A method, system and storage medium for judging sample boundary quality

Abstract

The present invention discloses a sample boundary quality judgment method, system and storage medium, belonging to the field of 3D printing technology. The method comprises: S1: boundary quality calibration stage; S2: boundary feature extraction stage, judging the boundary position of a standard raw embryo image data set and a non-standard raw embryo image data set to obtain a boundary information matrix; S3: boundary judgment value fitting stage, using the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain a boundary point set, and obtaining the residual value of each image data in the boundary information matrix through the boundary point set, and taking the largest residual value in the standard raw embryo image data set as the boundary judgment value; S4: boundary quality judgment stage. The subjectivity of human judgment is eliminated, the accuracy of evaluation is greatly improved, and the evaluation result is more objective and accurate; there is no need to rely on experienced evaluators, which reduces the dependence on human resources, and also reduces the cost of training and maintaining the evaluation team.

Description

Technical Field

[0001] The present invention relates to the technical field of 3D printing, and in particular to a sample boundary quality judgment method, system and storage medium. Background Art

[0002] In the field of 3D printing technology, with the rapid development and widespread application of technology, the control of product quality has become the focus of the industry. Traditional 3D printing product quality assessment mainly relies on the experience and judgment of professionals, especially the evaluation of product boundary quality. Although this method can reflect the quality status of the product to a certain extent, it is highly subjective and its accuracy is limited by the experience and skill level of the evaluator, making it difficult to ensure the consistency and accuracy of the evaluation results.

[0003] The evaluation methods in the existing technology mainly have the following problems: first, they are highly subjective, and there may be large differences in judgment between different evaluators; second, they are inefficient and rely on experienced evaluators, making it difficult to conduct evaluations quickly and in large quantities; third, they are costly, and it requires a large investment to cultivate and maintain a high-level quality assessment team.

[0004] In the current 3D printing industry, the evaluation of product boundary quality is particularly critical because it directly affects the overall performance and reliability of the product. However, due to the existence of the above problems, the existing technology has obvious deficiencies in ensuring product boundary quality. As the market's requirements for 3D printed product quality continue to increase, traditional subjective evaluation methods can no longer meet the needs of industry development. Summary of the invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a sample boundary quality judgment method, system and storage medium.

[0006] The object of the present invention is achieved through the following technical solutions: The first aspect of the present invention provides: a method for determining the quality of a sample boundary, comprising the following steps:

[0007] S1: In the boundary quality calibration stage, a raw embryo image dataset to be calibrated is obtained, and the raw embryo image dataset to be calibrated is divided into a standard raw embryo image dataset and a non-standard raw embryo image dataset according to a boundary quality standard value;

[0008] S2: Boundary feature extraction stage, determining the boundary position of the standard raw embryo image data set and the non-standard raw embryo image data set to obtain the boundary information matrix;

[0009] S3: Boundary judgment value fitting stage, using the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain a boundary point set, and using the boundary point set to obtain the residual value of each image data in the boundary information matrix, the largest residual value in the standard green embryo image data set is taken as the boundary judgment value;

[0010] S4: Boundary quality judgment stage, input the image to be judged, when the residual value of the image to be judged is greater than the boundary judgment value, it is judged that its boundary quality is unqualified; when the residual value of the image to be judged is less than the boundary judgment value, it is judged that its boundary quality is qualified.

[0011] Preferably, the boundary of the raw embryo to be calibrated in the raw embryo image data set to be calibrated is parallel to a certain edge of the raw embryo image data to which it belongs.

[0012] Preferably, the said S2: boundary feature extraction stage further comprises the following steps:

[0013] Determine the boundary position z of the target image:

[0014]

[0015] In the formula, f is the distribution function of the line grayscale sum of the target image, i , j are the horizontal and vertical values ​​of the target image respectively;

[0016] When the boundary of the green embryo to be calibrated is parallel to the longitudinal edge of the green embryo image data to which it belongs, the distribution of the line grayscale sum of its boundary is only j Related, let the gray value matrix of the target image be , where n=0,1,2…, N ,m=0,1,2…, M , N , M are all positive integers, we get:

[0017]

[0018]

[0019] In the formula, J is the vertical value of the boundary position in the target image, Max is the maximum value function, dz is the differential;

[0020] Use matrix B with i rows and k columns to record the boundary positions:

[0021]

[0022] In the formula, m1 =0,1,2,3…,23;

[0023] Then use the filter convolution kernel C to filter the target image and get the boundary information matrix :

[0024]

[0025] In the formula, n 1 =0,1,2…,N- k 1 +1, m 2 =0,1,2…,24- l 1 , C is the filter convolution kernel.

[0026] Preferably, the step S3: boundary judgment value fitting stage further includes the following steps:

[0027] Use the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain the boundary point set L :

[0028]

[0029] Then construct a function :

[0030]

[0031] Solution The minimum value of the boundary point set L The straight line with the smallest residual value , and we get:

[0032]

[0033] in yes j The average value of

[0034] straight line With boundary point set L The residual value of v :

[0035]

[0036] Obtain the residual value of each image data in the boundary information matrix v , take the largest residual value in the standard embryo image data set as the boundary judgment value v 0 .

[0037] The second aspect of the present invention provides: a sample boundary quality judgment system, which is used to implement any of the above-mentioned sample boundary quality judgment methods, comprising:

[0038] A boundary quality calibration module is used to obtain a raw embryo image data set to be calibrated, and divide the raw embryo image data set to be calibrated into a standard raw embryo image data set and a non-standard raw embryo image data set according to a boundary quality standard value;

[0039] A boundary feature extraction module is used to determine the boundary position of the standard raw embryo image data set and the non-standard raw embryo image data set to obtain a boundary information matrix;

[0040] The boundary judgment value fitting module is used to establish a set of data pairs using the vertical value and gray value of the data point in the boundary information matrix to obtain a boundary point set, obtain the residual value of each image data in the boundary information matrix through the boundary point set, and take the largest residual value in the standard green embryo image data set as the boundary judgment value;

[0041] The boundary quality judgment module is used to input the image to be judged. When the residual value of the image to be judged is greater than the boundary judgment value, it is judged that its boundary quality is unqualified; when the residual value of the image to be judged is less than the boundary judgment value, it is judged that its boundary quality is qualified.

[0042] The third aspect of the present invention provides: a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by a processor, any of the above-mentioned sample boundary quality judgment methods is implemented.

[0043] The beneficial effects of the present invention are:

[0044] 1) An algorithm that can convert the experience of professionals into purely objective numbers can greatly improve the accuracy compared to the existing technology that only relies on human subjective judgment to determine whether the boundary quality of 3D printed products is qualified, and there is no need to rely on experienced evaluators, making the company's production system more scientific and standardized.

[0045] 2) It eliminates the subjectivity of human judgment, greatly improves the accuracy of evaluation, and makes the evaluation results more objective and accurate; it does not need to rely on experienced evaluators, reduces the dependence on human resources, and also reduces the cost of training and maintaining the evaluation team; the application of the present invention makes the quality control of 3D printed products more scientific and standardized, which helps to improve the overall production efficiency and product quality stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of the sample boundary quality judgment method;

[0047] Figure 2This is a comparison chart between the standard green embryo image and the non-standard green embryo image;

[0048] Figure 3 Schematic diagram of the boundaries of ten calibrated green embryo images in the embodiment;

[0049] Figure 4 Schematic diagram of boundary information of ten calibrated green embryo images in the embodiment;

[0050] Figure 5 These are two images of green embryos to be judged. DETAILED DESCRIPTION

[0051] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0052] See also Figure 1-Figure 5 The first aspect of the present invention provides: a method for determining the quality of a sample boundary, comprising the following steps:

[0053] S1: In the boundary quality calibration stage, a raw embryo image dataset to be calibrated is obtained, and the raw embryo image dataset to be calibrated is divided into a standard raw embryo image dataset and a non-standard raw embryo image dataset according to a boundary quality standard value;

[0054] S2: Boundary feature extraction stage, determining the boundary position of the standard raw embryo image data set and the non-standard raw embryo image data set to obtain the boundary information matrix;

[0055] S3: Boundary judgment value fitting stage, using the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain a boundary point set, and using the boundary point set to obtain the residual value of each image data in the boundary information matrix, the largest residual value in the standard green embryo image data set is taken as the boundary judgment value;

[0056] S4: Boundary quality judgment stage, input the image to be judged, when the residual value of the image to be judged is greater than the boundary judgment value, it is judged that its boundary quality is unqualified; when the residual value of the image to be judged is less than the boundary judgment value, it is judged that its boundary quality is qualified.

[0057] In this embodiment, if Figure 2 As shown in the figure, to fit the boundary judgment value, it is first necessary to take multiple pictures of the green embryo boundary under fixed shooting conditions, and then obtain the boundary quality standard value manually or in advance to calibrate the green embryo boundary quality in the picture. The image with good boundary quality is the standard green embryo image, and the image with bad boundary quality is the non-standard green embryo image. The standard green embryo image is calibrated as 1, and the non-standard green embryo image is calibrated as -1.

[0058] In some embodiments, the boundary of the raw embryo to be calibrated in the raw embryo image data set to be calibrated is parallel to a certain edge of the raw embryo image data to which it belongs.

[0059] In this embodiment, in order to simplify calculation, when photographing the boundary of the green body, the boundary of the green body block is made parallel to a certain side of the picture.

[0060] In some embodiments, the S2: boundary feature extraction stage further includes the following steps:

[0061] Determine the boundary position z of the target image:

[0062]

[0063] In the formula, f is the distribution function of the line grayscale sum of the target image, i , j are the horizontal and vertical values ​​of the target image respectively;

[0064] When the boundary of the green embryo to be calibrated is parallel to the longitudinal edge of the green embryo image data to which it belongs, the distribution of the line grayscale sum of its boundary is only j Related, let the gray value matrix of the target image be , where n=0,1,2…, N ,m=0,1,2…, M , N , M are all positive integers, we get:

[0065]

[0066]

[0067] In the formula, J is the vertical value of the boundary position in the target image, Max is the maximum value function, dz is the differential;

[0068] Use matrix B with i rows and k columns to record the boundary positions:

[0069]

[0070] In the formula, m 1 =0,1,2,3…,23;

[0071] Then use the filter convolution kernel C to filter the target image and get the boundary information matrix :

[0072]

[0073] In the formula, n 1 =0,1,2…,N- k 1 +1, m 2 =0,1,2…,24- l 1 , C is the filter convolution kernel.

[0074] In this embodiment, if Figure 2 As shown, the boundary of the embryo is parallel to the vertical edge of the image. At this time, the distribution of the grayscale and the line of the boundary is only related to j The filter convolution kernel C is used to filter out the redundant information in the image. When K=L=3, .

[0075] In some embodiments, the S3: boundary judgment value fitting stage further includes the following steps:

[0076] Use the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain the boundary point set L :

[0077]

[0078] Then construct a function :

[0079]

[0080] Solution The minimum value of the boundary point set L The straight line with the smallest residual value , and we get:

[0081]

[0082] in yes j The average value of

[0083] straight line With boundary point set L The residual value of v :

[0084]

[0085] Obtain the residual value of each image data in the boundary information matrix v , take the largest residual value in the standard embryo image data set as the boundary judgment value v 0 .

[0086] In this embodiment, since it is a univariate distribution, the residuals obtained from different images can be v Arrange in ascending order, and then find the result with the largest residual value when the boundary quality is calibrated as 1 as the boundary judgment value v 0 When judging a new photo, its residual v Greater than v 0 When , the boundary quality is judged to be unqualified; its residual v Less than v 0 When , the boundary quality is judged to be qualified.

[0087] An embodiment is given below. Figure 3 As shown, there are ten photos of the boundaries of the raw embryo. The boundary with good quality is y= +1, and the boundary with poor quality is y= -1.

[0088] for Figure 3 Mesoblast a, i={1,2,3,…,336}, j={1,2,3,…,484}, J=449;

[0089] for Figure 3 Mesoembryo b, i={1,2,3,…,289}, j={1,2,3,…493}, J=438;

[0090] for Figure 3 Mesoblast c, i={1,2,3,…,298}, j={1,2,3,…,454}, J=436;

[0091] for Figure 3 Mesoembryo d, i={1,2,3,…,313}, j={1,2,3,…,483}, J=455;

[0092] for Figure 3 Mesoembryo e, i={1,2,3,…,283}, j={1,2,3,…,482}, J=449;

[0093] for Figure 3 Mesoembryo f, i={1,2,3,…,334}, j={1,2,3,…,503}, J=465;

[0094] for Figure 3 Mesoblast g, i={1,2,3,…,393}, j={1,2,3,…,600}, J=444;

[0095] for Figure 3 Mesoembryo h, i = {1, 2, 3, …, 324}, j = {1, 2, 3, …, 462}, J = 573;

[0096] for Figure 3 Mesoblast i, i={1,2,3,…,333}, j={1,2,3,…,499}, J=448;

[0097] for Figure 3 Mesoblast j, i={1,2,3,…,314}, j={1,2,3,…,600}, J=568;

[0098] like Figure 4 As shown, according to formulas (1-1, 1-2, 1-3, 1-4, 1-5), the boundary information of ten raw embryos can be obtained: .

[0099] for Figure 3 Mesoblast a, L a ={(7,0),(7,10)…,(7,260)};

[0100] for Figure 3 Mesoblastic b, with L b ={(10,0),(9,10)…,(17,260)};

[0101] for Figure 3 Mesoblast c, L c ={(14,0),(11,10)…,(4,260)};

[0102] for Figure 3 Mesozoic embryo d, L d ={(11,0),(8,10)…,(9,260)};

[0103] for Figure 3 Mesozoic embryo, L e ={(12,0),(12,10)…,(10,260)};

[0104] for Figure 3 Mesoblastic embryo f, L f ={(12,0),(10,10)…,(7,260)};

[0105] for Figure 3 Mesozoic embryo g, L g ={ (10,0),(9,10)…,(6,260)};

[0106] for Figure 3Mesoblast h, L h ={(11,0),(10,10)…,(6,260)};

[0107] for Figure 3 Mesozoic embryo i, L i ={(15,10),(14,10)…,(4,260)};

[0108] for Figure 3 Mesozoic embryo j, L j ={(16,10),(14,10)…,(6,260)};

[0109] According to formulas (1-6, 1-7, 1-8), all w 0 and b 0 :

[0110] for Figure 3 Mesoblast a, ;

[0111] for Figure 3 Mesoblastic b, ;

[0112] for Figure 3 Mesoblast c, ;

[0113] for Figure 3 Mesoblastic embryo d, ;

[0114] for Figure 3 Mesozoic embryo, ;

[0115] for Figure 3 Mesoblastic embryo f, ;

[0116] for Figure 3 Mesozoic embryo g, ;

[0117] for Figure 3 Mesoblast h, ;

[0118] for Figure 3 Mesoblast i, ;

[0119] for Figure 3 Mesozoic J, ;

[0120] According to formula (1-9), calculate all v :

[0121] for Figure 3 Mesoblast a, ;

[0122] for Figure 3 Mesoblastic b, ;

[0123] for Figure 3 Mesoblast c, ;

[0124] for Figure 3 Mesoblastic embryo d, ;

[0125] for Figure 3 Mesozoic embryo, ;

[0126] for Figure 3 Mesoblastic embryo f, ;

[0127] for Figure 3 Mesozoic embryo g, ;

[0128] for Figure 3 Mesoblast h, ;

[0129] for Figure 3 Mesoblast i, ;

[0130] for Figure 3 Mesozoic J, ;

[0131] Finally, by arranging the residual values ​​of different pictures in ascending order, it can be found that the largest residual value among the qualified boundaries is 18.218. In summary, through the above ten pictures, the boundary judgment value of the green boundary quality is obtained. v 0 =18.218, then we can judge the value based on the boundary v 0 , to judge the quality of the green surface. Figure 5 Shown are two green embryo boundaries to be tested.

[0132] for Figure 5 The test embryo boundary 1 (the embryo marked as 1 on the left), i={1,2,3,…,399},j={1,2,3,…,600}, J=573;

[0133] for Figure 5The test embryo boundary 2 (the embryo marked as 2 on the right), i={1,2,3,…,348},j={1,2,3,…576}, J=547;

[0134] According to formulas (1-1, 1-2, 1-3, 1-4, 1-5), we can get ;

[0135] for Figure 5 The test embryo boundary 1 (the embryo marked as 1 on the left), ;

[0136] for Figure 5 The embryo boundary 2 to be tested (the embryo marked as 2 on the right), ;

[0137] According to formula (1-9), calculate all v :

[0138] for Figure 5 The test embryo boundary 1 (the embryo marked as 1 on the left), ;

[0139] for Figure 5 The embryo boundary 2 to be tested (the embryo marked as 2 on the right), ;

[0140] because, v 1 < v 0 , v 2 > v 0 Then you can get Figure 5 The boundary 1 of the green embryo to be tested (the green embryo marked as 1 on the left) is the boundary of qualified quality. Figure 5 Boundary 2 of the green embryo to be tested (the green embryo marked as 2 on the right) is a boundary with unqualified quality.

[0141] The second aspect of the present invention provides: a sample boundary quality judgment system, which is used to implement any of the above-mentioned sample boundary quality judgment methods, comprising:

[0142] A boundary quality calibration module is used to obtain a raw embryo image data set to be calibrated, and divide the raw embryo image data set to be calibrated into a standard raw embryo image data set and a non-standard raw embryo image data set according to a boundary quality standard value;

[0143] A boundary feature extraction module is used to determine the boundary position of the standard raw embryo image data set and the non-standard raw embryo image data set to obtain a boundary information matrix;

[0144] The boundary judgment value fitting module is used to establish a set of data pairs using the vertical value and gray value of the data point in the boundary information matrix to obtain a boundary point set, obtain the residual value of each image data in the boundary information matrix through the boundary point set, and take the largest residual value in the standard green embryo image data set as the boundary judgment value;

[0145] The boundary quality judgment module is used to input the image to be judged. When the residual value of the image to be judged is greater than the boundary judgment value, it is judged that its boundary quality is unqualified; when the residual value of the image to be judged is less than the boundary judgment value, it is judged that its boundary quality is qualified.

[0146] The third aspect of the present invention provides: a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by a processor, any of the above-mentioned sample boundary quality judgment methods is implemented.

[0147] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for judging sample boundary quality, characterized in that: The following steps are involved: S1: In the boundary quality calibration stage, a raw embryo image dataset to be calibrated is obtained, and the raw embryo image dataset to be calibrated is divided into a standard raw embryo image dataset and a non-standard raw embryo image dataset according to a boundary quality standard value; S2: Boundary feature extraction stage, determining the boundary position of the standard raw embryo image data set and the non-standard raw embryo image data set to obtain the boundary information matrix; S3: Boundary judgment value fitting stage, using the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain a boundary point set, and using the boundary point set to obtain the residual value of each image data in the boundary information matrix, the largest residual value in the standard green embryo image data set is taken as the boundary judgment value; S4: Boundary quality judgment stage, input the image to be judged, when the residual value of the image to be judged is greater than the boundary judgment value, it is judged that its boundary quality is unqualified; when the residual value of the image to be judged is less than the boundary judgment value, it is judged that its boundary quality is qualified; The boundary quality standard value needs to be obtained in advance; The S2: boundary feature extraction stage further includes the following steps: Determine the boundary position z of the target image: In the formula, f is the distribution function of the line grayscale sum of the target image, i , j are the horizontal and vertical values ​​of the target image respectively; When the boundary of the green embryo to be calibrated is parallel to the longitudinal edge of the green embryo image data to which it belongs, the distribution of the line grayscale sum of its boundary is only j Related, let the gray value matrix of the target image be , where n=0,1,2…, N ,m=0,1,2…, M , N , M are all positive integers, we get: ‘ ； In the formula, J is the vertical value of the boundary position in the target image, Max is the maximum value function, dz is the differential; Use matrix B with i rows and k columns to record the boundary positions: ； In the formula, m 1=0,1,2,3…,23; Then use the filter convolution kernel C to filter the target image and get the boundary information matrix : ； In the formula, n 1=0,1,2…,N- k 1+1, m 2=0,1,2…,24- l 1, C is the filter convolution kernel; The step S3: boundary judgment value fitting stage further includes the following steps: Use the vertical value and gray value of the data point in the boundary information matrix to establish a set of data pairs to obtain the boundary point set L : ； Then construct a function : ； Solution The minimum value of the boundary point set L The straight line with the smallest residual value , and we can solve for: ； in yes j The average value of straight line With boundary point set L The residual value of v : ； Obtain the residual value of each image data in the boundary information matrix v , take the largest residual value in the standard embryo image data set as the boundary judgment value v 0.

2. The sample boundary quality judgment method according to claim 1, characterized in that: The boundary of the raw embryo to be calibrated in the raw embryo image data set to be calibrated is parallel to a certain edge of the raw embryo image data to which it belongs.

3. A sample boundary quality judgment system, characterized in that: The method for determining the sample boundary quality as claimed in claim 1 or 2 comprises: A boundary quality calibration module is used to obtain a raw embryo image data set to be calibrated, and divide the raw embryo image data set to be calibrated into a standard raw embryo image data set and a non-standard raw embryo image data set according to a boundary quality standard value; A boundary feature extraction module is used to determine the boundary position of the standard raw embryo image data set and the non-standard raw embryo image data set to obtain a boundary information matrix; The boundary judgment value fitting module is used to establish a set of data pairs using the vertical value and gray value of the data point in the boundary information matrix to obtain a boundary point set, obtain the residual value of each image data in the boundary information matrix through the boundary point set, and take the largest residual value in the standard green embryo image data set as the boundary judgment value; The boundary quality judgment module is used to input the image to be judged. When the residual value of the image to be judged is greater than the boundary judgment value, it is judged that its boundary quality is unqualified; when the residual value of the image to be judged is less than the boundary judgment value, it is judged that its boundary quality is qualified.

4. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by the processor, the sample boundary quality judgment method as claimed in claim 1 or 2 is implemented.

Images