CN113160223A

CN113160223A - Contour determination method, contour determination device, detection device and storage medium

Info

Publication number: CN113160223A
Application number: CN202110534946.6A
Authority: CN
Inventors: 陈鲁; 肖遥; 佟异; 张嵩
Original assignee: Shenzhen Zhongke Feice Technology Co Ltd
Current assignee: Skyverse Ltd; Shenzhen Zhongke Feice Technology Co Ltd
Priority date: 2021-05-17
Filing date: 2021-05-17
Publication date: 2021-07-23

Abstract

The application discloses a contour determination method. The method for determining the contour is used for determining the graphic contour of a target graphic in an image, and comprises the following steps: acquiring an initial contour of the target graph according to the shape information and the initial position information of the target graph; respectively selecting a plurality of calculation areas by taking each of a plurality of preset points on the initial contour as a center; determining a fitting point in each calculation region according to the distribution information of the characteristic values of the image pixels in each calculation region; and fitting the fitting points in the plurality of calculation regions to obtain a graph outline of the target graph. The application also discloses a contour determining device, a detecting device and a computer readable storage medium.

Description

Contour determination method, contour determination device, detection device and storage medium

Technical Field

The present disclosure relates to the field of industrial detection technologies, and in particular, to a method, an apparatus, a device and a computer readable storage medium for determining a contour.

Background

Semiconductor products generally comprise a plurality of characteristic patterns, and the characteristic patterns on the products need to be measured frequently so as to detect whether the semiconductor products are qualified or not by measuring whether the characteristic patterns meet standards, and in the process, it is important to accurately determine the outlines of the characteristic patterns in pictures. Some methods mark the contour of the feature pattern directly by a binarization method, but the binarization method has high requirements on the brightness, contrast, and the like of the picture, and is liable to cause low accuracy when marking the contour of the feature pattern.

Disclosure of Invention

The embodiment of the application provides a contour determining method, a contour determining device, a detecting device and a computer readable storage medium.

The contour determination method of the embodiment of the application is used for determining the figure contour of a target figure in an image, and comprises the following steps: acquiring an initial contour of the target graph according to the shape information and the initial position information of the target graph; selecting a plurality of calculation areas by respectively taking each of a plurality of preset points on the initial contour as a center; determining a fitting point in each calculation region according to distribution information of characteristic values of image pixels in each calculation region; and fitting the fitting points in the plurality of calculation regions to obtain a graph outline of the target graph.

In some embodiments, said selecting a plurality of calculation regions, respectively centered on each of a plurality of preset points on said initial contour, comprises: determining a deflection direction corresponding to each preset point according to the tangent direction of the initial contour on each preset point; setting a preset graph in the corresponding deflection direction by taking each preset point as a center; and selecting the area in the preset graph as the calculation area.

In some embodiments, the preset pattern is a rectangle, and the setting the preset pattern with each preset point as a center and the corresponding deflection direction respectively includes: setting the extending direction of the first edge of the preset graph to be parallel to the tangential direction; setting half of a second edge of the preset graph to be larger than a maximum offset, wherein the maximum offset is a preset maximum offset between the initial profile and the graph profile; and setting an overlapping area of two adjacent preset graphs.

In some embodiments, the determining a fitting point in each of the calculation regions according to distribution information of feature values of image pixels in each of the calculation regions includes: in each calculation region, calculating an accumulated value of the characteristic values of the pixels along a tangential direction, or an average value of the accumulated values, wherein the tangential direction is the tangential direction of the initial contour at the preset point; calculating a gradient value of the accumulated value or the average value in a direction perpendicular to the tangent; and determining a fit point in the calculation region based on the gradient values.

In some implementations, the determining a fit point in the calculation region based on the gradient values includes: and determining that the sign of the gradient value is a preset sign, and determining that the point at which the absolute value of the gradient value is greater than the gradient threshold value is a fitting point in the calculation region.

In some embodiments, the fitting points in the plurality of calculation regions to obtain the figure contour of the target figure includes: fitting a plurality of the fitting points to obtain a fitting outline of the target graph; a calculation step: respectively calculating the distances from a plurality of fitting points to the fitting contour; a first fitting step: eliminating the fitted points with the distance larger than a preset distance threshold value, and fitting the rest fitted points to obtain a new fitted contour; and after the calculating step and the first fitting step are executed circularly for preset times, taking the latest fitting contour as the graphic contour.

In some embodiments, the fitting points in the plurality of calculation regions to obtain the figure contour of the target figure includes: fitting a plurality of the fitting points to obtain a fitting outline of the target graph; a calculation step: respectively calculating the distances from a plurality of fitting points to the fitting contour; a second fitting step: according to the size relation between the distance and the distance threshold value, setting a fitting weight for each fitting point, and according to the fitting weight, re-fitting a plurality of fitting points to obtain a new fitting contour; and after the calculating step and the second fitting step are executed circularly for preset times, taking the latest fitting contour as the graphic contour.

The device for determining the contour is used for determining the contour of a target figure in an image, and comprises an acquisition module, a selection module, a determination module and a fitting module, wherein the acquisition module is used for acquiring the initial contour of the target figure according to the shape information and the initial position information of the target figure; the selecting module is used for selecting a plurality of calculation areas by respectively taking each of a plurality of preset points on the initial contour as a center; the determining module is used for determining a fitting point in each computing area according to the distribution information of the characteristic values of the image pixels in each computing area; the fitting module is used for fitting the fitting points in the plurality of calculation regions to obtain a graph outline of the target graph.

The detection equipment comprises an imaging device, a memory and a processor, wherein the imaging device is used for shooting an image of a piece to be detected; the memory is used for storing the image; the processor is in communication with the memory, and the processor is configured to perform the method for determining a contour according to any of the embodiments of the present application.

The non-transitory computer-readable storage medium of the embodiments of the present application stores a computer program that, when executed by one or more processors, causes the processors to perform the method of determining a contour as described in any of the embodiments of the present application.

In the method for determining a contour, the apparatus for determining a contour, the detection device, and the computer-readable storage medium according to the embodiments of the present application, the approximate position of the contour of a graphic is determined by obtaining an initial contour of a target image, then a calculation region is selected around the initial contour, a fitting point is determined according to distribution information of feature values of image pixels by analyzing feature values of image pixels in each calculation region more finely, and then the image contour is obtained by fitting according to the fitting point, so that the accuracy of the finally determined image contour is high.

Additional aspects and advantages of embodiments of the present application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present application.

Drawings

The above and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

FIG. 1 is a schematic flow chart of a method of determining a profile according to certain embodiments of the present application;

FIG. 2 is a block schematic diagram of a contour determination apparatus according to certain embodiments of the present application;

FIG. 3 is a schematic structural view of a detection apparatus according to certain embodiments of the present application;

FIGS. 4-9 are schematic illustrations of a method of determining according to certain embodiments of the present application;

FIG. 10 is a schematic flow chart diagram of a method of determining a profile according to certain embodiments of the present application;

FIG. 11 is a block diagram of a profile determining apparatus according to certain embodiments of the present application;

FIG. 12 is a schematic flow chart diagram of a method of determining a profile according to certain embodiments of the present application;

FIG. 13 is a block diagram of a profile determining apparatus according to certain embodiments of the present application;

FIG. 14 is a schematic flow chart diagram of a method of determining a profile according to certain embodiments of the present application;

FIG. 15 is a block diagram of a profile determining apparatus according to certain embodiments of the present application;

FIG. 16 is a schematic illustration of the principles of a determination method according to certain embodiments of the present application;

FIG. 17 is a schematic flow chart diagram of a method of determining a profile according to certain embodiments of the present application;

FIG. 18 is a block diagram of a profile determining apparatus according to certain embodiments of the present application;

FIG. 19 is a schematic flow chart diagram of a method of determining a profile according to certain embodiments of the present application;

FIG. 20 is a block diagram of a profile determining apparatus according to certain embodiments of the present application;

FIG. 21 is a schematic diagram of a computer-readable storage medium and a processor in accordance with certain embodiments of the present application.

Description of the main elements and symbols:

a determination device 10, an acquisition module 11,

The selection module 12, the determination unit 121, the first setting unit 122, the first setting subunit 1221, the second setting subunit 1222, the third setting subunit 1223, the selection unit 123, the determination unit, the first setting subunit 1221, the second setting subunit 1222, the third setting subunit 1223, the determination unit 123, the determination unit, the second setting subunit, the third setting subunit, and the third setting subunit,

A determination module 13, a first calculation unit 131, a second calculation unit 132, a second setting unit 133,

A fitting module 14, a first fitting unit 141, a third calculation unit 142, a second fitting unit 1431, a third fitting unit 1432, a first circulation unit 144, a second circulation unit 145, a,

A detection device 20, an imaging apparatus 21, a memory 22, a processor 23,

A computer readable storage medium 30, a computer program 31, a processor 40.

Detailed Description

Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below by referring to the drawings are exemplary only for the purpose of explaining the embodiments of the present application, and are not to be construed as limiting the embodiments of the present application.

Referring to fig. 1, an embodiment of the present application discloses a method for determining a contour, where the method for determining a contour of a target graphic in an image includes:

01: acquiring an initial contour of the target graph according to the shape information and the initial position information of the target graph;

02: respectively selecting a plurality of calculation areas by taking each of a plurality of preset points on the initial contour as a center;

03: determining a fitting point in each calculation region according to the distribution information of the characteristic values of the image pixels in each calculation region; and

04: fitting the fitting points in the plurality of calculation regions to obtain a figure contour of the target figure.

Referring to fig. 2, the present application further discloses a contour determination apparatus 10, and the determination apparatus 10 can be used to implement the contour determination method according to the present application. Specifically, the determination apparatus 10 includes an acquisition module 11, a selection module 12, a determination module 13, and a fitting module 14. The obtaining module 11 may be configured to perform step 01, that is, the obtaining module 11 may be configured to obtain an initial contour of the target graph according to the shape information and the initial position information of the target graph. The selection module 12 may be configured to perform step 02, i.e. the selection module 12 may be configured to select a plurality of calculation regions respectively centered on each of a plurality of preset points on the initial contour. The determining module 13 may be configured to implement step 03, that is, the determining module 13 may be configured to determine the fitting point in each calculation region according to the distribution information of the feature values of the image pixels in each calculation region. Fitting module 14 may be configured to perform step 04, that is, fitting module 14 may be configured to fit the fitting points in the plurality of calculation regions to obtain the figure contour of the target figure.

Referring to fig. 3, the present application further discloses a detection apparatus 20, where the detection apparatus 20 includes an imaging device 21, a memory 22, and a processor 23. The imaging device 21 is used to take an image of the object 100 to be inspected. The memory 22 is used to store images. The processor 23 is communicatively connected to the memory 22, and the processor 23 can be used to execute the contour determination method according to the embodiment of the present application. Specifically, the processor 23 may be configured to implement the

steps

01, 02, 03 and 04.

In the method for determining a contour, the apparatus 10 for determining a contour, and the detection device 20 according to the embodiments of the present application, an approximate position of a contour of a figure is determined by obtaining an initial contour of a target image, a calculation region is selected around the initial contour, a feature value of an image pixel in each calculation region is analyzed more finely, a fitting point is determined according to distribution information of the feature value of the image pixel, an image contour is obtained according to fitting of the fitting point, and accuracy of the finally determined image contour is high.

Referring to fig. 3 and 4, the inspection apparatus 20 may be a visual inspection apparatus, and in one example, the inspection apparatus 20 may be used to inspect whether a defect exists on the object 100 to be inspected, or measure the size of some features on the object 100 to be inspected, wherein the object 100 to be inspected may be a semiconductor device, such as a wafer, a circuit board, a panel, a substrate, a chip, and the like, without limitation.

The imaging device 21 may be used to capture an image of the object to be inspected, it being understood that the image may include a graphic representation of the detail feature of the object 100 to be inspected. As shown in fig. 4, the image P1 of the object 100 includes various patterns with different shapes and positions, such as G1, G2, G3, G4, and so on. By analyzing the position, size and other characteristics of the pattern, it can be determined whether the quality of the object 100 to be inspected meets the requirements. Accurately identifying the contour of the target pattern is an important basis for analyzing the characteristics of the target pattern.

The image captured by the imaging device 21 may be sent to the memory 22 for storage, and of course, some standard information of the target image, status parameters and control instructions of the imaging device 21, and related information sent by upper software may also be stored in the memory 22. The processor 23 can accurately determine the figure outline of the target figure by processing the image stored in the imaging device 21.

In step 01, an initial contour of the target pattern is obtained according to the shape information and the initial position information of the target pattern. Specifically, the shape information and the position information of the target pattern may be provided by a preceding software flow or may be already stored in the memory 22, that is, the shape information and the initial position information of the target pattern are known in the contour determination method according to the embodiment of the present application. Specifically, the shape information of the target figure, which is used to represent the outline shape of the target figure, may be a straight line, a circle, a rectangle, or the like. The initial position information of the target pattern may be rough position information of the target pattern, which may be deviated from precise position information of the target pattern, but an approximate range of the deviation is known. For target graphics with different shape information, the representation form of the initial position information may be different, for example, if the shape information of the target graphics is a line segment, the initial position information includes two end points of the line segment respectively; if the shape information of the target graph is a circle, the initial position information comprises the coordinate and the diameter of the circle center; if the shape information of the target figure is a rectangle, the initial position information includes coordinates of the center of the rectangle, lengths of the length and width, and an angle of rotation. Of course, different target patterns may have different corresponding shape information and initial position information, and are not limited herein.

It can be understood that, according to the shape information and the initial position information of the target pattern, the initial contour of the target pattern can be determined, and the initial contour can be used to roughly determine the pattern contour of the target pattern, reduce the range of determining the pattern contour, and reduce the calculation amount of the algorithm for determining the pattern contour. In the image P shown in fig. 5, the target figure is a figure G, the figure G is substantially rectangular, and after the step 01 is performed, an initial contour E (indicated by a dotted line frame in fig. 5) of the target figure can be obtained, and the initial contour E is deviated from the figure contour of the target figure G to a certain extent.

In step 02, a plurality of calculation regions are selected, each centered on each of a plurality of preset points on the initial contour. As described above, the initial contour of the target graphic may be deviated from the graphic contour, and the graphic contour may be distributed within a certain range around the initial contour, so that the calculation region is selected by centering on the preset point on the initial contour, and the approximate range where the point on the graphic contour is located may be actually selected without performing subsequent calculation on the entire image, thereby reducing the calculation amount. Specifically, the preset points are all set on the initial contour, and each preset point can determine a calculation area. The preset points may be distributed at equal intervals or at unequal intervals, and the intervals between the preset points may be set according to actual requirements, which is not limited herein. In the examples shown in fig. 6 and 7, a plurality of preset points are preset on the initial contour E, and taking the preset point D at the upper left corner as an example, a calculation area selected according to the preset point D is a, and points in the calculation area a including the graphic contour of the graphic G are respectively centered on the plurality of preset points, so that a plurality of calculation areas can be selected, as shown in fig. 7.

In step 03, fitting points in each calculation region are determined according to distribution information of feature values of image pixels in each calculation region. Each calculation region is selected by taking a preset point as a center, points on the graphic outline may exist in each calculation region, and fitting points on the graphic outline can be determined through step 03 to be used for finally fitting to obtain the graphic outline. Specifically, the target graph is a shape presented by a certain feature on the to-be-detected piece, the feature presents a characteristic different from a background on the image, and the characteristic can be embodied by distribution information of a feature value of an image pixel. The characteristic value may be a gray scale value, a brightness value, a pixel value of a certain channel, etc., and is not limited herein. According to the distribution information of the characteristic values of the image pixels, fitting points in the calculation area can be determined so as to determine boundary points between the target graph and the image background, and finally the boundary points are used for fitting to obtain the graph outline. In the examples shown in fig. 7 and 8, one fitting point N can be calculated for each calculation region, and a plurality of fitting points N can be calculated for a plurality of calculation regions, the plurality of fitting points N being shown as open dots in fig. 8.

In step 04, fitting points in the plurality of calculation regions to obtain a figure contour of the target figure. As described above, each fitting point is actually a boundary point between the target pattern and the image background, and by fitting a plurality of fitting points, a pattern profile of the target pattern can be obtained, which can reflect the actual profile of the target pattern more accurately than the initial profile. In the example shown in fig. 9, by fitting a plurality of fitting points, the figure contour F of the target figure G can be obtained.

Referring to fig. 10, in some embodiments, step 02: selecting a plurality of calculation areas respectively by taking each of a plurality of preset points on the initial contour as a center, and comprising the following steps of:

021: determining a deflection direction corresponding to each preset point according to the tangential direction of the initial contour on each preset point;

022: setting a preset graph in a corresponding deflection direction by taking each preset point as a center; and

023: and selecting an area in the preset graph as a calculation area.

Referring to fig. 11, in some embodiments, the selecting module 12 includes a determining unit 121, a first setting unit 122, and a selecting unit 123. The determining unit 121 may be configured to implement step 021, that is, the determining unit 121 may be configured to determine the deflection direction corresponding to the preset point according to the tangential direction of the initial contour at each preset point. The first setting unit 122 may be configured to perform step 022, i.e., the first setting unit 122 may be configured to set the preset pattern with the corresponding deflection direction respectively centered at each preset point. The selecting unit 123 may be used to implement step 023, i.e. the selecting unit 123 may be used to select an area within the preset pattern as the calculation area.

In addition, referring to fig. 2, in some embodiments, the processor 23 in the detection apparatus 20 can also be used to implement

steps

021, 022 and 023.

By setting the preset graph and selecting the area in the preset graph as the calculation area, the shapes of the calculation areas corresponding to each preset point are unified, and the fitting points can be obtained by adopting a unified algorithm for a plurality of calculation areas subsequently.

Specifically, in step 021, a deflection direction corresponding to each preset point is determined according to a tangential direction of the initial contour at each preset point, the tangential direction of the preset point is related to the extension state of each preset point, and the deflection direction is determined according to the extension state, so that a calculation area more consistent with the extension state of each preset point can be obtained.

In step 022, a preset pattern is set around each preset point and in the corresponding deflection direction, where the preset pattern may be preset by a user, for example, a rectangle, a circle, a square, a diamond, a triangle, and the like, without limitation. Each preset point is provided with a preset graph in the deflection direction, so that the preset graphs can be quickly set in batches, and a plurality of calculation areas can be quickly selected.

In step 023, the area in the preset pattern is selected as the calculation area, so that the calculation area corresponding to each preset point is clear in boundary.

Referring to the example shown in fig. 6, the tangential direction of the preset point D is the horizontal direction in fig. 6, after the horizontal direction is determined, the preset pattern K is set with the horizontal direction as the deflection direction, and the area in the preset pattern K is the calculation area a. Referring to the example shown in fig. 7, a calculation area can be obtained according to each preset point on the initial contour E, and the area covered by the calculation areas includes all areas around the initial contour E.

Referring to fig. 12, in some embodiments, the predetermined pattern is a rectangle, and step 022: respectively taking each preset point as a center, setting a preset graph in a corresponding deflection direction, and comprising the following steps of:

0221: setting the extending direction of the first edge of the preset graph to be parallel to the tangential direction;

0222: setting half of a second edge of a preset graph to be larger than the maximum offset, wherein the maximum offset is the preset maximum offset between the initial profile and the graph profile; and

0223: and setting an overlapping area of two adjacent preset graphs.

Referring to fig. 13, in some embodiments, the predetermined pattern is a rectangle, and the first setup unit 122 includes a first setup subunit 1221, a second setup subunit 1222, and a third setup subunit 1223. The first setting subunit 1221 may be used to perform step 0221, that is, the first setting subunit 1221 may be used to set an extending direction of the first edge of the preset pattern to be parallel to the tangential direction. The second setting subunit 1222 may be configured to implement step 0222, i.e., the second setting subunit 1222 may be configured to set half of the second side of the preset pattern to be greater than a maximum offset, which is a preset maximum offset between the initial contour and the contour of the pattern. The third setting subunit 1223 may be configured to perform step 0223, that is, the third setting subunit 1223 may be configured to set an overlapping area where two adjacent preset patterns exist.

Referring to FIG. 2, in some embodiments, processor 23 of detection apparatus 20 may also be used to perform

steps

0221, 0222, and 0223.

The preset graph is set to be rectangular, and the shape of the calculation region defined by the rectangle is more regular, so that the subsequent calculation of the characteristic value of the image pixel is easier to implement, and the regions around the initial contour are easier to be completely covered by a plurality of calculation regions. Step 0221, step 0222, and step 0223 may be performed in any one, two, or three of them, and step 0221, step 0222, and step 0223 are not necessarily required to be performed in the order in fig. 12.

In step 0221, the extending direction of the first side of the preset pattern is set to be parallel to the tangential direction. The first side may be a long side or a short side of the rectangle, and the extending direction of the first side is set to be parallel to the tangential direction, so that the calculation region at least extends in the extending direction of the preset point.

In step 0222, half of the second edge of the preset graph is set to be greater than the maximum offset, where the maximum offset is the preset maximum offset between the initial contour and the graph contour, and it should be noted that the maximum offset is the offset between the initial contour and the graph contour, and there may not be a direct correlation with whether the center of the whole graph is offset, for example, when there is a non-zero offset between a corresponding set of edges in the initial contour and the graph contour, the center of the initial contour and the center of the graph contour may be coincident or may not be coincident. The second side is perpendicular to the first side, and when the first side is a long side of the rectangle, the second side is a short side of the rectangle, and when the first side is a short side of the rectangle, the second side is a long side of the rectangle. The maximum offset may be an empirical value, or the maximum offset may be a maximum value of the offsets between a set of graphic profiles and the initial profile in an image of a known graphic profile, without limitation. And setting half of the second side of the preset graph to be larger than the maximum offset, so that the finally selected calculation area can cover part of the graph outline, and a fitting point can be calculated through each calculation area.

In step 0223, two adjacent preset graphs are set to have an overlapping region, and two adjacent preset graphs have an overlapping region, that is, two adjacent calculation regions have an overlapping region, so that the same region around the initial contour may be covered by multiple calculation regions, and the feature value of the image pixel in the region may be used for multiple calculations, thereby increasing the number of samples when the fitting point is obtained by calculation, and further increasing the accuracy of obtaining the fitting point by calculation.

Referring to the example shown in fig. 7, the extending direction of the short side of the rectangular preset pattern K is parallel to the tangential direction of the preset point, the extending direction of the long side of the preset pattern K is perpendicular to the tangential direction of the preset point, each preset pattern K can surround at least a part of the pattern contour, and an overlapping area exists between two adjacent preset patterns K.

Referring to fig. 14, in some embodiments, step 03: determining a fitting point in each calculation region according to distribution information of characteristic values of image pixels in each calculation region, comprising the steps of:

031: in each calculation region, calculating an accumulated value of the characteristic values of the pixels along the tangential direction, or an average value of the accumulated values, wherein the tangential direction is the tangential direction of the initial contour at a preset point;

032: calculating the gradient value of the accumulated value or the average value along the direction vertical to the tangent; and

033: a fit point in the calculation region is determined based on the gradient values.

Referring to fig. 15, in some embodiments, the determining module 13 includes a first calculating unit 131, a second calculating unit 132 and a second setting unit 133. The first calculation unit 131 may be configured to implement step 031, that is, the first calculation unit 131 may be configured to calculate, in each calculation region, an accumulated value of the feature values of the pixels in the tangential direction, or an average value of the accumulated values. The second calculation unit 132 may be used to implement step 032, i.e. the second calculation unit 132 may be used to calculate a gradient value of the accumulated value or the average value along the direction perpendicular to the tangent. A second setup unit 133 may be used to implement step 033, i.e., the second setup unit 133 may be used to determine a fitting point in the calculation region based on the gradient values.

Referring to fig. 2, in some embodiments, the processor 23 in the detection apparatus 20 can also be used to implement

steps

031, 032 and 033.

In the image of the object to be measured, the target graph and the background have obvious difference on the characteristic value, the difference inside the target graph is not obvious, and the difference inside the background is not obvious. By calculating the gradient value of the characteristic value along the direction vertical to the tangent line, the position of the sudden change of the characteristic value in the calculation area can be calculated to obtain a fitting point, and further, a more accurate graph outline can be obtained.

Specifically, in step 031, in each calculation region, the accumulated value or the average value of the accumulated values of the feature values of the pixels along the tangential direction is calculated, where in the same calculation region, the accumulated value is calculated for each line of pixels along the tangential direction, or, in the same calculation region, the average value of the accumulated values is calculated for each line of pixels along the tangential direction. In different calculation regions, there may be a case where one calculation region calculates the accumulated values and another calculation region calculates the average value of the accumulated values. The accumulated value or the average value of the accumulated values is used for subsequent calculation, the acquired data has higher stability and smaller noise

In step 032, a gradient value of the accumulated value or the average value along a direction perpendicular to the tangent is calculated, it can be understood that a change rate of the feature value of the image pixel along the tangent direction is smaller, and a change rate of the feature value of the image pixel along the direction perpendicular to the tangent direction is larger.

In step 033, fitting points in the calculation region are determined based on the gradient values to determine positions of the fitting points on the graph, so that the graph contour can be obtained through fitting of the fitting points subsequently.

Referring to the example of one calculation region a shown in fig. 16, taking the feature value as the gray-scale value, the accumulated values of the feature values of the image pixels along the tangential direction are calculated first in step 031, for example, the numbers on the right side of the calculation region a in fig. 16 indicate the accumulated values of the feature values of the pixels in multiple rows in the calculation region a. By implementing step 032, a gradient value of the accumulated value in the direction perpendicular to the tangent line can be calculated (a gradient value is obtained by differentiating the front and rear arrays), and then by implementing step 033, a fitting point in the calculation region can be determined as point a based on the gradient value.

In one example, step 033: determining a fitted point in the calculation region based on the gradient values, comprising the steps of: and determining the sign of the gradient value as a preset sign, and determining the point at which the absolute value of the gradient value is greater than the gradient threshold value as a fitting point in the calculation region. Specifically, the position where the absolute value of the gradient value is larger may be caused by the characteristic value changing from large to small (for example, the image changes from white to black) or caused by the characteristic value changing from small to large (for example, the image changes from black to white), and in an actual algorithm, it may be determined by means of a preset symbol that the symbol of the gradient value conforms to the change rule of the accumulated value or the average value of the accumulated value when the image changes from the background of the image to the target pattern. For example, taking the characteristic value as the gray-level value, and taking the gray-level value of the target graphic relative to the background as a smaller example, when the background changes into the target graphic, the gradient of the gray-level value should be negative, and the predetermined sign is negative.

Meanwhile, the absolute value of the gradient value needs to be larger than the gradient threshold value so as to avoid the influence of defects such as bad points and the like existing in the image on the calculation of the fitting point. The gradient threshold may be set according to an empirical value, and is not limited herein.

When the positions of the pixel rows along the tangential direction with gradient values meeting the requirement are calculated, the position of the pixel in the middle of the pixel row can be used as the position of the fitting point.

Referring to fig. 17, in some embodiments, step 04: fitting the fitting points in the plurality of calculation regions to obtain a figure contour of the target figure, comprising the steps of:

041: fitting the fitting points to obtain a fitting outline of the target graph;

042: a calculation step: respectively calculating the distances from the fitting points to the fitting contour;

0431: a first fitting step: eliminating fitted points with the distance larger than a preset distance threshold value, and fitting the remaining fitted points to obtain a new fitted contour; and

044: and after the calculation step and the first fitting step are executed circularly for preset times, the latest fitting contour is used as the graphic contour.

Referring to fig. 18, in some embodiments, the fitting module 14 includes a first fitting unit 141, a third calculating unit 142, a second fitting unit 1431, and a first circulation unit 144. The first fitting unit 141 may be configured to perform step 041, that is, the first fitting unit 141 may be configured to fit the plurality of fitting points to obtain a fitting contour of the target graph. The third calculation unit 142 may be configured to implement step 042, that is, the third calculation unit 142 may be configured to calculate distances from the plurality of fitting points to the fitted contour, respectively. The second fitting unit 1431 may be configured to implement step 0431, that is, the second fitting unit 1431 may be configured to eliminate fitted points whose distances are greater than a preset distance threshold, and fit the remaining fitted points to obtain a new fitted contour. The first loop unit 144 may be configured to perform the step 044, that is, the first loop unit 144 may be configured to loop the calculation step (042) and the first fitting step (0431) for a predetermined number of times, and then use the latest fitted contour as the graph contour.

Referring to fig. 2, in some embodiments, the processor 23 in the detection apparatus 20 may also be configured to perform step 041, step 042, step 0431 and step 044.

In step 041, the fitting points are fitted to obtain a fitting profile of the target graph, the fitting profile may be obtained by least square fitting, and fitting may be performed in combination with shape information of the target graph during fitting, so that the fitting profile conforms to the shape information, for example, the fitting profile obtained by fitting is a rectangle, a circle, a straight line, or the like. The fitted contour can reflect the figure contour to a certain extent, so that the fitted contour can be used as the final figure contour when the accuracy required by the algorithm is low.

In step 042 (calculating step), the distances from the fitting points to the fitting contour are calculated respectively, and the distances from the fitting points to the fitting contour can be used to reflect the fitting degree of the fitting points to the fitting contour, so as to facilitate the subsequent further screening of the fitting points.

In step 0431 (the first fitting step), the fitted points whose distance is greater than the preset distance threshold are removed, and the remaining fitted points are fitted to obtain a new fitted contour, and for the fitted points whose distance is greater than the distance threshold, the quality of the fitted points is considered to be poor, which may be due to calculation deviation or noise generated by defects of the image itself, and such fitted points are removed, so that the influence of erroneous fitted points on the fitting result can be reduced, and the remaining fitted points are fitted again to obtain a new fitted contour, which is more accurate than the previous fitted contour.

In step 044, after the calculating step and the first fitting step are performed in a loop for a predetermined number of times, the latest fitted contour is used as the graphic contour. As described above, the new fitting profile has better accuracy than the previous fitting profile, and therefore, the calculation step and the first fitting step are performed in a loop for a predetermined number of times, so that the accuracy of the latest fitting profile is further improved. The preset times can be set individually by the user according to the requirements for precision and accuracy, and are not limited herein.

Referring to fig. 19, in some embodiments, step 04: fitting the fitting points in the plurality of calculation regions to obtain a figure contour of the target figure, comprising the steps of:

0432: a second fitting step: according to the size relation between the distance and the distance threshold value, setting a fitting weight for each fitting point, and according to the fitting weight, re-fitting the fitting points to obtain a new fitting contour; and

045: and after the calculating step and the second fitting step are executed circularly for preset times, the latest fitting contour is used as the graphic contour.

Referring to fig. 20, in some embodiments, the fitting module 14 includes a first fitting unit 141, a third calculating unit 142, a third fitting unit 1432, and a second circulation unit 145. The first fitting unit 141 may be configured to perform step 041, that is, the first fitting unit 141 may be configured to fit the plurality of fitting points to obtain a fitting contour of the target graph. The third calculation unit 142 may be configured to implement step 042, that is, the third calculation unit 142 may be configured to calculate distances from the plurality of fitting points to the fitted contour, respectively. The third fitting unit 1432 may be configured to implement step 0432, that is, the third fitting unit 1432 may be configured to set a fitting weight for each fitted point according to a magnitude relationship between the distance and the distance threshold, and re-fit the plurality of fitted points according to the fitting weights to obtain a new fitted contour. The second loop unit 145 may be configured to perform the step 045, that is, the second loop unit 145 may be configured to loop the calculation step (042) and the second fitting step (0432) for a predetermined number of times, and then use the latest fitted contour as the graph contour.

Referring to fig. 2, in some embodiments, the processor 23 in the detection apparatus 20 may also be configured to perform step 041, step 042, step 0432 and step 045.

Specifically, step 041 and step 042 can refer to the above description, and are not described herein again.

In step 0432 (second fitting step), a fitting weight is set for each fitted point according to a magnitude relation between the distance and the distance threshold, and a plurality of fitted points are re-fitted according to the fitting weight to obtain a new fitted contour, as described above, the distance from the fitted point to the fitted contour may be used to reflect the degree of fit between the fitted point and the fitted contour, and for fitted points whose distance is less than the distance threshold, the fitted point itself may be considered to be more accurate, and may take a higher weight in the fitting process of re-obtaining the fitted contour; for the fitted points with the distance greater than the distance threshold, the accuracy of the fitted points per se can be considered to be low, lower weight can be occupied in the fitting process of obtaining the fitted contour again, the weight is set for different fitted points according to the distance, the influence generated by noise in the fitted points can be weakened, and the fitted contour obtained through fitting is more accurate.

Specifically, how to set the fitting weight for each fitting point according to the magnitude relation between the distance and the distance threshold value can adopt different algorithms according to requirements, for example, the weight of each fitting point can be set as the ratio of the distance threshold value to the distance; or the weight of the fitted point whose distance is less than or equal to the distance threshold may be set to 1, and the weight of the fitted point whose distance is greater than the distance threshold may be set to a ratio of the distance threshold to the distance, which is not limited herein.

In one example, the preset distance threshold is 5, the distance between the fitting point 1 and the fitting contour is 2, and the distance between the fitting point 2 and the fitting contour is 10. For the fitting point 1, since the distance between the fitting point 1 and the fitting contour is smaller than the distance threshold, the weight of the fitting point 1 may be set to 1. For the fitted point 2, since the distance between the fitted point 2 and the fitted contour is greater than the distance threshold, the fitting weight of the fitted point 2 may be set to 0.5, i.e., the ratio of the distance threshold to the distance.

In step 045, after the calculating step and the second fitting step are performed in a loop for a predetermined number of times, the latest fitted contour is used as the graphic contour. As described above, the new fitting profile has better accuracy than the previous fitting profile, and therefore, the calculation step and the second fitting step are performed in a loop for a predetermined number of times, so that the accuracy of the latest fitting profile is further improved.

Referring to fig. 21, the embodiment of the present application further discloses a non-volatile computer readable storage medium 30, the computer readable storage medium 30 stores a computer program 31, and when the computer program 31 is executed by one or more processors 40, the processor 40 is caused to execute the method for determining the contour according to any embodiment of the present application.

For example, when the processor 40 executes the computer program 31, the processor 40 executes the

steps

01, 02, 03, 04, 021, 022, 023, 0221, 0222, 0223, 031, 032, 033, 041, 042, 0431, 044, 0432 and 045.

In summary, in the method for determining a contour, the apparatus for determining a contour, the detection device, and the computer-readable storage medium according to the embodiments of the present application, the approximate position of the contour of a graphic is determined by obtaining an initial contour of a target image, then a calculation region is selected around the initial contour, a feature value of an image pixel in each calculation region is analyzed more finely, a fitting point is determined according to distribution information of the feature value of the image pixel, and then the image contour is obtained by fitting according to the fitting point, so that the accuracy of the finally determined image contour is high.

In the description herein, reference to the description of the terms "one embodiment," "some embodiments," "an illustrative embodiment," "an example," "a specific example" or "some examples" or the like means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and the scope of the preferred embodiments of the present application includes other implementations in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present application.

Although embodiments of the present application have been shown and described above, it is to be understood that the above embodiments are exemplary and not to be construed as limiting the present application, and that changes, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present application.

Claims

1. A method for determining a contour of a target feature in an image, the method comprising:

acquiring an initial contour of the target graph according to the shape information and the initial position information of the target graph;

selecting a plurality of calculation areas by respectively taking each of a plurality of preset points on the initial contour as a center;

determining a fitting point in each calculation region according to distribution information of characteristic values of image pixels in each calculation region; and

fitting the fitting points in the plurality of calculation regions to obtain a figure contour of the target figure.

2. The method for determining a contour according to claim 1, wherein said selecting a plurality of calculation regions centered on each of a plurality of preset points on said initial contour comprises:

determining a deflection direction corresponding to each preset point according to the tangent direction of the initial contour on each preset point;

setting a preset graph in the corresponding deflection direction by taking each preset point as a center; and

and selecting the area in the preset graph as the calculation area.

3. The method for determining the profile according to claim 2, wherein the preset pattern is a rectangle, and the setting of the preset pattern with each of the preset points as a center and the corresponding deflection direction comprises:

setting the extending direction of the first edge of the preset graph to be parallel to the tangential direction;

setting half of a second edge of the preset graph to be larger than a maximum offset, wherein the maximum offset is a preset maximum offset between the initial profile and the graph profile; and

and setting an overlapping area of two adjacent preset graphs.

4. The method for determining the contour according to claim 1, wherein the determining the fitting point in each of the calculation regions according to the distribution information of the feature values of the image pixels in each of the calculation regions comprises:

in each calculation region, calculating an accumulated value of the characteristic values of the pixels along a tangential direction, or an average value of the accumulated values, wherein the tangential direction is the tangential direction of the initial contour at the preset point;

calculating a gradient value of the accumulated value or the average value in a direction perpendicular to the tangent; and

a fit point in the calculation region is determined based on the gradient values.

5. The method of determining a contour according to claim 4, wherein said determining a fitting point in the calculation region based on the gradient values comprises:

and determining that the sign of the gradient value is a preset sign, and determining that the point at which the absolute value of the gradient value is greater than the gradient threshold value is a fitting point in the calculation region.

6. The method for determining the contour according to any one of claims 1 to 5, wherein the fitting points in the plurality of calculation regions to obtain the figure contour of the target figure comprises:

fitting a plurality of the fitting points to obtain a fitting outline of the target graph;

a calculation step: respectively calculating the distances from a plurality of fitting points to the fitting contour;

a first fitting step: eliminating the fitted points with the distance larger than a preset distance threshold value, and fitting the rest fitted points to obtain a new fitted contour; and

and after the calculating step and the first fitting step are executed circularly for preset times, taking the latest fitting contour as the graphic contour.

7. The method for determining the contour according to any one of claims 1 to 5, wherein the fitting points in the plurality of calculation regions to obtain the figure contour of the target figure comprises:

a second fitting step: according to the size relation between the distance and the distance threshold value, setting a fitting weight for each fitting point, and according to the fitting weight, re-fitting a plurality of fitting points to obtain a new fitting contour; and

and after the calculating step and the second fitting step are executed circularly for preset times, taking the latest fitting contour as the graphic contour.

8. A contour determining apparatus for determining a figure contour of a target figure in an image, the contour determining apparatus comprising:

the acquisition module is used for acquiring the initial contour of the target graph according to the shape information and the initial position information of the target graph;

a selecting module, configured to select a plurality of calculation regions with each of a plurality of preset points on the initial contour as a center, respectively;

the determining module is used for determining a fitting point in each calculation region according to the distribution information of the characteristic values of the image pixels in each calculation region; and

and the fitting module is used for fitting the fitting points in the plurality of calculation regions to obtain the graph outline of the target graph.

9. A detection apparatus, the detection apparatus comprising:

the imaging device is used for shooting an image of the piece to be detected;

a memory for storing the image; and

a processor communicatively coupled to the memory, the processor configured to perform the method of determining a contour of any of claims 1-7.

10. A non-transitory computer-readable storage medium storing a computer program which, when executed by one or more processors, causes the processors to perform the method of determining a contour of any one of claims 1 to 7.