Light alloy casting flaw detection shooting, evaluation and positioning marking full-process detection method

Abstract

The invention belongs to the related field of casting quality detection technology, and discloses a light alloy casting flaw detection shooting, evaluation and positioning marking full-process detection method, which comprises the following steps: acquiring a real flaw detection imaging image and a three-dimensional model of a casting; the real flaw detection imaging image is preprocessed, and then identification of the typecasting of the single piece number is executed; generating a virtual defect-free flaw detection image, differencing the virtual defect-free flaw detection image and a real flaw detection imaging image, and positioning to obtain the position of a defect; sequentially carrying out secondary detection, defect false removal, category correction and hazard quantitative rating on real flaw detection defects; a real flaw detection imaging image is converted into a three-dimensional image from a two-dimensional image, the three-dimensional position and depth information of a real flaw detection defect is obtained, and then the real defect is correspondingly marked. According to the invention, intelligent detection in the whole process can be realized, and the system has the advantages of high efficiency, high precision, stability, reliability and the like.

Description

Technical Field

[0001] The present invention belongs to the field related to casting quality detection technology, and more specifically, relates to a full-process detection method for light alloy castings, including flaw detection, filming, film evaluation and positioning marking. Background Art

[0002] With the rapid development of advanced equipment manufacturing industry, the quality control requirements of light alloy metal components are getting higher and higher. These components often have complex designs, and these design features greatly increase the difficulty of internal defect detection. In the existing technology, the commonly used detection process is mainly: "X-ray flaw detection two-dimensional DR manual partition adjustment parameter filming + manual visual experience evaluation + manual trial and error three-dimensional positioning".

[0003] However, further research shows that the above-mentioned prior art still has the following defects or shortcomings:

[0004] First, in the filming process, the manual method often has problems such as missing key parts of castings and large fluctuations in imaging quality. Secondly, in the film evaluation process, due to the mixture of various defects and the similar morphology of various defects, defects are easily missed or misdetected. Finally, in the positioning process, due to the unknown depth and size of the defects, it is easy to have problems such as low-quality and low-efficiency positioning and repair, and possible secondary damage to the castings. In addition, the filming, film evaluation and positioning links are often completed by different personnel, and there is a lack of close connection between the links, resulting in a disconnected process, further reducing the overall detection efficiency and reliability.

[0005] Accordingly, many problems existing in existing detection methods have become the main bottleneck restricting the intelligent, efficient and high-quality batch production of high-end equipment. This field urgently needs further research and improvement in order to better meet the quality requirements of modern casting. Summary of the invention

[0006] In response to one or more of the above defects or needs in the prior art, the present invention provides a full-process inspection method for flaw detection, filming, film evaluation and positioning marking of light alloy castings, wherein the entire inspection process including flaw detection, filming, film evaluation and positioning marking is redesigned and targeted improvements are made to key steps by fully combining the processing characteristics and specific needs of light alloy castings, thereby realizing intelligent inspection of the entire process, and having the advantages of high efficiency, high precision, stability and reliability, and is therefore particularly suitable for quality inspection applications of light alloy castings with various complex structures.

[0007] To achieve the above object, according to the present invention, a full-process detection method for flaw detection, filming, film evaluation and positioning marking of light alloy castings is provided, characterized in that the method comprises the following steps:

[0008] S1. For the light alloy casting to be inspected, obtain the real flaw detection imaging images at different flaw detection angles relative to the radiation source, and simultaneously obtain the corresponding three-dimensional model of the casting;

[0009] S2, performing image preprocessing on the real flaw detection imaging image, and then performing recognition of the single piece number casting character;

[0010] S3, constructing a virtual imaging system to generate a flawless flaw detection image; performing grayscale fitting and difference calculation on the flawless flaw detection image and the real flaw detection imaging image to locate the position of the real flaw detection defect;

[0011] S4. Perform secondary inspection on the obtained real flaw detection defects to increase the number of real flaw detection defects and classify the real flaw detection defects;

[0012] S5. Defect false detection and category correction are performed on the real flaw detection defects obtained;

[0013] S6. Carry out quantitative hazard rating on the actual flaws obtained;

[0014] S7, converting the real flaw detection imaging image from two dimensions into three dimensions, and obtaining the three-dimensional position and depth information of the real flaw detection defect;

[0015] S8. Based on the three-dimensional position and depth information of the real flaw detection defect, the defect is marked using casting defect marking equipment, thereby completing the entire detection process.

[0016] As a further preferred embodiment of the present invention, in step S2, the image preprocessing is preferably implemented by a homomorphic filtering method and a local adaptive binarization method.

[0017] As a further preferred embodiment of the present invention, in step S2, the recognition process of the single piece number casting is preferably designed as follows:

[0018] S21, detecting the edge of the character in the real flaw detection imaging image and generating a binary image containing edge information; extracting the outline information of the character, i.e., the character contour line, from the binary image, and then finding its minimum circumscribed rectangular frame according to the contour information, performing a segmentation operation to obtain a minimum unit character image;

[0019] S22, build a deep learning model and perform recognition training on all characters in advance;

[0020] S23, dividing the real flaw detection imaging image into multiple areas, and then using the above-mentioned trained deep learning model to perform optical character recognition of the single piece number casting inside each area, and identify the specific content of the single piece number casting.

[0021] As a further preferred embodiment of the present invention, in step S2, it is preferred to also include realizing automatic mapping operations of each process of flaw detection filming, film evaluation and positioning marking with casting models and batches based on the recognition result of the single piece number casting.

[0022] As a further preferred embodiment of the present invention, in step S3, before using the virtual imaging system to generate the defect-free flaw detection image, it is preferred to further include a filming path planning operation, and the process is preferably designed as follows:

[0023] a. The stress concentration areas and defect-prone areas of the casting are taken as key quality control points, and the features of these key quality control points and similar structures are extracted and searched, and then they are used as the necessary points of the filming path of virtual imaging;

[0024] b. Classify the three-dimensional models of castings, and make castings classified into the same category share the same points on the filming path;

[0025] c. According to the filming path including the necessary points of the above-mentioned filming path and the corresponding filming parameters, a complete defect-free flaw detection image is generated using the virtual imaging system.

[0026] As a further preferred embodiment of the present invention, in step S3, the grayscale fitting difference calculation process is preferably designed as follows:

[0027] S31, respectively calculating the grayscale histograms of the defect-free flaw detection image and the real flaw detection imaging image;

[0028] S32, using the grayscale histogram, respectively calculating the number and proportion of pixels in the defect-free flaw detection image and the real flaw detection imaging image whose grayscale level is less than a preset value, and simultaneously constructing a mapping relationship between the two images;

[0029] S33, based on the mapping relationship, converting the gray level of the defect-free flaw detection image to obtain a fitted defect-free flaw detection image;

[0030] S34, performing pixel difference comparison between the fitted non-defective flaw detection image and the real flaw detection imaging image, and the area where the pixel difference is greater than the threshold is the location of the real flaw detection defect.

[0031] As a further preference of the present invention, in step S4, it is preferred to establish a network model for deep learning of defect detection, and increase the number of defects by means of casting simulation and virtual flaw detection simulation. At the same time, in the network model, the defects are secondary located and classified according to the defect generation mechanism and defect morphology characteristics.

[0032] As a further preference of the present invention, in step S5, it is preferred to construct a corresponding defect multidimensional feature matrix based on features such as defect contour, internal morphology and grayscale value, and use the feature matrix to implement the above-mentioned defect de-aliasing and category correction operations.

[0033] As a further preferred embodiment of the present invention, in step S5, it is preferred to further include a method of automatically associating defects of the same category in multi-position flaw detection images based on the diverse hierarchical features of the defects.

[0034] As a further preferred embodiment of the present invention, in step S6, it is preferred to extract information such as grayscale, texture, area and discreteness of the defect, and implement the above-mentioned quantitative hazard rating operation based on this information.

[0035] In general, the above technical solution conceived by the present invention has the following technical advantages compared with the prior art:

[0036] 1. The present invention redesigns the entire detection process including flaw detection filming, film evaluation and positioning marking and makes targeted improvements to key steps, thereby realizing intelligent detection of the entire process and having the advantages of high efficiency, high precision, stability and reliability;

[0037] 2. The present invention further optimizes the design of the identification process of single-piece casting characters, which can realize the identification of casting characters and contents more accurately and quickly, and can also realize the automatic mapping of flaw detection film shooting, film evaluation and positioning marking with casting models and batches;

[0038] 3. The present invention can realize accurate positioning of defects and detection of all key parts of castings by establishing a virtual imaging system and comparing non-virtual and real flaw detection imaging images; in addition, the present invention also makes targeted designs for related algorithms of grayscale fitting and difference calculation and filming path planning in the virtual imaging process, which can further improve the accuracy of defect positioning;

[0039] 4. The present invention further constructs a multi-dimensional feature matrix of defects to achieve defect elimination and category correction, effectively solving the problem of missed defect detection; combines casting knowledge, production requirements and relevant standards to design defect hazard assessment indicators, and achieves quantitative rating of defect hazards;

[0040] 5. The whole process detection method of the present invention is easy to operate as a whole, and human errors are avoided to the greatest extent. It can realize the X-ray flaw detection of light alloy castings with "complete, clear, accurate, stable, detailed, fast and accurate positioning". Therefore, it is particularly suitable for high-reliability detection applications such as large light alloy complex castings for high-end national defense equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1It is a flow chart of the whole process detection method of light alloy casting flaw detection filming, film evaluation and positioning marking according to the present invention;

[0042] Figure 2 It is an application scenario diagram for exemplifying the full-process detection method of the present invention;

[0043] Figure 3 It is an application scenario diagram used to demonstrate the linkage between the virtual imaging system and the flaw detection equipment for filming;

[0044] Figure 4 This is an application scenario diagram used to demonstrate the reconstruction and positioning of defect three-dimensional morphology. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0046] Figure 1 This is a flow chart of the whole process detection method of light alloy casting flaw detection, filming, film evaluation and positioning marking according to the present invention. Figure 1 To explain the present invention in more detail.

[0047] Step 1: For the light alloy casting to be inspected, obtain the real flaw detection imaging images at different flaw detection angles relative to the radiation source, and simultaneously obtain the corresponding casting three-dimensional model.

[0048] In this step, the radiation source may be, for example, X-rays, and all critical parts of the casting may be inspected at different flaw detection angles.

[0049] Step 2: Preprocess the real flaw detection imaging image and then perform identification of the casting number.

[0050] In this step, according to a preferred embodiment of the present invention, the obtained real flaw detection imaging image can be preprocessed by using a homomorphic filtering method and a local adaptive binarization method, and then the image text, i.e., the single piece number casting character, is recognized as follows:

[0051] First, for example, the Canny edge detection method is used to detect the edges of the text in the image and generate a binary image containing edge information. Based on the edge image, the contour information of the characters, that is, the character contour lines, are extracted using the contour extraction algorithm. These contour lines are the boundaries between the characters and the background. Based on the contour lines of the characters, the minimum circumscribed rectangular frame of the contour is found, and then the segmentation operation is performed along the minimum circumscribed rectangular frame to obtain the smallest unit character image;

[0052] Next, a deep learning model can be constructed to pre-train all characters for recognition. For example, a convolutional neural network model can be constructed based on character contours and the pre-processed relevant data can be input into the model for training to increase the accuracy of letter recognition. The model principle and specific process are well known in the art and will not be described in detail.

[0053] Next, the real flaw detection imaging image is divided into multiple regions, and then the optical character recognition of the single piece number casting is performed in each region using the above-mentioned trained deep learning model, and the specific content of the single piece number casting is identified. In addition, it can also realize the automatic mapping of film shooting, film evaluation, positioning information and casting model, batch, etc.

[0054] Step three, construct a virtual imaging system and generate a flawless flaw detection image; perform grayscale fitting and difference calculation on the flawless flaw detection image and the real flaw detection imaging image to locate the position of the real flaw detection defect.

[0055] In this step, a virtual imaging system can be constructed first. Specifically, for example, the physical imaging process and principle can be used as a blueprint to simulate the processes of ray emission, penetration, and imaging plate reception, so as to achieve complete data tracking and filming simulation for the abstract filming process, and build a model that can generate high-quality defect images. As those skilled in the art will appreciate, there are many ways to implement the virtual imaging system, and it is preferred to use the technical solutions disclosed in the accepted patents CN202310335552.7 and "A method for generating virtual radiographic images for internal defect detection of castings" to implement it.

[0056] Next, according to a preferred embodiment of the present invention, a filming path planning operation may also be included, and the process is preferably designed as follows:

[0057] a. Take the stress concentration areas and defect-prone areas of the casting as key quality control points, extract and search for features of these key quality control points and similar structures, and then use them as the necessary points of the virtual imaging filming path; for example, finite element analysis (FEA) can be used to predict potential stress concentration areas, and analyze and find out the areas where defects frequently occurred in the past as key quality control points of the casting;

[0058] b. Classify the three-dimensional models of castings, and make the castings classified into the same category share the same points on the filming path; for example, a three-dimensional model classification algorithm for castings can be established to calculate the similarity between the three-dimensional model of castings and different model categories, and complete the classification of the three-dimensional model through operator spectrum analysis;

[0059] c. Use the virtual imaging system to generate complete flawless flaw detection images according to the filming path that includes the necessary points of the above filming path and the supporting filming parameters. For example, you can formulate the virtual system-flaw detection equipment linkage criteria, generate complete filming images and shooting path previews in the virtual imaging system, use the recommended path to program the filming equipment, control the movement of the flaw detection equipment, and realize the automatic conversion of the recommended flaw detection strategy and CNC numerical control flaw detection.

[0060] Next, grayscale fitting is performed on the defect-free flaw detection image and the real flaw detection imaging image to obtain the position of the real flaw detection defect.

[0061] More specifically, according to another preferred embodiment of the present invention, the process can be designed as follows:

[0062] Calculate the grayscale histogram of the virtual flaw detection image and the real flaw detection image. For example, if the grayscale level of the image is r, and the number of pixels with grayscale level r in the image is h(r), then the grayscale histogram can be expressed as H(r) = h(r);

[0063] Suppose the gray level of the image is r, and the number of pixels with gray level less than or equal to r in the image is c(r), then the cumulative distribution function is Where N is the total number of pixels in the image. Assuming the grayscale level of the virtual flaw detection image is r and the grayscale level of the real flaw detection image is z, a mapping function T(r) can be constructed so that CDF 虚拟 (r)=CDF 真实 (z), thereby realizing the mapping relationship between the two images;

[0064] Perform grayscale fitting to find the difference, use the mapping relationship to convert the grayscale level of the virtual flaw detection image, and then obtain the fitted image. Assume that the pixel value of the flawless flaw detection image is I1(x, y), and the pixel value of the flaw detection image is I2(x, y), then the pixel difference can be expressed as D(x, y) = |I1(x, y)-I2(x, y)|. The area where the pixel difference is greater than the threshold T is the defect area.

[0065] Step 4: Perform secondary inspection on the obtained real flaw detection defects to increase the number of real flaw detection defects and classify the real flaw detection defects.

[0066] In this step, according to another preferred embodiment of the present invention, a network model for deep learning of defect detection can be established, and the number of defects can be increased by using casting simulation and virtual flaw detection simulation. At the same time, in the network model, the defects can be secondary located and classified according to the defect generation mechanism and defect morphological characteristics. Accordingly, problems such as unbalanced defect categories and single morphology can be solved, and problems such as "biased" defect detection models can be avoided, while achieving high-precision defect location and classification.

[0067] Step five, perform defect de-false detection and category correction on the real flaws obtained.

[0068] In this step, according to another preferred embodiment of the present invention, a corresponding multi-dimensional feature matrix of defects can be constructed based on features such as defect contours, internal morphology, and grayscale values, and the feature matrix can be used to implement the above-mentioned defect removal and category correction operations. In addition, image processing and feature extraction methods such as Gaussian filtering, binarization, and morphological processing can be integrated to accurately capture the diverse hierarchical features of defects, create a multi-image same-defect identification algorithm, and realize intelligent automatic association of the same defect in multi-part flaw detection images.

[0069] Step six, conduct quantitative hazard rating on the actual flaws obtained.

[0070] In this step, according to another preferred embodiment of the present invention, information such as the grayscale, texture, area, and degree of discreteness of the defect can be extracted, and the above-mentioned quantitative hazard rating operation can be implemented based on this information. More specifically, for example, different hazard assessment indicators and feature matrices can be designed for each defect in combination with casting knowledge, production requirements, and relevant standards. Denoising, sharpening, and binarization are performed on the defect sub-images to extract information such as the grayscale, texture, area, and degree of discreteness of the defect, so as to implement quantitative hazard rating of the defect.

[0071] The above-mentioned defect multidimensional feature matrix and hazard assessment index matrix can be expressed as follows:

[0072]

[0073] Step seven, converting the real flaw detection imaging image from two dimensions to three dimensions, and obtaining the three-dimensional position and depth information of the real flaw detection defect.

[0074] In this step, there are many ways to construct a three-dimensional model from a two-dimensional image. Preferably, it can be implemented by using the technical solution disclosed in the accepted patent CN202310399158.X, "Three-dimensional intelligent positioning method and system for casting defects based on multi-angle flaw detection images".

[0075] Next, the three-dimensional generation of defects is carried out. For example, the two-dimensional X-ray image of the casting can be converted into three-dimensional point cloud data, the intersection of the ray source and the casting model can be calculated, and the precise position of the intersection can be determined by the ray projection method. On this basis, the boundary detection algorithm is used to identify the key boundary points around the intersection. These boundary points form a characteristic polyhedron, which represents the preliminary three-dimensional outline of the defect. In order to accurately reconstruct the defect morphology, edge fitting technology is used, such as B-spline surface fitting, to smooth and refine these boundary points. Finally, the ray angle is adjusted for multiple projections and optimizations to refine the three-dimensional position and morphology of the defect, and finally the precise three-dimensional position and depth information of the defect is obtained.

[0076] Step eight, based on the three-dimensional position and depth information of the actual flaw detection defect, the defect is marked using casting defect marking equipment, thereby completing the entire detection process.

[0077] In this step, for example, a robotic arm casting defect marking equipment can be used, high-precision sensors and image processing technology can be used to avoid accumulated errors during the marking process, and a visual system can be used to capture subtle surface features of the casting in real time, accurately identify the defect position, depth and other information on the casting, accurately match the actual casting position with the defect information on the casting, and automatically mark the defects.

[0078] In summary, the full-process detection method for flaw detection, filming, film evaluation and positioning marking of light alloy castings according to the present invention can realize intelligent detection of the whole process, and can realize X-ray flaw detection of light alloy castings with "complete, clear, accurate, stable, detailed, fast and accurate positioning". Therefore, it is particularly suitable for high-reliability detection applications such as large light alloy complex castings of high-end national defense equipment and has good practical value and application prospects.

[0079] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A full-process detection method for light alloy casting flaw detection, filming, film evaluation and positioning marking, characterized in that: The method comprises the following steps: S1. For the light alloy casting to be inspected, obtain the real flaw detection imaging images at different flaw detection angles relative to the radiation source, and simultaneously obtain the corresponding three-dimensional model of the casting; S2, performing image preprocessing on the real flaw detection imaging image, and then performing recognition of the single piece number casting character; S3, build a virtual imaging system to generate defect-free flaw detection images; Perform grayscale fitting and difference calculation on the defect-free flaw detection image and the real flaw detection imaging image to locate the position of the real flaw detection defect; S4. Perform secondary inspection on the obtained real flaw detection defects to increase the number of real flaw detection defects and classify the real flaw detection defects; S5. Defect false detection and category correction are performed on the real flaw detection defects obtained; S6. Carry out quantitative hazard rating on the actual flaws obtained; S7, converting the real flaw detection imaging image from two dimensions into three dimensions, and obtaining the three-dimensional position and depth information of the real flaw detection defect; S8. Based on the three-dimensional position and depth information of the real flaw detection defect, the defect is marked using casting defect marking equipment, thereby completing the entire detection process.

2. The method according to claim 1, characterized in that: In step S2, the identification process of the piece number casting is preferably designed as follows: S21, detecting the edges of characters in the real flaw detection imaging image and generating a binary image containing edge information; extracting the contour information of the characters in the binary image, and then finding its minimum circumscribed rectangular frame according to the contour information, performing a segmentation operation to obtain a minimum unit character image; S22, build a deep learning model and perform recognition training on all characters in advance; S23, dividing the real flaw detection imaging image into multiple regions, and then using the above-mentioned trained deep learning model to perform optical character recognition of the single piece number casting inside each region, and identify the specific content of the single piece number casting.

3. The method according to claim 1 or 2, characterized in that In step S2, it is preferred to further include realizing automatic mapping operations of flaw detection filming, film evaluation and positioning marking processes with casting models and batches based on the recognition results of the single piece number casting characters.

4. The method according to any one of claims 1 to 3, characterized in that: In step S3, before using the virtual imaging system to generate the defect-free flaw detection image, it is preferred to further include a filming path planning operation, and the process is preferably designed as follows: a. Take the stress concentration areas and defect-prone areas of the casting as key quality control points, extract and search for the features of these key quality control points and similar structures, and then use them as the necessary points of the virtual imaging path; b. Classify the three-dimensional models of castings, and make castings classified into the same category share the same points on the filming path; c. According to the filming path including the necessary points of the above-mentioned filming path and the corresponding filming parameters, a complete defect-free flaw detection image is generated using the virtual imaging system.

5. The method according to any one of claims 1 to 4, characterized in that: In step S3, the grayscale fitting difference calculation process is preferably designed as follows: S31, respectively calculating the grayscale histograms of the defect-free flaw detection image and the real flaw detection imaging image; S32, using the grayscale histogram, respectively calculating the number and proportion of pixels in the defect-free flaw detection image and the real flaw detection imaging image whose grayscale level is less than a preset value, and simultaneously constructing a mapping relationship between the two images; S33, based on the mapping relationship, converting the gray level of the defect-free flaw detection image to obtain a fitted defect-free flaw detection image; S34, performing pixel difference comparison between the fitted non-defective flaw detection image and the real flaw detection imaging image, and the area where the pixel difference is greater than the threshold is the location of the real flaw detection defect.

6. The method according to any one of claims 1 to 5, characterized in that: In step S4, it is preferred to establish a network model for deep learning of defect detection, and increase the number of defects by using casting simulation and virtual flaw detection simulation. At the same time, in the network model, the defects are secondary located and classified according to the defect generation mechanism and defect morphology characteristics.

7. The method according to any one of claims 1 to 6, characterized in that: In step S5, a corresponding multi-dimensional feature matrix of defects is preferably constructed based on features such as defect contour, internal morphology and grayscale value, and the above-mentioned defect de-aliasing and category correction operations are implemented using the feature matrix.

8. The method according to any one of claims 1 to 7, characterized in that: In step S5, it is preferred to further include a method of automatically associating defects of the same category in multi-position flaw detection images based on the diverse hierarchical features of the defects.

9. The method according to any one of claims 1 to 8, characterized in that: In step S6, it is preferred to extract information such as grayscale, texture, area and discreteness of the defect, and implement the above-mentioned quantitative hazard rating operation based on this information.

Images