2, wherein
In the step of performing a contrast operation on the original image and obtaining an equalized image, one of the following functions is used: gamma transform function, linear transform function, histogram normalization function.
5. The method of claim 2, wherein one of the following functions is employed in the step of convolving the equalized image and obtaining a smoothed image: gaussian blur function, mean filter function, median filter function, custom filter function.
6. The method of claim 1, wherein
The process of obtaining the contour data of the bright spots in the original image further comprises: and setting a screening threshold value according to the area distribution of the contour so as to reject contour data with larger deviation from the center of the area distribution of the contour.
7. The method of claim 1, wherein
The fitting to obtain the precise position of the central point of the bright spot comprises: setting a fitting model as a circular Gaussian function, taking the pixel values in the area containing the bright spots as data, and obtaining undetermined parameters of the fitting model by using the fitting function, wherein the circular Gaussian function is as follows:
wherein x0And y0For the accurate position of the central point of the bright spot to be fitted, Amp is the amplitude of a Gaussian function, sigma is the area parameter of the bright spot, and offset is the translation parameter of the central point of the bright spot in the original image.
8. The method of claim 7, wherein the fitting function employs a least squares curve fitting method.
9. The method of claim 1, further comprising:
and obtaining the central positions of other atoms according to the central positions of a plurality of adjacent bright atoms and the lattice type of the material.
10. The method of claim 9, wherein the step of obtaining the center positions of the other atoms comprises:
establishing a kd-tree data structure and establishing mapping of each bright atom and the bright atoms adjacent to the bright atom;
calculating estimated central point positions of other atoms among the plurality of neighboring bright atoms according to the lattice structure relationship of the material and the accurate positions of the central points of the plurality of neighboring bright atoms;
extracting pixel values of corresponding areas in the original image according to the estimated central point position, and fitting to obtain the accurate central point positions of the other atoms, wherein the fitting to obtain the accurate central point positions of the other atoms comprises:
setting a fitting model as a circular Gaussian function, taking the pixel values in the region as data, and obtaining undetermined parameters of the fitting model by using the fitting function, wherein the circular Gaussian function is as follows:
wherein x0And y0For the accurate position of the central point of the bright spot to be fitted, Amp is the amplitude of a Gaussian function, sigma is the area parameter of the bright spot, and offset is the translation parameter of the central point of the bright spot in the original image.
11. A transmission electron microscope image analysis apparatus comprising a memory and a processor, wherein the memory has stored therein a computer program which, when executed, carries out the method of one of claims 1 to 10.
12. A computer-readable storage medium, on which a computer program is stored which, when executed, implements the method of one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011008851.2A CN112116581B (en) | 2020-09-23 | 2020-09-23 | Method and device for acquiring atomic position in atomic imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011008851.2A CN112116581B (en) | 2020-09-23 | 2020-09-23 | Method and device for acquiring atomic position in atomic imaging |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112116581A true CN112116581A (en) | 2020-12-22 |
| CN112116581B CN112116581B (en) | 2023-09-08 |
Family
ID=73801094
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011008851.2A Active CN112116581B (en) | 2020-09-23 | 2020-09-23 | Method and device for acquiring atomic position in atomic imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112116581B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5797396A (en) * | 1995-06-07 | 1998-08-25 | University Of Florida Research Foundation | Automated method for digital image quantitation |
| US20030193022A1 (en) * | 2002-04-12 | 2003-10-16 | Houge Erik Cho | Calibration standard for high resolution electron microscopy |
| US20100254611A1 (en) * | 2009-03-30 | 2010-10-07 | Carl Zeiss Sms Gmbh | Method and device for determining the position of an edge of a marker structure with subpixel accuracy in an image, having a plurality of pixels, of the marker structure |
| CN103679167A (en) * | 2013-12-18 | 2014-03-26 | 杨新锋 | Method for processing CCD images |
| CN106204544A (en) * | 2016-06-29 | 2016-12-07 | 南京中观软件技术有限公司 | A kind of automatically extract index point position and the method and system of profile in image |
| CN107784669A (en) * | 2017-10-27 | 2018-03-09 | 东南大学 | A kind of method that hot spot extraction and its barycenter determine |
| CN109523565A (en) * | 2018-11-15 | 2019-03-26 | 湖北工业大学 | A kind of diffraction light-free Moire fringe center positioning method and system |
| KR101969242B1 (en) * | 2017-12-29 | 2019-04-15 | 포항공과대학교 산학협력단 | Apparatus and method for analyzing atomic structure |
| CN110487225A (en) * | 2019-08-27 | 2019-11-22 | 上海交通大学 | Atomic scale strain calculation method and atomic scale strain calculation device |
| CN110533601A (en) * | 2019-07-15 | 2019-12-03 | 江苏大学 | A kind of Position of Laser-Spot Center and profile acquisition methods |
| CN111089537A (en) * | 2019-12-23 | 2020-05-01 | 中国科学院西安光学精密机械研究所 | Method and system for detecting imaging position offset of Doppler differential interferometer |
| CN111562408A (en) * | 2020-06-02 | 2020-08-21 | 江南大学 | Atomic force microscope long-distance operation method based on local scanning imaging |
-
2020
- 2020-09-23 CN CN202011008851.2A patent/CN112116581B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5797396A (en) * | 1995-06-07 | 1998-08-25 | University Of Florida Research Foundation | Automated method for digital image quantitation |
| US20030193022A1 (en) * | 2002-04-12 | 2003-10-16 | Houge Erik Cho | Calibration standard for high resolution electron microscopy |
| US20100254611A1 (en) * | 2009-03-30 | 2010-10-07 | Carl Zeiss Sms Gmbh | Method and device for determining the position of an edge of a marker structure with subpixel accuracy in an image, having a plurality of pixels, of the marker structure |
| CN103679167A (en) * | 2013-12-18 | 2014-03-26 | 杨新锋 | Method for processing CCD images |
| CN106204544A (en) * | 2016-06-29 | 2016-12-07 | 南京中观软件技术有限公司 | A kind of automatically extract index point position and the method and system of profile in image |
| CN107784669A (en) * | 2017-10-27 | 2018-03-09 | 东南大学 | A kind of method that hot spot extraction and its barycenter determine |
| KR101969242B1 (en) * | 2017-12-29 | 2019-04-15 | 포항공과대학교 산학협력단 | Apparatus and method for analyzing atomic structure |
| CN109523565A (en) * | 2018-11-15 | 2019-03-26 | 湖北工业大学 | A kind of diffraction light-free Moire fringe center positioning method and system |
| CN110533601A (en) * | 2019-07-15 | 2019-12-03 | 江苏大学 | A kind of Position of Laser-Spot Center and profile acquisition methods |
| CN110487225A (en) * | 2019-08-27 | 2019-11-22 | 上海交通大学 | Atomic scale strain calculation method and atomic scale strain calculation device |
| CN111089537A (en) * | 2019-12-23 | 2020-05-01 | 中国科学院西安光学精密机械研究所 | Method and system for detecting imaging position offset of Doppler differential interferometer |
| CN111562408A (en) * | 2020-06-02 | 2020-08-21 | 江南大学 | Atomic force microscope long-distance operation method based on local scanning imaging |
Non-Patent Citations (2)
| Title |
|---|
| 南虎;卢江波;刘明;井红梅;汤少杰;王大威;贾春林;: "高分辨原子像中原子峰位置的精确自动判定与电极化畴的快速可视化", pages 191 - 200 * |
| 周鑫: "透射电镜图像定量化分析", pages 020 - 1137 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112116581B (en) | 2023-09-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978839B (en) | Method for detecting wafer low-texture defects | |
| US6111981A (en) | Method and apparatus for processing pattern image data by SEM | |
| Ahmed | Comparative study among Sobel, Prewitt and Canny edge detection operators used in image processing | |
| US8699784B2 (en) | Inspection recipe generation and inspection based on an inspection recipe | |
| CN113963042B (en) | Metal part defect degree evaluation method based on image processing | |
| CN112132791A (en) | Shale scanning electron microscope image processing method, device, equipment and storage medium | |
| CN115797300A (en) | Edge detection method and device based on adaptive gradient threshold canny operator | |
| KR20170040983A (en) | Method and apparatus of image denoising using multi-scale block region detection | |
| CN111429372A (en) | Method for enhancing edge detection effect of low-contrast image | |
| CN112116581B (en) | Method and device for acquiring atomic position in atomic imaging | |
| CN111340741A (en) | Particle swarm optimization gray level image enhancement method based on quaternion and L1 norm | |
| CN116363097A (en) | Defect detection method and system for photovoltaic panel | |
| Kumar et al. | Robust filtering and particle picking in micrograph images towards 3D reconstruction of purified proteins with cryo-electron microscopy | |
| CN116523816A (en) | Sea surface ship target detection method | |
| El-Sayed | Study of edge detection based on 2D entropy | |
| CN113393387A (en) | Mathematical morphology self-adaptive structural element construction method | |
| CN107292854B (en) | Gray level image enhancement method based on local singularity quantitative analysis | |
| Nieniewski et al. | Segmentation of spinal cord images by means of watershed and region merging together with inhomogeneity correction | |
| CN117274241B (en) | Brake drum surface damage detection method and device based on rapid image analysis | |
| Tangsakul et al. | Improving single image haze removal based on cellular automata model | |
| Pang et al. | Prediction of the suitability for image-matching based on self-similarity of vision contents | |
| CN111476821B (en) | Target tracking method based on online learning | |
| Li | Classification of SAR images using morphological texture features | |
| CN116205940B (en) | Digital image target detection method and system based on medical examination | |
| CN111754524B (en) | Remote sensing image rapid sea and land segmentation method based on shoreline data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |