Laser selective melting powder bed spreading quality testing method for metal additive manufacturing

Abstract

This invention relates to the field of image processing technology, specifically to a method for detecting the powder bed spreading quality in laser selective melting molding for metal additive manufacturing. The method involves acquiring various local regions of a grayscale image of the powder bed surface, determining the light reflection anomaly coefficient of each local region based on pixel differences to filter out areas with abnormal light reflection; determining the edge detail prominence coefficient and shadow index parameters of these areas, and combining these with the light reflection anomaly coefficient to determine the edge detail loss coefficient; performing edge detection on these areas, correcting the edge point retention rate based on the edge detail loss coefficient, and performing quality inspection based on the edge detection results. This invention improves the accuracy of powder bed spreading quality detection by considering the influence of shadow noise on the powder bed surface and preserving more edge detail information in areas with abnormal light reflection.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and specifically to a method for detecting the quality of laser selective melting powder bed for metal additive manufacturing. Background Technology

[0002] Laser additive manufacturing of metallic materials mainly includes three methods: selective laser melting (SLM), laser metal deposition, and laser near-net-shape forming. SLM is a powder bed-based metal additive manufacturing technology that uses a high-energy laser beam to melt a pre-designed two-dimensional cross-section of metal alloy powder, bonding it with the underlying solidified material, accumulating layer by layer until a three-dimensional solid is formed. During SLM, it is crucial to ensure the quality of each layer of metal alloy powder. However, various defects may occur during powder spreading due to factors such as unstable laser energy and residual printing residue, which can affect the quality and performance of the parts. Therefore, inspecting the powder spreading quality of each layer is particularly important to ensure the quality and performance of the parts.

[0003] Current methods for inspecting the powder bed spreading quality in laser selective melting powder beds typically involve acquiring surface image data of the powder layer, using image processing and pattern recognition to extract and identify defects based on the features of these edge curves, thus achieving powder bed quality inspection. However, because the powder bed surface consists of irregularly shaped particles with varying particle packing states, their light reflection effects differ, resulting in significant shadow noise in the acquired images. This noise obscures the true defect areas, leading to a loss of edge detail information and a substantial error in the final quality inspection results. Summary of the Invention

[0004] The purpose of this invention is to provide a method for detecting the powder bed quality of laser selective melting forming powder bed used in metal additive manufacturing. This method addresses the problem that existing methods suffer from significant errors in powder bed quality detection due to the loss of edge detail information in real defect areas caused by shadow noise in the powder bed surface image.

[0005] To address the aforementioned technical problems, in a first aspect, the present invention provides a method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing, comprising the following steps:

[0006] A grayscale image of the powder-coated surface is obtained, and the grayscale image is segmented to obtain various local regions;

[0007] Based on the pixel differences in different local regions, the light reflection anomaly coefficient of the local region is determined, and based on the light reflection anomaly coefficient, the light reflection anomaly regions in each local region are filtered out.

[0008] The edge features in the abnormal light reflection region are analyzed to determine the edge detail prominence coefficient of the abnormal light reflection region;

[0009] Based on the grayscale correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image, and the difference in the number of edge pixels between the light reflection abnormal region and other adjacent light reflection abnormal regions, the shadow index parameters of the light reflection abnormal region are determined.

[0010] By integrating the light reflection anomaly coefficient, edge detail prominence coefficient, and shadow index parameters of the light reflection anomaly region, the edge detail loss coefficient of the light reflection anomaly region is determined.

[0011] Based on the edge detail loss coefficient, edge detection is performed on the light reflection abnormal area. The larger the edge detail loss coefficient, the greater the amount of edge points retained during the edge detection process. The powder spreading quality is then detected based on the edge detection results.

[0012] In conjunction with the first aspect above, in some possible implementations, determining the light reflection anomaly coefficient of the local region includes:

[0013] Get the average pixel value of each local region, and get the average pixel value of the entire grayscale image;

[0014] The light reflection anomaly coefficient of each local region is determined based on the difference between the average pixel value of each local region and the average pixel value of the entire grayscale image. The larger the difference, the larger the corresponding light reflection anomaly coefficient.

[0015] In conjunction with the first aspect above, in some possible implementations, the edge features of the abnormal light reflection region are analyzed to determine the edge detail prominence coefficient of the abnormal light reflection region, including:

[0016] Edge detection is performed on the abnormal light reflection areas to obtain the edge curves;

[0017] The edge smoothing coefficient of the abnormal light reflection region is determined based on the goodness of fit between the edge curve and its fitted straight line.

[0018] The edge detail prominence coefficient of the light reflection abnormality region is determined based on the average edge length of all edge curves in the region, the number of all edge curves, and the edge smoothing coefficient. The number of curves is positively correlated with the edge detail prominence coefficient, while the average edge length and the edge smoothing coefficient are both negatively correlated with the edge detail prominence coefficient.

[0019] In conjunction with the first aspect above, in some possible implementations, determining the edge detail prominence coefficient of the light reflection anomalous region includes:

[0020] Determine the ratio of the number of all edge curves within the abnormal light reflection region to the average edge length of all edge curves;

[0021] The product of the ratio and the negative correlation mapping result of the edge smoothing coefficient is used as the edge detail prominence coefficient of the light reflection abnormal region.

[0022] In conjunction with the first aspect above, in some possible implementations, determining the shadow index parameters of the abnormal light reflection region includes:

[0023] The pixel correlation coefficient of the light reflection abnormal region is determined based on the gray-level correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image.

[0024] Determine the number of edge pixels within the abnormal light reflection region, and the average number of edge pixels in each of the other adjacent abnormal light reflection regions.

[0025] Based on the difference between the number of edge pixels in the abnormal light reflection area and the average number of edge pixels, and the pixel correlation coefficient of the abnormal light reflection area, the shadow index parameter of the abnormal light reflection area is determined. The difference is negatively correlated with the shadow index parameter, and the pixel correlation coefficient is positively correlated with the shadow index parameter.

[0026] In conjunction with the first aspect above, in some possible implementations, determining the pixel correlation coefficient of the light reflection anomalous region includes:

[0027] Determine the average pixel value of the same region in the grayscale image of the first set number of historical powder-coated surface images of the abnormal light reflection region to obtain the first pixel average value sequence.

[0028] Determine the average pixel value of the same location region in the grayscale image of each adjacent light reflection abnormal region in a first set number of historical powder-coated surface images to obtain a second pixel mean sequence.

[0029] The correlation parameters between the first pixel mean sequence and the second pixel mean sequence are determined, and the pixel correlation coefficient of the light reflection abnormal region is determined based on the overall distribution of the correlation parameters.

[0030] In conjunction with the first aspect above, in some possible implementations, the relevant parameter is the absolute value of the Pearson correlation coefficient between the first pixel mean sequence and the second pixel mean sequence.

[0031] In conjunction with the first aspect above, in some possible implementations, the product value of the edge detail prominence coefficient, pixel correlation coefficient, and shadow index parameter of the light reflection abnormal region is determined, and the product value is used as the edge detail loss coefficient of the light reflection abnormal region.

[0032] In conjunction with the first aspect above, in some possible implementations, edge detection is performed on the abnormal light reflection region based on the edge detail loss coefficient, including:

[0033] The edge detection of the abnormal light reflection region is performed using the Canny algorithm to obtain the edge line of the abnormal light reflection region. During the edge detection process:

[0034] The gradient magnitudes of each edge point and all pixels in the abnormal light reflection region are obtained, where each edge point is a local maximum point obtained by the non-maximum suppression method in the Canny algorithm.

[0035] The number of edge points to be added is determined based on the number of edge points and the edge detail loss coefficient of the abnormal light reflection area;

[0036] Identify candidate pixels (excluding edge points) within all pixels in the abnormal light reflection region. Based on the gradient magnitude of the candidate pixels, select a number of edge points with the largest gradient magnitude from the candidate pixels.

[0037] All the aforementioned edge points, along with the number of edge points to be added, are taken as the final edge points.

[0038] In conjunction with the first aspect mentioned above, some possible implementations include powder spreading quality detection based on edge detection results, including:

[0039] Extract key features of edge lines from edge detection results;

[0040] The extracted key features are compared with the key features of the edge lines extracted from the standard powder-coated surface image, and the powder-coating quality is determined to be abnormal based on the comparison results.

[0041] To address the aforementioned technical problems, in a second aspect, the present invention also provides a device for detecting the quality of powder bed laying in laser selective melting for metal additive manufacturing, the device comprising:

[0042] The local region acquisition module is used to acquire a grayscale image of the powder-coated surface, and to segment the grayscale image to obtain various local regions;

[0043] An abnormal region acquisition module is used to determine the light reflection anomaly coefficient of the local region based on the pixel differences of different local regions, and to filter out the light reflection anomaly regions in each local region based on the light reflection anomaly coefficient.

[0044] The edge detail prominence coefficient acquisition module is used to analyze the edge features in the light reflection abnormal region and determine the edge detail prominence coefficient of the light reflection abnormal region;

[0045] The shadow index parameter acquisition module is used to determine the shadow index parameters of the light reflection abnormal area based on the grayscale correlation between the light reflection abnormal area and other adjacent light reflection abnormal areas in the historical powder-coated surface image, and the difference in the number of edge pixels between the light reflection abnormal area and other adjacent light reflection abnormal areas.

[0046] The edge detail loss coefficient acquisition module is used to fuse the light reflection anomaly coefficient, edge detail prominence coefficient and shadow index parameter of the light reflection anomaly region to determine the edge detail loss coefficient of the light reflection anomaly region.

[0047] The quality inspection module is used to perform edge detection on the abnormal light reflection area based on the edge detail loss coefficient. The larger the edge detail loss coefficient, the greater the amount of edge points retained during the edge detection process. The module also performs powder spreading quality inspection based on the edge detection results.

[0048] To address the aforementioned technical problems, in a third aspect, the present invention also provides a laser selective melting powder bed quality inspection system for metal additive manufacturing, comprising a memory and a processor. The memory stores executable program code, and the processor retrieves and runs the executable program code from the memory, causing the device to perform the methods described in the first aspect or any possible implementation thereof.

[0049] To address the aforementioned technical problems, in a fourth aspect, the present invention also provides a computer program product comprising: computer program code, which, when executed on a computer, causes the computer to perform the method described in the first aspect or any possible implementation thereof.

[0050] To address the aforementioned technical problems, in a fifth aspect, the present invention also provides a computer-readable storage medium storing computer program code that, when executed on a computer, causes the computer to perform the method described in the first aspect or any possible implementation thereof.

[0051] The present invention has the following beneficial effects: The present invention performs region segmentation on the grayscale image of the powder-coated surface image to obtain multiple local regions. Due to powder coating defects or shadow noise, abnormal light reflection will cause changes in the pixel performance of local regions in the powder-coated surface image, resulting in significant pixel differences from normal powder-coated areas. Therefore, based on the pixel differences of different local regions, the light reflection anomaly coefficient of local regions is determined, thereby initially screening out areas with abnormal light reflection. Furthermore, on the one hand, since the edge detail information of the powder spreading defect area is more prominent than that of the normal area, the subsequent edge detection needs to pay special attention to the area with prominent edge detail information. Therefore, the edge features of the light reflection defect area are analyzed to determine the edge detail prominence coefficient of the light reflection defect area. On the other hand, since the shadow noise area is usually caused by the local distribution state of the powder, the pixel changes in this area are highly correlated with the pixel changes of the neighboring light reflection defect area. Moreover, on the powder bed spreading surface, the shadow part usually causes the loss of powder edge detail features to varying degrees. For the light reflection defect area with high correlation of pixel changes, the number of edge pixels in the shadow area is often less than that in the defect area compared to other adjacent light reflection defect areas. Therefore, the shadow index parameters of the light reflection defect area are determined based on the grayscale correlation between the light reflection defect area and other adjacent light reflection defect areas in the historical powder spreading surface image, as well as the difference in the number of edge pixels between the light reflection defect area and other adjacent light reflection defect areas. Finally, by combining the light reflection anomaly coefficient, edge detail prominence coefficient, and shadow index parameters of the light reflection anomaly area, the edge detail loss coefficient of the light reflection anomaly area is determined to reflect the degree of edge detail loss in the light reflection anomaly area. Then, based on the edge detail loss coefficient, edge detection is performed on the light reflection anomaly area. The larger the edge detail loss coefficient, the greater the amount of edge points retained in the edge detection process, thereby reducing the influence of shadow noise on the powder-coated surface, retaining more edge detail information, and ultimately improving the accuracy of powder-coated quality detection. Attached Figure Description

[0052] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 This is a flowchart illustrating the steps of a laser selective melting and forming powder bed quality inspection method for metal additive manufacturing, according to an embodiment of the present invention.

[0054] Figure 2 This is a grayscale image of a powder-coated surface captured by a camera in an embodiment of the present invention.

[0055] Figure 3 This is a schematic diagram of the structure of the laser selective melting and forming powder bed quality detection device for metal additive manufacturing according to an embodiment of the present invention;

[0056] Figure 4 This is a schematic diagram of the structure of a laser selective melting powder bed quality inspection system for metal additive manufacturing, according to an embodiment of the present invention. Detailed Implementation

[0057] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings.

[0058] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the invention.

[0059] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0060] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0061] It should be noted that the concepts of "first" and "second" mentioned in this invention are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0062] Although operations or steps are described in a specific order in the accompanying drawings in the embodiments of the present invention, this should not be construed as requiring these operations or steps to be performed in the specific order or serial order shown, or requiring all of the shown operations or steps to be performed to obtain the desired result. In the embodiments of the present invention, these operations or steps may be performed serially; they may be performed in parallel; or a portion of these operations or steps may be performed.

[0063] Furthermore, it is understood that the data involved in the technical solutions of this invention (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions. Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains, and all parameters or indicators in the formulas involved in this invention are normalized values ​​that have eliminated the influence of dimensions.

[0064] To address the problem of significant errors in powder bed quality inspection results caused by shadow noise in powder bed surface images, this invention provides a method for inspecting the powder bed quality of laser selective melting forming powder beds used in metal additive manufacturing. This method acquires various local regions of a grayscale image of the powder bed surface. Based on the pixel differences in different local regions, it determines the light reflection anomaly coefficient of each local region to filter out areas with abnormal light reflection. It then determines the edge detail prominence coefficient and shadow index parameters of these areas, and, combined with the light reflection anomaly coefficient, determines the edge detail loss coefficient. Based on the edge detail loss coefficient, it performs edge detection on these areas. A higher edge detail loss coefficient results in a greater amount of edge points retained during edge detection, thereby reducing the impact of shadow noise on the powder bed surface and preserving more edge detail information in areas with abnormal light reflection. Finally, it performs quality inspection based on the edge detection results, effectively improving the accuracy of powder bed quality inspection.

[0065] The following will describe in detail, with reference to the accompanying drawings, a method for detecting the quality of laser selective melting powder bed laying for metal additive manufacturing, provided by an embodiment of the present invention.

[0066] Figure 1 This diagram illustrates the basic flow chart of a laser selective melting powder bed quality inspection method for metal additive manufacturing, provided by an embodiment of the present invention. Figure 1 As shown, the method specifically includes the following steps:

[0067] Step S100: Obtain a grayscale image of the powder-coated surface, and perform region segmentation on the grayscale image to obtain various local regions.

[0068] Specifically, after the laser selective melting and forming of the powder bed is completed, a CCD industrial camera is installed directly above the powder bed to acquire real-time images of the powder bed surface. In this embodiment, the camera acquisition frequency is set to acquire an image of the powder bed surface every 2 seconds. Figure 2 The image shown is a grayscale representation of a powder-coated surface captured by a camera. Geometric transformations are performed on the captured powder-coated surface image to correct image distortion caused by camera position and distortion. This preprocessing of the powder-coated surface image ensures the accuracy of subsequent analysis. The powder-coated surface image referred to below is the preprocessed image.

[0069] During the powder bed spreading process, various defects may occur due to factors such as unstable laser energy and residual printing residue. Therefore, defect feature identification of the powder bed spreading surface image can achieve quality inspection of the powder bed spreading. However, in the powder bed spreading surface image, irregular or unevenly distributed powder particles in local areas can cause abnormal light reflection, resulting in shadow noise regions. These shadow noise regions obscure the edge details of the defective areas, thus affecting the subsequent extraction of defect features and ultimately impacting the accuracy of powder bed spreading quality inspection.

[0070] Because of the abnormal light reflection in the local area of ​​the powder-coated surface, the pixel appearance of the local area in the powder-coated surface image will change, and there will be a significant pixel difference from the normal powder-coated area. Therefore, image processing algorithms can be used to obtain all local areas with similar pixel features in the powder-coated surface image.

[0071] Based on the above analysis, the powder-spreading surface image is converted to grayscale to obtain a grayscale image of the powder-spreading surface. Superpixel block segmentation is then performed on the grayscale image to obtain multiple superpixel block regions. Each superpixel block region is treated as a local region, thereby achieving the goal of obtaining all local regions with similar pixel features in the powder-spreading surface image. Since the grayscale conversion process and the superpixel block segmentation process are existing technologies, they will not be elaborated upon here.

[0072] Step S200: Determine the light reflection anomaly coefficient of the local region based on the pixel differences of the different local regions, and filter out the light reflection anomaly regions in each local region based on the light reflection anomaly coefficient.

[0073] Specifically, due to factors such as the shape and distribution of anomalous powder particles, the pixel appearance of a specific area differs from the distribution of normal powder particles in the overall image. Furthermore, the distribution of anomalous powder particles is localized, while normal powder particles are widely distributed. Therefore, if the pixel distribution in a localized area differs significantly from the overall image, it suggests that this area may contain a higher level of pixel anomalies, implying that it might be a region of abnormal light reflection.

[0074] In this embodiment, the above-mentioned steps for determining the light reflection anomaly coefficient of the local region based on the pixel differences of different local regions include:

[0075] Get the average pixel value of each local region, and get the average pixel value of the entire grayscale image;

[0076] The light reflection anomaly coefficient of each local region is determined based on the difference between the average pixel value of each local region and the average pixel value of the entire grayscale image. The larger the difference, the larger the corresponding light reflection anomaly coefficient.

[0077] Regarding the above steps, as an example, for any local region, i.e., the superpixel region, with the first... Taking a local region as an example, obtain the first local region. The average pixel value of a local region, where all subsequent pixel references refer to grayscale values, is also obtained. Simultaneously, the average pixel value of the entire grayscale image is acquired, and this is used to calculate the average pixel value of the first local region. The pixel mean of each local region is compared with the pixel mean of the entire grayscale image to obtain the pixel mean of the first local region. Light reflection anomaly coefficient in a local area:

[0078] ;

[0079] in, Indicates the first The light reflection anomaly coefficient of a local area; Indicates the first The average pixel value of a local region; This represents the average pixel value of the entire grayscale image; This represents the logarithmic function with base 2, used to prevent values ​​from becoming too large.

[0080] In the above calculation formula, when the first... The greater the difference between the average pixel value of a local region and the average pixel value of the entire grayscale image, the more it indicates that the... The more likely a local area is to have an abnormal light reflection, the larger the value of the light reflection anomaly coefficient.

[0081] Following the above method, the light reflection anomaly coefficients of all local areas on the powder-coated surface image can be determined. An anomaly coefficient threshold is preset; the specific value of this threshold can be reasonably set according to actual conditions. In this embodiment, the threshold value is set to 0.3. The light reflection anomaly coefficient of each local area is compared with this threshold. If the light reflection anomaly coefficient is greater than the threshold, the corresponding local area is marked as a light reflection anomaly area. In this way, all light reflection anomaly areas on the powder-coated surface image can be filtered out.

[0082] Step S300: Analyze the edge features in the abnormal light reflection region to determine the edge detail prominence coefficient of the abnormal light reflection region.

[0083] Specifically, in powder-coated surface images, both actual powder-coated defects and shadow noise can alter the pixel representation in local areas. Shadow noise causes varying degrees of edge information loss, and when shadow areas cover powder defect areas, the edge information of the defect areas is more severely affected, leading to errors in subsequent judgments. Therefore, after acquiring all areas with abnormal light reflection in the powder-coated surface image, further analysis of the edge information loss in these areas is necessary to accurately identify the defect areas.

[0084] Analysis revealed that the edge structures of local areas in powder-coated surface images exhibited different characteristics due to variations in powder distribution. For instance, defective areas, caused by powder accumulation or voids, displayed irregular and rough edge curves; while normal areas showed uniform powder distribution, more regular edge contours, and good overall connectivity. Therefore, in subsequent edge extraction, special attention should be paid to areas with prominent edge details.

[0085] In this embodiment, the steps of analyzing the edge features in the abnormal light reflection region and determining the edge detail prominence coefficient of the abnormal light reflection region include:

[0086] Edge detection is performed on the abnormal light reflection areas to obtain the edge curves;

[0087] The edge smoothing coefficient of the abnormal light reflection region is determined based on the goodness of fit between the edge curve and its fitted straight line.

[0088] The edge detail prominence coefficient of the light reflection abnormality region is determined based on the average edge length of all edge curves in the region, the number of all edge curves, and the edge smoothing coefficient. The number of curves is positively correlated with the edge detail prominence coefficient, while the average edge length and the edge smoothing coefficient are both negatively correlated with the edge detail prominence coefficient.

[0089] Regarding the above steps, as an example, for any region with abnormal light reflection, let the first... Taking the first abnormal light reflection region as an example, the Canny edge detection algorithm is used to analyze this region. Edge detection is performed on the first region of abnormal light reflection to obtain the edge of the first region. All edge curves in the region of abnormal light reflection are obtained. The least squares method and linear fitting are used to obtain the edge curves of this region. The fitted straight line for all edge curves in the region of abnormal light reflection is then calculated. The goodness of fit of all edge curves in an area of ​​abnormal light reflection to their corresponding fitted straight lines is used as the smoothing coefficient of the edge curves. The larger the value of the smoothing coefficient, the closer the edge curve is to its fitted straight line, i.e., the higher the smoothness of the curve. Since the specific process of obtaining the goodness of fit is existing technology, it will not be described in detail here.

[0090] Furthermore, determine the first The mean of the smoothing coefficients of all edge curves in the nth region of abnormal light reflection is denoted as the nth mean. The edge smoothing coefficient of the first light reflection abnormality region. Simultaneously, obtain the first... The number of all edge curves in a region of abnormal light reflection and the average length of all edge curves are calculated, and this average length is denoted as the nth... The average edge length of each region with abnormal light reflection.

[0091] Furthermore, based on this first The edge smoothing coefficient and mean edge length of the first light reflection anomalous region, and the first... The number of curves in all edge curves within a region of abnormal light reflection is used to determine the number of curves in that region. Edge detail enhancement factor for areas with abnormal light reflection:

[0092] ;

[0093] in, Indicates the first The edge detail enhancement factor of an area with abnormal light reflection; Indicates the first The number of curves in all edge curves within a region of abnormal light reflection; Indicates the first The average edge length of the nth region with abnormal light reflection, which is the nth The average length of all edge curves in a region of abnormal light reflection; Indicates the first The edge smoothing coefficient of an area with abnormal light reflection.

[0094] In the above calculation formula, if there are a large number of short edge curves inside the abnormal light reflection area, and the difference between these edge curves and their fitted straight line is large, it indicates that these edge curves may contain more detailed information compared to other more widely distributed edge curves in the powder surface image. The higher the prominence of the edge details, the more prominent the details of the abnormal light reflection area.

[0095] Step S400: Based on the grayscale correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image, and the difference in the number of edge pixels between the light reflection abnormal region and other adjacent light reflection abnormal regions, determine the shadow index parameters of the light reflection abnormal region.

[0096] Specifically, since shadow noise regions are typically caused by localized powder distribution, pixel changes in these regions are highly correlated with pixel changes in adjacent areas of abnormal light reflection, indicating that shadow noise is closely related to lighting conditions and powder distribution. Therefore, when analyzing powder spreading quality, the influence of light reflection needs to be comprehensively considered to improve the accuracy of defect detection.

[0097] Meanwhile, considering that under normal circumstances, shadowed areas on the powder bed surface typically lead to varying degrees of loss of powder edge detail features, for light reflection anomaly areas with high pixel correlation coefficients, the number of edge pixels in the shadowed areas is often less than that in the defective areas compared to other adjacent light reflection anomaly areas. By comparing the number of edge pixels in a light reflection anomaly area with that in other adjacent light reflection anomaly areas, the specific affected areas and the relationship between them can be effectively identified.

[0098] In this embodiment, the above-mentioned method of determining the shadow index parameters of the light reflection abnormal region based on the grayscale correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image, and the difference in the number of edge pixels between the light reflection abnormal region and other adjacent light reflection abnormal regions, includes the following steps:

[0099] The pixel correlation coefficient of the light reflection abnormal region is determined based on the gray-level correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image.

[0100] Determine the number of edge pixels within the abnormal light reflection region, and the average number of edge pixels in each of the other adjacent abnormal light reflection regions.

[0101] Based on the difference between the number of edge pixels in the abnormal light reflection area and the average number of edge pixels, and the pixel correlation coefficient of the abnormal light reflection area, the shadow index parameter of the abnormal light reflection area is determined. The difference is negatively correlated with the shadow index parameter, and the pixel correlation coefficient is positively correlated with the shadow index parameter.

[0102] Regarding the above steps, as an example, for any region with abnormal light reflection, the same applies to the first... Taking the first area of ​​abnormal light reflection as an example, obtain the first... The average pixel value of the same location in the grayscale image of a first predetermined number of historical powder-coated surface images for each region with abnormal light reflection is used to determine the first pixel value sequence. In this embodiment, the first predetermined number is set to 50. These average pixel values ​​are then sorted in chronological order to obtain the first pixel value sequence. Determine the first The Euclidean distance between the centroid coordinates of the first light reflection anomaly region and any other light reflection anomaly region is obtained. Based on this Euclidean distance, the distance between the first and second light reflection anomaly regions is calculated. A light reflection abnormality region is located closest to a second predetermined number of other light reflection abnormality regions. These other light reflection abnormality regions are denoted as the nth region. Each region with abnormal light reflection is adjacent to other regions with abnormal light reflection, forming a set of adjacent abnormal regions. In this embodiment, the second preset number is set to 8. For adjacent abnormal region sets... Any adjacent area of ​​abnormal light reflection within the set any of the innermost numbers Taking the first adjacent area of ​​abnormal light reflection as an example, obtain the first... The average pixel values ​​of adjacent areas with abnormal light reflection in the grayscale images of a first predetermined number of historical powder-coated surface images are taken from the same location. These average pixel values ​​are then sorted in chronological order to obtain a second pixel value sequence. Determine the mean sequence of the first pixel. With the second pixel mean sequence The absolute value of the Pearson correlation coefficient between them The absolute value of the Pearson correlation coefficient As the first pixel mean sequence With the second pixel mean sequence The larger the value of the relevant parameter, the stronger the... The area with abnormal light reflection and the first The more likely there is a correlation between pixel changes between adjacent areas of abnormal light reflection.

[0103] Based on this, obtain the first A set of light reflection anomalous regions and their adjacent anomalous regions. The mean of the relevant parameters among all adjacent areas of abnormal light reflection is used to determine the mean of the relevant parameters as the first... Pixel correlation coefficient of a region with abnormal light reflection .

[0104] The above method can be used to determine the pixel correlation coefficient of any region with abnormal light reflection. The higher the value of the pixel correlation coefficient, the more likely the pixel performance of the corresponding region with abnormal light reflection may be affected by pixels in other regions with abnormal light reflection. However, the specific areas of this influence and the affected areas are not yet clear, so further differentiation and screening are required.

[0105] Furthermore, obtain the first The number of edge pixels in the abnormal light reflection area is obtained, and the number of edge pixels in the abnormal light reflection area is obtained. A set of adjacent abnormal regions of a region with abnormal light reflection. The average number of edge pixels in all adjacent areas of abnormal light reflection is denoted as the average number of edge pixels. This is achieved by comparing the values ​​of these edge pixels with the average number of edge pixels in the first region. The magnitude of the number of edge pixels in each abnormal light reflection region and the average number of edge pixels, combined with the first... The pixel correlation coefficient of the first light reflection abnormality region can be used to determine the first... Shadow index parameters for areas with abnormal light reflection:

[0106] ;

[0107] in, Indicates the first Shadow index parameters for areas with abnormal light reflection; Indicates the first The number of edge pixels in an area of ​​abnormal light reflection. Indicates the first The average number of edge pixels in the nth region of abnormal light reflection, i.e., the nth A set of adjacent abnormal regions of a region with abnormal light reflection. The average number of edge pixels in all other areas of abnormal light reflection; Indicates the first The pixel correlation coefficient of an area with abnormal light reflection.

[0108] In the above calculation formula, the first... Number of edge pixels in areas of abnormal light reflection With the A set of adjacent abnormal regions of a region with abnormal light reflection. The average number of edge pixels in all other areas of abnormal light reflection The smaller the difference, the better. Pixel correlation coefficient of a region with abnormal light reflection The larger the value, the more it indicates the number of... The higher the likelihood of shadow noise affecting an area with abnormal light reflection, the larger the value of the corresponding shadow index parameter.

[0109] Step S500: Combine the light reflection anomaly coefficient, edge detail prominence coefficient, and shadow index parameters of the light reflection anomaly region to determine the edge detail loss coefficient of the light reflection anomaly region.

[0110] Specifically, the degree of edge detail loss in each abnormal light reflection region is evaluated based on the edge detail prominence coefficient, pixel correlation coefficient, and shadow index parameters, thus obtaining the edge detail loss coefficient for each abnormal light reflection region.

[0111] Similarly, with the first Taking an abnormal light reflection area as an example, the first one is fused. The light reflection anomaly coefficient, edge detail prominence coefficient, and shadow index parameters of each light reflection anomaly region are used to determine the first region. Edge detail loss coefficient of an area with abnormal light reflection:

[0112] ;

[0113] in, Indicates the first The coefficient of edge detail loss in areas of abnormal light reflection; Indicates the first The light reflection anomaly coefficient of a local area; Indicates the first The edge detail enhancement factor of an area with abnormal light reflection; Indicates the first The shadow index parameter of an area with abnormal light reflection.

[0114] In the above calculation formula, the first... When the light reflection anomaly coefficient of an area with abnormal light reflection is high, it indicates that there may be powdering defects or shadows in the area. Furthermore, if the area has obvious edge details and a high shadow index, it indicates that there is significant loss of edge details in the area, and the corresponding edge detail loss coefficient is large. Therefore, in the subsequent edge detection process, more edge details need to be preserved in order to effectively eliminate the impact of shadow occlusion.

[0115] The above method can be used to determine the edge detail loss coefficient of all areas with abnormal light reflection.

[0116] Step S600: Based on the edge detail loss coefficient, perform edge detection on the abnormal light reflection area. The larger the edge detail loss coefficient, the greater the amount of edge points retained during the edge detection process. And perform powder spreading quality detection based on the edge detection results.

[0117] Specifically, the Canny algorithm is used to perform edge detection on areas with abnormal light reflection. The edge lines in the edge detection results can be compared with the edge lines extracted from the standard powder-coated surface image to determine whether there are defects in the areas with abnormal light reflection, thereby achieving powder-coated quality inspection.

[0118] In the existing Canny algorithm for edge detection, the image is first smoothed using a Gaussian filter to reduce noise. Next, the Sobel operator is used to calculate the gradient magnitude and direction of each pixel. Then, based on the gradient magnitude and direction of each pixel, a non-maximum suppression method is applied to search for local maxima and suppress non-maximums, initially identifying edge points. Finally, double thresholding is used to mark edge points as strong or weak edges, thus ultimately determining the edge pixels and obtaining the edge lines in the image. However, due to the presence of significant shadow noise on the powdered surface, the non-maximum suppression method in the existing Canny algorithm may overlook edge features of minor defects, leading to poor reliability of the final defect extraction results. Therefore, the existing Canny algorithm can be improved by utilizing the edge detail loss coefficient of all abnormal light reflection regions to effectively retain more detail information and reduce the loss of edge detail information caused by shadow noise.

[0119] In this embodiment, the above-mentioned edge detection of the abnormal light reflection region based on the edge detail loss coefficient includes the following steps:

[0120] The edge detection of the abnormal light reflection region is performed using the Canny algorithm to obtain the edge line of the abnormal light reflection region. During the edge detection process:

[0121] The gradient magnitudes of each edge point and all pixels in the abnormal light reflection region are obtained, where each edge point is a local maximum point obtained by the non-maximum suppression method in the Canny algorithm.

[0122] The number of edge points to be added is determined based on the number of edge points and the edge detail loss coefficient of the abnormal light reflection area;

[0123] Identify candidate pixels (excluding edge points) within all pixels in the abnormal light reflection region. Based on the gradient magnitude of the candidate pixels, select a number of edge points with the largest gradient magnitude from the candidate pixels.

[0124] All the aforementioned edge points, along with the number of edge points to be added, are taken as the final edge points.

[0125] Regarding the above steps, as an example, for any region with abnormal light reflection, the same applies to the first... Taking the first abnormal light reflection region as an example, the Canny algorithm is used to analyze this region. Edge detection is performed on the first region of abnormal light reflection to obtain the edge of the first region. The edge line of the region with abnormal light reflection. In the existing Canny algorithm, this is based on the edge line of the region with abnormal light reflection. The gradient magnitude and direction of all pixels in a region of abnormal light reflection are analyzed. A non-maximum suppression method is applied to search for local maxima, suppressing non-maximums and initially determining each edge point. In this embodiment, to effectively retain more detailed information, thereby reducing the loss of edge detail information due to shadow noise and ultimately improving the accuracy of powder spreading quality detection, according to the first... The magnitude of the edge detail loss coefficient in an area of ​​abnormal light reflection is used to further select pixels with larger gradient magnitudes from the pixels outside the initially determined edge points, and these pixels are also selected as edge points. The larger the value of the edge detail loss coefficient, the larger the number of pixels with larger gradient magnitudes selected, thus preserving the edge features of small defects and improving the reliability of the final defect extraction result.

[0126] Based on the above analysis, the first Within each region of abnormal light reflection, all pixels except for edge points (i.e., local maxima obtained through non-maximum suppression) are considered as candidate pixels. These candidate pixels are arranged in descending order of edge magnitude to obtain a candidate pixel sequence. Simultaneously, the ... The rounded product of the edge detail loss coefficient of the light reflection anomalous region and the total number of local maxima obtained by the non-maximum suppression method is used as the number of edge points to be added. The first n candidate pixels to be added are selected from the candidate pixel sequence. All edge points and the n n edge points to be added are then used as the final edge points. Based on these final edge points, the double thresholding process in the Canny algorithm is used to mark the edge points as strong or weak edges, thus finally determining the nth edge point. All edge lines of the area with abnormal light reflection.

[0127] Following the above method, all edge lines of each area with abnormal light reflection can be obtained, and the powder spreading quality can be detected based on these edge lines. The specific steps for powder spreading quality detection based on edge lines can be reasonably selected as needed, and are not specifically limited here. In this embodiment, for the edge lines of each area with abnormal light reflection, firstly, filtering techniques such as Gaussian filtering or Savitzky-Golay filtering are used to remove noise to obtain smooth edge lines; then, key features such as curvature, edge length, and angle are extracted from the smoothed edge lines to analyze the shape and characteristics of the powder bed; then, the key features of the extracted edge lines are compared with the key features of the edge lines extracted from the standard powder spreading surface image, and the presence of abnormalities in powder spreading quality is determined based on the comparison results.

[0128] Specifically, when the difference in key features exceeds a pre-set threshold, it indicates a powder-laying defect in the corresponding area with abnormal light reflection; otherwise, it indicates no powder-laying defect in the corresponding area. The final powder-laying quality test result is determined based on the presence or absence of powder-laying defects in each area with abnormal light reflection. If any area with abnormal light reflection has a powder-laying defect, the powder-laying quality is deemed abnormal; if none of the areas with abnormal light reflection have powder-laying defects, the powder-laying quality is deemed normal. Finally, the test results are fed back to technical personnel, who analyze the causes of the powder-laying defects and adjust the powder-laying process parameters to optimize the powder bed quality.

[0129] Based on the same inventive concept, embodiments of the present invention also provide a device for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing, such as... Figure 3 As shown, the device includes:

[0130] The local region acquisition module is used to acquire a grayscale image of the powder-coated surface, and to segment the grayscale image to obtain various local regions;

[0131] An abnormal region acquisition module is used to determine the light reflection anomaly coefficient of the local region based on the pixel differences of different local regions, and to filter out the light reflection anomaly regions in each local region based on the light reflection anomaly coefficient.

[0132] The edge detail prominence coefficient acquisition module is used to analyze the edge features in the light reflection abnormal region and determine the edge detail prominence coefficient of the light reflection abnormal region;

[0133] The shadow index parameter acquisition module is used to determine the shadow index parameters of the light reflection abnormal area based on the grayscale correlation between the light reflection abnormal area and other adjacent light reflection abnormal areas in the historical powder-coated surface image, and the difference in the number of edge pixels between the light reflection abnormal area and other adjacent light reflection abnormal areas.

[0134] The edge detail loss coefficient acquisition module is used to fuse the light reflection anomaly coefficient, edge detail prominence coefficient and shadow index parameter of the light reflection anomaly region to determine the edge detail loss coefficient of the light reflection anomaly region.

[0135] The quality inspection module is used to perform edge detection on the abnormal light reflection area based on the edge detail loss coefficient. The larger the edge detail loss coefficient, the greater the amount of edge points retained during the edge detection process. The module also performs powder spreading quality inspection based on the edge detection results.

[0136] It should be noted that the device provided in the above embodiments is only an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above.

[0137] Based on the same inventive concept, embodiments of the present invention also provide a laser selective melting powder bed quality inspection system for metal additive manufacturing, such as... Figure 4 As shown, the control system includes: a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the system can perform any of the aforementioned laser selective melting powder bed quality detection methods for metal additive manufacturing.

[0138] In this embodiment of the invention, the system can be divided into functional modules according to the above method example. For example, each module can correspond to a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0139] Based on the same inventive concept, embodiments of the present invention also provide a computer program product, which includes: computer program code, which, when run on a computer, causes the computer to execute any of the aforementioned laser selective melting molding powder bed quality detection methods for metal additive manufacturing.

[0140] Based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium storing computer program code, which, when executed on a computer, causes the computer to perform any of the aforementioned laser selective melting molding powder bed quality detection methods for metal additive manufacturing.

[0141] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for detecting the quality of powder bed laying in laser selective melting for metal additive manufacturing, characterized in that, Includes the following steps: A grayscale image of the powder-coated surface is obtained, and the grayscale image is segmented to obtain various local regions; Based on the pixel differences in different local regions, the light reflection anomaly coefficient of the local region is determined, and based on the light reflection anomaly coefficient, the light reflection anomaly regions in each local region are filtered out. The edge features in the abnormal light reflection region are analyzed to determine the edge detail prominence coefficient of the abnormal light reflection region; Based on the grayscale correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image, and the difference in the number of edge pixels between the light reflection abnormal region and other adjacent light reflection abnormal regions, the shadow index parameters of the light reflection abnormal region are determined. By integrating the light reflection anomaly coefficient, edge detail prominence coefficient, and shadow index parameters of the light reflection anomaly region, the edge detail loss coefficient of the light reflection anomaly region is determined. Based on the edge detail loss coefficient, edge detection is performed on the light reflection abnormal area. The larger the edge detail loss coefficient, the greater the amount of edge points retained during the edge detection process. The powder spreading quality is then detected based on the edge detection results. Determining the shadow index parameters of the area with abnormal light reflection includes: The pixel correlation coefficient of the light reflection abnormal region is determined based on the gray-level correlation between the light reflection abnormal region and other adjacent light reflection abnormal regions in the historical powder-coated surface image. Determine the number of edge pixels within the abnormal light reflection region, and the average number of edge pixels in each of the other adjacent abnormal light reflection regions. Based on the difference between the number of edge pixels in the abnormal light reflection area and the average number of edge pixels, and the pixel correlation coefficient of the abnormal light reflection area, the shadow index parameter of the abnormal light reflection area is determined. The difference is negatively correlated with the shadow index parameter, and the pixel correlation coefficient is positively correlated with the shadow index parameter. The product of the edge detail prominence coefficient, pixel correlation coefficient, and shadow index parameter of the light reflection abnormal region is determined, and the product value is used as the edge detail loss coefficient of the light reflection abnormal region.

2. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 1, characterized in that, Determining the light reflection anomaly coefficient of the local region includes: Get the average pixel value of each local region, and get the average pixel value of the entire grayscale image; The light reflection anomaly coefficient of each local region is determined based on the difference between the average pixel value of each local region and the average pixel value of the entire grayscale image. The larger the difference, the larger the corresponding light reflection anomaly coefficient.

3. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 1, characterized in that, Analyzing the edge features in the anomalous light reflection region to determine the edge detail prominence coefficient of the anomalous light reflection region includes: Edge detection is performed on the abnormal light reflection areas to obtain the edge curves; The edge smoothing coefficient of the abnormal light reflection region is determined based on the goodness of fit between the edge curve and its fitted straight line. The edge detail prominence coefficient of the light reflection abnormality region is determined based on the average edge length of all edge curves in the region, the number of all edge curves, and the edge smoothing coefficient. The number of curves is positively correlated with the edge detail prominence coefficient, while the average edge length and the edge smoothing coefficient are both negatively correlated with the edge detail prominence coefficient.

4. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 3, characterized in that, Determining the edge detail prominence coefficient of the region with abnormal light reflection includes: Determine the ratio of the number of all edge curves within the abnormal light reflection region to the average edge length of all edge curves; The product of the ratio and the negative correlation mapping result of the edge smoothing coefficient is used as the edge detail prominence coefficient of the light reflection abnormal region.

5. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 1, characterized in that, Determining the pixel correlation coefficient of the region with abnormal light reflection includes: Determine the average pixel value of the same region in the grayscale image of the first set number of historical powder-coated surface images of the abnormal light reflection region to obtain the first pixel average value sequence. Determine the average pixel value of the same location region in the grayscale image of each adjacent light reflection abnormal region in a first set number of historical powder-coated surface images to obtain a second pixel mean sequence. The correlation parameters between the first pixel mean sequence and the second pixel mean sequence are determined, and the pixel correlation coefficient of the light reflection abnormal region is determined based on the overall distribution of the correlation parameters.

6. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 5, characterized in that, The relevant parameter is the absolute value of the Pearson correlation coefficient between the first pixel mean sequence and the second pixel mean sequence.

7. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 1, characterized in that, Based on the edge detail loss coefficient, edge detection is performed on the abnormal light reflection region, including: The edge detection of the abnormal light reflection region is performed using the Canny algorithm to obtain the edge line of the abnormal light reflection region. During the edge detection process: The gradient magnitudes of each edge point and all pixels in the abnormal light reflection region are obtained, where each edge point is a local maximum point obtained by the non-maximum suppression method in the Canny algorithm. The number of edge points to be added is determined based on the number of edge points and the edge detail loss coefficient of the abnormal light reflection area; Identify candidate pixels (excluding edge points) within all pixels in the abnormal light reflection region. Based on the gradient magnitude of the candidate pixels, select a number of edge points with the largest gradient magnitude from the candidate pixels. All the aforementioned edge points, along with the number of edge points to be added, are taken as the final edge points.

8. The method for detecting the powder bed quality of laser selective melting forming powder for metal additive manufacturing according to claim 1, characterized in that, The powder spreading quality is inspected based on the edge detection results, including: Extract key features of edge lines from edge detection results; The extracted key features are compared with the key features of the edge lines extracted from the standard powder-coated surface image, and the powder-coating quality is determined to be abnormal based on the comparison results.

Images