Method and system for identifying concave marks on forged workpieces, storage medium, data processing terminal

Abstract

The application discloses a kind of concave mark methods and systems on forging workpiece, storage medium, data processing terminal.Recognition, 3D laser sensor is used to carry out full coverage scanning to the surface of forging workpiece and obtains original point cloud data, after the original point cloud data obtained is preprocessed after denoising, carry out layered processing, obtain multiple data layers;For each data layer, the likelihood of the neighbor point of the identification point in the coordinate range is obtained using local fitting projection projection method;The relative position and gradient of the centroid of each identification point and the likelihood of the neighbor point are compared, the degree of concave is calculated, and the initial candidate points are obtained from the identification points according to the degree of concave;The initial candidate points obtained by analyzing multiple data layers are analyzed, and the invalid candidate points in the initial candidate points are filtered out to obtain the maximum probability of the effective candidate points as the identified mark effective points;Based on the mark point, the concave mark on the workpiece is identified.The application can realize the fast and effective identification of the concave mark on the surface of forging workpiece.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of forging workpiece identification, in particular to a forging workpiece recessed mark method and system, a storage medium and a data processing terminal. BACKGROUND

[0002] In the whole process of processing key metal components, the data cannot be fused with single-piece granularity, the data utilization value is low, and accurate on-site process data cannot be provided for process optimization. Therefore, it is of great significance to carry out single-piece tracking technology research and form a single-piece forging process database. The surface marking of the workpiece is crucial to the quality tracking system of the workpiece. The workpiece needs to go through multiple cold working processes after being discharged from the forging area. Therefore, workpiece surface marking is an important link to realize the whole process quality tracking of metal components. At present, the surface marking of metal workpieces adopts a pneumatic dot matrix marking method, and a DataMatrix two-dimensional code is marked on the workpiece surface in the form of a dot matrix. The two-dimensional code marking on the workpiece surface is easily affected by the environment, such as mold release agent dust, surface discoloration after the workpiece is modulated, cutting fluid, oil stains, etc. During the transfer process on the logistics roller, it is easy to be affected by wear. The single-piece tracking function of metal components realized through the workpiece surface marking is highly dependent on the stable reading rate, accuracy and error correction ability of the two-dimensional code. Therefore, the adaptive recognition technology of the workpiece surface marking is very important.

[0003] In the production process of forged workpieces, a camera or a video camera is usually used to take pictures of the two-dimensional code mark or the bar code mark formed by printing on the forged workpiece, obtain a two-dimensional code or bar code information picture, and then use a picture recognition software containing a recognition model to recognize the two-dimensional code or bar code, and obtain the workpiece information contained in the two-dimensional code or bar code.

[0004] Because the recognition area of the forged workpiece is heated or worn during the processing of the forged workpiece, the surface of the workpiece will change complexly. Therefore, the two-dimensional code or bar code mark picture obtained by the camera or video camera will have a large amount of interference texture due to the above changes, which will cause the picture recognition software or recognition program to fail to correctly distinguish the two-dimensional code or bar code mark and the corresponding interference. The recognition of the recessed structure mark usually fails. SUMMARY

[0005] The purpose of the present application is to solve the problems in the prior art, and to provide a forged workpiece recessed mark method and system, a storage medium and a data processing terminal, which are used to solve the recognition problem of the surface mark of the metal forging after processing and cooling.

[0006] The first aspect of the application provides a method for identifying concave marks on a forged workpiece, which comprises the following steps: a 3D laser sensor is used to perform full-coverage scanning on the surface of the forged workpiece to obtain original point cloud data, and then the following data processing method is used to identify the concave marks on the forged workpiece:

[0007] The obtained original point cloud data is subjected to denoising preprocessing;

[0008] The point cloud data obtained after denoising preprocessing is subjected to layered processing to obtain multiple data layers;

[0009] For each data layer, a local fitting projection method is used to obtain the likely nearest neighbor points of the identification points within the coordinate range;

[0010] The relative positions and gradients of the centroids of each identification point and the likely nearest neighbor points are compared, the concave degree is calculated, and the initial candidate points are obtained from the identification points according to the concave degree;

[0011] The initial candidate points obtained by analyzing the multiple data layers are analyzed to filter out invalid candidate points, and the most probable valid candidate points are obtained as the identified mark effective points;

[0012] The concave marks on the workpiece are identified based on the mark points.

[0013] In the step of layered processing of the point cloud data obtained after denoising preprocessing to obtain multiple data layers, the number, direction and thickness of the layered processing of the point cloud data are determined according to the normal vector and data density distribution characteristics of the original point cloud data, and then the point cloud data is subjected to layered processing according to the determined number, direction and thickness of the layered processing of the point cloud data.

[0014] The invalid candidate points include skinning points.

[0015] The identification of the valid points and the skinning points and the screening thereof are realized based on the following distinguishing strategies:

[0016] The valid points are concentrated and form a continuous curved surface, and the skinning points are dispersed and discontinuous;

[0017] The total number of the valid points is less than that of the skinning points,

[0018] The randomness of the concave degree of the skinning points is higher than that of the valid points.

[0019] The 3D laser sensor is driven to work by a linear module through an encoder synchronization mode, and the scanning lens of the 3D laser sensor faces the scanning area of the forged workpiece.

[0020] The wavelength of the 3D laser sensor is selected to be blue light.

[0021] The concave mark on the forged workpiece includes a two-dimensional code and a bar code.

[0022] In a second aspect of the present application, a system for identifying a concave mark on a forged workpiece is provided, which includes a 3D laser sensor and a data processing device. The 3D laser sensor is driven by a linear module. After the linear module synchronously drives the 3D laser sensor to perform full coverage scanning on the surface of the forged workpiece to obtain original point cloud data, the data processing device identifies the concave mark on the forged workpiece by using the data processing method described in the first aspect of the present application.

[0023] In a third aspect of the present application, a storage medium is provided, which stores at least one instruction, at least one program, a code set or an instruction set. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to realize the data processing method described in the first aspect of the present application to identify the concave mark on the forged workpiece.

[0024] In a fourth aspect of the present application, a data processing terminal is provided, which includes a processor and a memory. The memory stores at least one instruction, at least one program, a code set or an instruction set. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to realize the data processing method described in the first aspect of the present application to identify the concave mark on the forged workpiece.

[0025] The method and system for identifying a concave mark on a forged workpiece can quickly and effectively identify the concave mark on the surface of the forged workpiece by using the 3D laser sensor to perform full coverage scanning on the surface of the forged workpiece to obtain original point cloud data and by using the data processing method to identify the concave mark on the forged workpiece. Thus, the information of the forged workpiece in the production process can be effectively identified, and the problem of identifying the surface mark caused by the change of the surface image after the forging is cooled is solved. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 FIG. 1 is a flowchart of the method for identifying a concave mark on a forged workpiece according to an embodiment of the present application.

[0027] Figure 2 FIG. 4 is a schematic diagram of layering point cloud data according to an embodiment of the present application.

[0028] Figure 3 FIG. 6 is a schematic diagram of layering point cloud data in a predetermined direction according to an embodiment of the present application.

[0029] Figure 4 FIG. 8 is a schematic diagram of selecting a point and surrounding points in point cloud data and projecting the point and the surrounding points to determine a likely nearest neighbor point according to an embodiment of the present application.

[0030] Figure 5 is a schematic diagram of identifying the effective mark point from the skinning point.

[0031] Figure 6 is a structural principle schematic diagram of the identification system for the concave mark of the forged workpiece. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be described clearly and completely below in combination with the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments of the present application.

[0033] The present application is based on the relatively stable attribute of the depth information of the mark area of the forged workpiece before and after hot working, proposes to use a laser sensor to perform three-dimensional scanning, use three-dimensional information to identify mark point cloud data, extract the characteristics of the identified mark, and increase the spatial depth information during identification, thereby improving the identification degree, so as to realize the identification of the concave mark on the forged workpiece, and solve the problem that the concave mark on the forged workpiece cannot be effectively identified by using a camera to freely shoot pictures in the prior art.

[0034] In a first aspect of the embodiments of the present application, a method for identifying a concave mark on a forged workpiece is provided, which uses a 3D laser sensor to perform full-coverage scanning on the surface of the forged workpiece to obtain original point cloud data, and then identifies the concave mark on the forged workpiece by the following data processing method:

[0035] The obtained original point cloud data is preprocessed by denoising;

[0036] The point cloud data obtained after the denoising preprocessing is processed in layers to obtain multiple data layers;

[0037] For each data layer, a local fitting projection projection method is used to obtain the likely nearest neighbor points of the identification points within the coordinate range;

[0038] The relative positions and gradients of the centroids of each identification point and the likely nearest neighbor points are compared, the concave degree is calculated, and the initial candidate points are obtained from the identification points according to the concave degree;

[0039] The initial candidate points obtained by analyzing the multiple data layers are analyzed, the invalid candidate points in the initial candidate points are filtered out, the most probable effective candidate points are obtained as the identified mark effective points;

[0040] The concave mark on the workpiece is identified based on the mark point.

[0041] In the point cloud data denoising, a statistical denoising method can be used to denoise the point cloud data, avoiding affecting the signal point features.

[0042] The 3D laser sensor is driven to work by a linear module through an encoder synchronization mode, the scanning lens of the 3D laser sensor faces the scanning area of the forged workpiece, and the motor of the 3D laser sensor and the linear module works through the encoder synchronization mode, so that the scanning precision of the point data of the workpiece can be provided, higher spatial precision data can be obtained, and the foundation for subsequent point cloud data processing is laid.

[0043] In the embodiment of the application, the forged workpiece is produced in a high-temperature environment during the production and processing process, therefore, the wavelength of the 3D laser sensor is selected as blue light to ensure the anti-interference of signal acquisition in a high-temperature scene.

[0044] In some embodiments, in the step of obtaining a plurality of data layers by layering the point cloud data obtained after the denoising and prediction processing, a layering model or a tomography model established based on an empirical rule or empirical data is used to determine the number, direction and thickness of layering the point cloud data according to the normal vector and data density distribution characteristics of the original point cloud data, and then the point cloud data is layering processed according to the determined number, direction and thickness of layering the point cloud data. Through layering of the point cloud data and separate processing of each data layer after layering, the data dimension can be reduced, the data processing speed can be improved, and at the same time, to improve the processing speed of each data layer, a local fitting projection lamp method is used to simplify the data complexity. Through layering analysis of the data, a "thin layer" data can be obtained, the amount of data processed each time can be reduced, and through combination analysis of the multi-layer data, the feature loss caused by over-thinning layering can be made up, the feature weakening caused by label filling can be adapted, the data loss can be avoided, and the accuracy or accuracy of identification can be ensured.

[0045] Specific layering process, please refer to Figure 2 、 3 , Figure 2 shows the layering processing of the point cloud data, please refer to Figure 2 , the point cloud data 2 is layered by a plurality of layering lines 1, and the three layering lines shown in the figure can divide the point cloud data into a plurality of layers with the same or different thicknesses according to different layering directions, Figure 3The processing of the hierarchical direction of the point cloud data is shown, and a hierarchical line 1 is inclined at a certain angle with the horizontal plane to layer the point cloud data at a preset angle or direction.

[0046] In the prior art, the identification of the identified point cloud data is mainly to identify the identification point relative to the "adjacent point" with the spatial "concave" feature. The adjacent point refers to calculating the distance between a point and all other points in three-dimensional space, and taking the K nearest points as the adjacent points. The identification of the concave feature is achieved by using the following two evaluation methods: 1. Calculate the geometric feature expectation value of the adjacent points, such as the centroid or the fitting plane, and compare the relationship between the identified point and the expectation value, such as the distance and the direction, to determine the "concave" degree feature, 2. Calculate the gradient value of each point and the surrounding points. The traditional "concave" feature identification algorithm needs to establish a "neighbor" relationship tree for each three-dimensional point, and the algorithm processing is complex, requiring high computing power. When scanning point cloud data containing more than 100+ million data points, the algorithm processing data volume is large, the traditional algorithm is very slow when using CPU processing, and using GPU increases the cost, has poor universality, and the identified concave feature is weakened due to different degrees of filling phenomenon. In addition, due to the easy peeling of the workpiece in the thermal additive process, similar concave features similar to the identification may be caused, which may lead to misjudgment. Therefore, in the present application, in order to quickly and effectively identify the identification point and reduce the pressure on the processor, an accurate "neighbor" structure is not established, and a "likely" neighbor is quickly obtained. The efficiency is improved by using layer analysis and surface parameterization method to reduce the data volume and complexity. Based on the correct identification point, the identification point should appear repeatedly in different layers. Through probability arbitration, "misidentified" points or invalid candidate points are filtered out, and finally effective candidate points, i.e. identification effective points, are obtained.

[0047] Specifically, in each layer of data, local likely neighbor points are used to calculate the "concave" degree, and a complete and accurate neighbor relationship tree is not established. Instead, the "likely" neighbor points of the selected points are obtained by projection and coordinate range setting, the calculation amount is reduced, the processing speed is improved, and then the centroid relative position and gradient of each point and the neighbor points are compared to calculate the concave degree. Finally, all candidate points obtained by analyzing each layer of data are analyzed, including identification effective points (candidate effective points) and peeling points (candidate invalid points). Figure 4 As shown, a selected point 3 is taken as an example for analysis, a selected projection point 31 is formed by projection, and surrounding projection points 41 of point 4 around the selected projection point 31 are formed by projection. Then, the corresponding surrounding projection points 41 are selected in the coordinate range 5 with the selected projection point 31 as the center, and the points in the coordinate range 5 are the likely neighbor points.

[0048] The identification of recesses in point cloud data requires the establishment of a neighboring point relationship tree. The traditional establishment of a neighboring tree requires the calculation of the "distance" of all points to the identified point, followed by screening. In the case of a large amount of point cloud data, a large amount of time and computing power is consumed. In the present application, the "likelihood" of the neighboring point is calculated by the likelihood neighboring method, which can greatly reduce the amount of calculation and improve the processing speed.

[0049] Due to the large amount of "peeling" on the surface of the workpiece after hot working, there is a large amount of false depth information or recess information at the "peeling" site, that is, the true marking point and the peeling point may have similar recess degrees, and the filtering speed based on "neighborhood" and "frequency" is slow, and it is easy to be mistaken, therefore, the invalid candidate point is the peeling point, by filtering out the peeling point as an invalid candidate point, the effectiveness of the obtained candidate point can be determined. In some embodiments, the features of the peeling point and the features of the marking point can be different, and the following distinguishing strategies can be used to realize the identification of the effective marking point 6 Figure 5 The small-area approximate circular multiple color deep black points) and the peeling point 7 Figure 5 The large-area approximate circular or elliptical multiple color gray points) are distinguished and screened, as shown in Figure 5

[0050] The marking effective points are continuously distributed, and the peeling points are dispersed and discontinuously distributed; the total number of the marking effective points is less than that of the peeling points, and the recess degree of the peeling points is higher than that of the marking effective points.

[0051] The embodiment of the present application filters the peeling point 7 from all candidate points based on the distribution of the feature values of the statistical marking point and the peeling point, and analyzes the spatial continuity between the candidate points, to obtain more accurate marking effective points, so that the peeling point 7 can be finally filtered from all candidate points based on the difference between the peeling point and the marking point, to finally obtain the candidate point with the maximum probability, that is, the marking effective point 6. The identification of the recess marking is realized through the analysis of the marking effective point.

[0052] The recess marking on the forged workpiece includes a two-dimensional code, a bar code or other similar marking code formed by mechanical processing on the surface of the workpiece.

[0053] The recess marking method of the forged workpiece in the embodiment of the present application can realize the rapid and effective identification of the recess marking on the surface of the forged workpiece by performing full coverage scanning on the surface of the forged workpiece through the 3D laser sensor to obtain the original point cloud data, and identifying the recess marking on the forged workpiece through the data processing method, so as to ensure the effective identification of the information of the forged workpiece in the production process, and facilitate the smooth production.

[0054] ​The second aspect of the embodiment of the present application provides a system for identifying concave marks on a forged workpiece, referring to Figure 6 As shown, the system comprises a 3D laser sensor and a data processing device: the 3D laser sensor is drivenly connected with a linear module, and after the linear module synchronously drives the 3D laser sensor to perform full-coverage scanning on the surface of the forged workpiece to obtain original point cloud data, the data processing device identifies the concave marks on the forged workpiece by using the data processing method according to the first aspect of the embodiment of the present application, and the specific processing steps are shown in Figure 1 .

[0055] The system for identifying concave marks on a forged workpiece according to the embodiment of the present application can quickly and effectively identify the concave marks on the surface of the forged workpiece after the 3D laser sensor performs full-coverage scanning on the surface of the forged workpiece to obtain original point cloud data, thereby ensuring effective identification of the information of the forged workpiece in the production process and facilitating smooth production.

[0056] The third aspect of the embodiment of the present application provides a storage medium, wherein the storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the data processing method for identifying the concave marks on the forged workpiece according to the first aspect of the present application.

[0057] The instruction or program in the storage medium is loaded and executed by the processor, and the method or steps for identifying the concave marks on the forged workpiece are shown in the identification processing steps of the data processing method involved in the method for identifying the concave marks on the forged workpiece according to the embodiment of the present application, which will not be described here.

[0058] The fourth aspect of the embodiment of the present application provides a data processing terminal or a data processing device, wherein the data processing terminal comprises a processor and a memory, and the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the data processing method for identifying the concave marks on the forged workpiece according to the first aspect of the present application.

[0059] The instruction or program in the storage medium is loaded and executed by the processor, and the method or steps for identifying the concave marks on the forged workpiece are shown in the identification processing steps of the data processing method involved in the method for identifying the concave marks on the forged workpiece according to the embodiment of the present application, which will not be described here.

[0060] The above merely describes the preferred embodiments of the present application, and it should be pointed out that, for those skilled in the art, several improvements and refinements can be made without departing from the principles of the present application, and these improvements and refinements should also be considered as falling within the protection scope of the present application.

Claims

1. A method of identifying a concave mark on a forged workpiece, characterized by, After the 3D laser sensor is used to scan the surface of the forged workpiece to obtain original point cloud data, the following data processing method is used to identify the concave marks on the forged workpiece: the original point cloud data obtained is preprocessed by denoising; The point cloud data obtained after the preprocessing by denoising is processed in layers to obtain multiple layers of data; For each layer of data, the local fitting projection method is used to obtain the likely neighbor points of the identification points within the coordinate range; The relative positions and gradients of the centroids of each identification point and the likely neighbor points are compared, the concave degree is calculated, and the initial candidate points are obtained from the identification points according to the concave degree; The initial candidate points obtained by analyzing the multiple layers of data are analyzed, and the invalid candidate points in the initial candidate points are filtered out to obtain the effective candidate points with the maximum probability as the identified mark effective points; The concave marks on the workpiece are identified based on the mark effective points; The invalid candidate points are skinning points; The identification of the mark effective points and the skinning points is realized based on the following distinguishing strategies: the mark effective points are concentrated and form a continuous curved surface, and the skinning points are scattered and discontinuously distributed; The total number of the mark effective points is less than that of the skinning points, The randomness of the concave degree of the skinning points is higher than that of the mark effective points.

2. The method of claim 1, wherein In the step of processing the point cloud data obtained after the preprocessing by denoising in layers to obtain multiple layers of data, the number, direction and thickness of the layers of the point cloud data are determined according to the normal vector and the characteristics of the data density distribution of the original point cloud data, and then the point cloud data is processed in layers according to the determined number, direction and thickness of the layers of the point cloud data.

3. The method of claim 1, wherein the step of identifying the indentation mark on the forged workpiece is characterized by, The 3D laser sensor is driven to work by a linear module through an encoder synchronization mode, and the scanning lens of the 3D laser sensor faces the scanning area of the forged workpiece.

4. The method of claim 1, wherein The wavelength of the 3D laser sensor is blue light.

5. The method of claim 1, wherein the step of identifying the indentation mark on the forged workpiece is characterized by, The concave marks on the forged workpiece include two-dimensional codes and bar codes.

6. A system for identifying recessed markings on forged workpieces, characterized in that, The data processing device includes a 3D laser sensor and a linear module driving connection, and the 3D laser sensor is synchronously driven by the linear module to scan the surface of the forged workpiece to obtain original point cloud data, and then the data processing device identifies the concave marks on the forged workpiece by the data processing method of any one of claims 1 to 5.

7. A storage medium, characterized by The storage medium stores at least one instruction, at least one program, a code set or an instruction set, which is loaded and executed by the processor to realize the data processing method of any one of claims 1 to 5 to identify the concave marks on the forged workpiece.

8. A data processing terminal, characterized by The data processing terminal includes a processor and a memory, and the memory stores at least one instruction, at least one program, a code set or an instruction set, which is loaded and executed by the processor to realize the data processing method of any one of claims 1 to 5 to identify the concave marks on the forged workpiece.

Images