A multi-dimensional real-time visual detection system and method for a turning machine machining process

Abstract

The application relates to the technical field of precision machining and machine vision detection, in particular to a multi-dimensional real-time visual detection system and method for a cam machine machining process, which comprises a multi-dimensional visual detection unit, an industrial computer, a standard workpiece database, a cam machine control system interface and a man-machine interaction unit. A plurality of industrial cameras and optical sensors are arranged above, on the side and on the end surface of a cam machine machining station to realize multi-dimensional image and contour information collection of a workpiece. The detection method is as follows: after each machining process is completed, the current workpiece image is collected, pretreatment and multi-feature extraction are completed, the extracted features are compared with the standard features of the corresponding process in the standard workpiece database in terms of similarity, and the workpiece cutting in-place state is detected. When an abnormality occurs, the machine is stopped and an alarm is triggered. The application can improve the existing cam machine machining quality detection lag, realize inter-process detection and lack of machining state monitoring, and improve the detection precision.

Description

Technical Field

[0001] This invention relates to the technical field of precision machining and machine vision inspection, and in particular to a multi-dimensional real-time visual inspection system and method for Swiss-type lathe machining processes. Background Technology

[0002] Swiss-type lathes operate on the principle of "material moves while tool remains stationary," where the spindle holds the bar stock, which rotates while being fed linearly through various tool stations to complete the machining process. Swiss-type lathes possess powerful multi-processing capabilities, capable of completing almost all operations such as turning, milling, drilling, tapping, and grooving in a single setup. They are particularly suitable for the efficient production of precision, slender shafts and irregularly shaped shafts, and are widely used in high-end manufacturing fields such as medical devices, electronic connectors, and automotive parts.

[0003] Currently, quality inspection during the Swiss-type lathe machining process mainly relies on manual sampling or offline inspection after machining. This method not only suffers from significant lag, only detecting problems after the entire batch of parts has been machined, leading to a large number of defective products and a double waste of materials and time, but also suffers from significant influence on inspection accuracy due to operator subjectivity. It is difficult to accurately identify minute dimensional deviations and surface defects, making it difficult to guarantee inspection accuracy and consistency. More importantly, the existing inspection method cannot achieve inter-process inspection, failing to promptly assess the machining quality of each process after completion. This leads to subsequent processes continuing to process defective semi-finished products, further increasing production costs. Furthermore, the lack of real-time monitoring of the workpiece's machining status makes it impossible to promptly detect abnormalities such as whether the workpiece is fully machined or whether the cut-off was successful, easily leading to equipment failures and safety accidents.

[0004] Although there have been attempts to apply visual inspection technology to Swiss-type lathe machining, most of them use a single camera for overall inspection, which has the problem of limited inspection dimensions. It can only obtain an image of the workpiece from one angle and cannot comprehensively inspect the machining quality of each surface of the workpiece. It is easy to miss defects on the side and end faces. At the same time, the comparison technology used is relatively simple, with high false detection and false detection rates. Furthermore, there is no corresponding detection algorithm and process designed for the characteristics of different machining processes of Swiss-type lathes. The inspection efficiency and accuracy are difficult to meet the actual production needs of precision manufacturing. Summary of the Invention

[0005] To address the issues of lagging quality inspection, inability to perform inter-process inspection, and lack of processing status monitoring in existing Swiss-type lathes, and to improve inspection accuracy, this application provides a multi-dimensional real-time visual inspection system and method for the Swiss-type lathe machining process.

[0006] This application provides a multi-dimensional real-time visual inspection system and method for the machining process of Swiss-type lathes, which adopts the following technical solution:

[0007] A multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe includes the following steps:

[0008] S1: System initialization, based on the model of the workpiece to be processed, retrieves the corresponding standard image and feature data from the standard workpiece database;

[0009] S2: The Swiss-type lathe starts processing. After completing a processing step, the control system of the Swiss-type lathe sends a process completion signal to the industrial computer through the Swiss-type lathe control system interface.

[0010] S3: The industrial computer controls the multi-dimensional vision inspection unit to collect image information and three-dimensional contour information of the current workpiece from different angles;

[0011] S4: Preprocess the acquired images, including image denoising, brightness normalization, geometric correction, and image enhancement;

[0012] S5: Extract feature data from the preprocessed image, including edge features, texture features, shape features, and color features;

[0013] S6: Compare the extracted feature data with the standard feature data of the corresponding process in the standard workpiece database, and calculate the similarity.

[0014] S7: Determine whether the processing quality of this process is qualified based on the similarity. If it is not qualified, generate a defect report and display it through the human-machine interaction unit, and send a stop command to the Swiss-type lathe control system at the same time.

[0015] S8: Detects whether the previous workpiece has been completely cut off and left the processing station. If it has not been cut off, it sends an alarm signal to the Swiss-type lathe control system and prohibits the next workpiece from starting processing.

[0016] S9: If the processing quality of this process is qualified and the previous workpiece has been cut off normally, the Swiss-type lathe will continue to process the next process.

[0017] S10: Repeat steps S2 to S9 until all processing steps are completed.

[0018] Preferably, the preprocessing in step S4 specifically includes:

[0019] S41: Use median filtering algorithm to remove salt-and-pepper noise from the image;

[0020] S42: The image brightness is normalized using a histogram equalization algorithm to eliminate the influence of illumination changes;

[0021] S43: Based on the position information of the Swiss-type lathe spindle and the calibration parameters of the camera, perform geometric correction on the image to eliminate the influence of workpiece posture deviation;

[0022] S44: The Retinex algorithm is used to enhance the image, improving its contrast and clarity.

[0023] Preferably, the specific method for extracting the feature data in step S5 includes:

[0024] S51: The Canny edge detection algorithm is used to extract the edge features of the image;

[0025] S52: Extract texture features of the image using the Gray-Level Co-occurrence Matrix (GLCM), including contrast, correlation, energy, and entropy;

[0026] S53: The shape features of the image are extracted using Hu invariant moments, which have invariance to translation, rotation and scaling.

[0027] S54: Convert the image to the HSV color space and extract color histogram features.

[0028] Preferably, the comparison method in step S6 is as follows:

[0029] S61: For edge features, the improved Hausdorff distance is used to calculate similarity;

[0030] S62: For texture features and color features, cosine similarity is used to calculate similarity;

[0031] S63: For shape features, Euclidean distance is used to calculate similarity;

[0032] S64: The similarity of the different features mentioned above is fused using a weighted fusion method to obtain a comprehensive similarity.

[0033] Preferably, the defect report in step S7 includes the defect type, defect location, defect size, and defect severity.

[0034] Preferably, step S8 is as follows:

[0035] S81: Collect the background image of the processing station as a reference image;

[0036] S82: After the current process is completed, acquire real-time images of the processing station;

[0037] S83: Perform differential processing on the real-time image and the reference image to obtain a differential image;

[0038] S84: Perform binarization and morphological operations on the difference image to extract the foreground region;

[0039] S85: Calculate the area of ​​the foreground region. If the area is greater than the preset threshold, it is determined that the workpiece has not been cut off; otherwise, it is cut off.

[0040] Preferably, the tool wear detection step is also included: the tool wear is indirectly detected by analyzing the texture features and temperature distribution features of the workpiece surface. If the similarity of the roughness features is lower than a preset threshold or the cutting temperature is higher than a preset threshold, the tool wear is judged to be severe, and a tool wear alarm message is generated.

[0041] A system for implementing the above-mentioned multi-dimensional real-time visual inspection method for Swiss-type lathe machining processes includes:

[0042] The multi-dimensional vision inspection unit includes at least three industrial cameras and corresponding light source systems, which are respectively installed above, on the side and end face of the Swiss-type machining station to acquire image information of the workpiece from different angles;

[0043] An industrial control computer is communicatively connected to the multi-dimensional vision detection unit, used to receive and process the acquired image information, and execute image comparison algorithms and detection logic;

[0044] A standard workpiece database is stored in the industrial control computer and is used to store standard images and feature data of different types of workpieces after each processing step.

[0045] The Swiss-type lathe control system interface is used to realize communication between the industrial computer and the control system of the Swiss-type lathe, receive processing signals and send detection results and control commands;

[0046] The human-machine interaction unit is connected to the industrial control computer and is used to display detection results, set system parameters, and perform manual intervention.

[0047] Preferably, the multi-dimensional visual detection unit further includes:

[0048] A 3D laser contour sensor is installed above the machining station to collect the three-dimensional contour information of the workpiece and achieve accurate measurement of dimensions.

[0049] Infrared thermal imaging cameras are used to collect information on the surface temperature distribution of workpieces after processing, in order to help detect processing defects and tool wear.

[0050] Preferably, the light source system includes:

[0051] A ring light source, mounted around the lens of the industrial camera above the processing station, is used to provide uniform frontal illumination;

[0052] A strip light source, mounted on one side of the industrial camera at the side of the processing station, is used to provide side lighting to highlight the surface texture and edge features of the workpiece;

[0053] A coaxial light source is installed in front of the lens of the industrial camera at the end face of the processing station to eliminate reflections on the workpiece end face and improve image contrast.

[0054] In summary, this application includes the following beneficial technical effects:

[0055] 1. This invention enables immediate quality inspection after each processing step of the Swiss-type lathe is completed, promptly identifying problems in the processing process and avoiding further processing on unqualified semi-finished products, thus greatly reducing production costs and material waste.

[0056] 2. This invention uses multiple industrial cameras to acquire workpiece images from different angles such as above, side and end face. Combined with a 3D laser contour sensor and an infrared thermal imaging camera, it can comprehensively detect the processing quality and dimensional accuracy of each surface of the workpiece, effectively improving the problem of missed detection.

[0057] 3. This invention adopts a combination of multiple feature extraction and similarity calculation methods, including edge features, texture features, shape features and color features, and obtains a comprehensive similarity through weighted fusion, which greatly improves the accuracy of image comparison and anti-interference ability. It also has good anti-interference ability against interference factors such as changes in lighting, workpiece posture deviation, and surface oil stains.

[0058] 4. This invention can detect in real time whether the previous workpiece has been completely cut off and left the processing station, effectively avoiding equipment failures and safety accidents caused by workpieces not being cut off, and improving the safety and stability of production.

[0059] 5. This invention can automatically generate defect reports, displaying the defect type, location, size, and severity, and can automatically control the operation of the Swiss-type lathe based on the detection results, reducing manual intervention and improving production efficiency and automation level. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the multi-dimensional real-time visual inspection system for the Swiss-type lathe machining process in this invention;

[0061] Figure 2 This is a flowchart of the multi-dimensional real-time visual inspection method for the Swiss-type lathe machining process in this invention;

[0062] Figure 3 This is a flowchart of the image preprocessing steps in this invention;

[0063] Figure 4 This is a flowchart of the feature data extraction steps in this invention;

[0064] Figure 5 This is a flowchart of the feature comparison and analysis steps in this invention;

[0065] Figure 6 This is a flowchart of the workpiece cutting status detection step in this invention.

[0066] Explanation of reference numerals in the attached diagram: 1. Swiss-type lathe; 2. Multi-dimensional vision inspection unit; 3. Industrial computer; 31. Standard workpiece database; 32. Swiss-type lathe control system interface; 33. Human-machine interaction unit. Detailed Implementation

[0067] To enable those skilled in the art to better understand the present invention, the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0068] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation or specific orientational structure and operation. Therefore, they should not be construed as limitations on the invention. The terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, unless otherwise explicitly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0069] This application discloses a multi-dimensional real-time visual inspection system and method for the machining process of a Swiss-type lathe.

[0070] Reference Figure 1 The multi-dimensional real-time visual inspection system for the Swiss-type lathe machining process includes a multi-dimensional visual inspection unit 2, an industrial computer 3, a standard workpiece database 31, a control system interface for the Swiss-type lathe 1, and a human-machine interaction unit 33.

[0071] The multi-dimensional vision inspection unit 2 is connected to the industrial control computer 3 for acquiring image information and three-dimensional contour information of the workpiece from different angles. The industrial control computer 3 is the core of the entire system, used to receive and process the acquired image information, execute image comparison algorithms and detection logic. The standard workpiece database 31 is stored in the industrial control computer 3 for storing standard images and feature data of different types of workpieces after each processing step. The Swiss-type lathe 1 control system interface is used to realize communication between the industrial control computer 3 and the Swiss-type lathe 1 control system, receive processing step signals and send detection results and control commands. The human-machine interaction unit 33 is connected to the industrial control computer 3 for displaying detection results, setting system parameters and performing manual intervention, usually using a touch screen or industrial display.

[0072] The multi-dimensional visual inspection unit 2 includes an upper industrial camera, a side industrial camera, an end-face industrial camera, a 3D laser contour sensor, an infrared thermal imaging camera, and a corresponding light source system. The upper industrial camera is installed directly above the machining station of the Swiss-type lathe 1 to acquire a top view of the workpiece; the side industrial camera is installed on the side of the machining station to acquire a side view of the workpiece; the end-face industrial camera is installed on the end face of the machining station to acquire an end-face view of the workpiece; the 3D laser contour sensor is installed above the machining station, side by side with the upper industrial camera, to acquire the three-dimensional contour information of the workpiece and achieve accurate measurement of dimensional accuracy; the infrared thermal imaging camera is installed above the machining station to acquire the surface temperature distribution information of the workpiece after machining, assisting in the detection of machining defects and tool wear.

[0073] The light source system includes a ring light source, a bar light source, and a coaxial light source. The ring light source is installed around the lens of the upper industrial camera to provide uniform front lighting and reduce shadows. The bar light source is installed on one side of the side industrial camera, parallel to the axis of the workpiece, to provide side lighting and highlight the surface texture and edge features of the workpiece. The coaxial light source is installed in front of the lens of the end face industrial camera to eliminate reflections on the end face of the workpiece and improve image contrast.

[0074] Among them, the industrial camera adopts a high-resolution global shutter industrial camera with a resolution of no less than 5 million pixels and a frame rate of no less than 30fps, which can meet the image acquisition requirements in high-speed processing. The 3D laser contour sensor adopts a line laser contour sensor with a measurement accuracy of no less than ±0.005mm, which can realize high-precision measurement of workpiece size. The infrared thermal imaging camera adopts an uncooled infrared thermal imaging camera with a resolution of no less than 320×240 pixels and a temperature measurement range of -20℃ to 300℃.

[0075] Example 1

[0076] This embodiment provides a multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe, referring to... Figures 2-6This includes the following steps:

[0077] S1: System initialization. Based on the model of the workpiece to be processed, the corresponding standard image and feature data are retrieved from the standard workpiece database 31. The standard image and feature data are collected during the system debugging phase by collecting images of qualified workpieces after each processing step, extracting features, and storing them in the database.

[0078] S2: The Swiss-type lathe 1 starts processing. After completing a processing operation, such as external turning, end milling, drilling, etc., the control system of the Swiss-type lathe 1 sends a process completion signal to the industrial computer 3 through the control system interface of the Swiss-type lathe 1.

[0079] S3: The industrial computer 3 controls the multi-dimensional vision inspection unit 2 to collect image information and three-dimensional contour information of the current workpiece from different angles. Specifically, the top industrial camera collects the top view of the workpiece, the side industrial camera collects the side view of the workpiece, the end face industrial camera collects the end face view of the workpiece, the 3D laser contour sensor collects the three-dimensional contour data of the workpiece, and the infrared thermal imaging camera collects the surface temperature distribution image of the workpiece.

[0080] S4: Preprocess the acquired images, such as... Figure 3 As shown, it specifically includes:

[0081] S41: The median filtering algorithm is used to remove salt and pepper noise from the image. The window size of the median filter is 3×3.

[0082] S42: The image brightness is normalized using a histogram equalization algorithm to eliminate the influence of illumination changes;

[0083] S43: Based on the position information of the main spindle of the Swiss-type lathe 1 and the calibration parameters of the camera, perform geometric correction on the image to eliminate the influence of workpiece posture deviation. The camera calibration is performed after the system is installed. The intrinsic and extrinsic parameters of the camera are obtained through the calibration board, and the mapping relationship between the image coordinate system and the world coordinate system is established.

[0084] S44: The Retinex algorithm is used to enhance the image, improving its contrast and clarity. In particular, it can effectively remove the effects of oil stains on workpieces with oily surfaces.

[0085] S5: Extract feature data from the preprocessed image, such as Figure 4 As shown, it specifically includes:

[0086] S51: The Canny edge detection algorithm is used to extract the edge features of the image. The high and low thresholds of the Canny algorithm are set to 0.7 and 0.3, respectively.

[0087] S52: Use the Gray-Level Co-occurrence Matrix (GLCM) to extract the texture features of the image, including contrast, correlation, energy and entropy. The step size of the GLCM is 1, and the orientation is 0°, 45°, 90° and 135°.

[0088] S53: The shape features of the image are extracted using Hu invariant moments. Hu invariant moments have translation, rotation and scaling invariance, which can effectively describe the shape features of the workpiece.

[0089] S54: Convert the image to the HSV color space, extract the color histogram features, and set the number of bins in the color histogram to 16.

[0090] Simultaneously, the 3D contour data acquired by the 3D laser contour sensor is processed to extract the workpiece's dimensional features, including diameter, length, step height, and aperture. The temperature distribution images acquired by the infrared thermal imaging camera are processed to extract temperature features, including maximum temperature, minimum temperature, average temperature, and temperature distribution uniformity.

[0091] S6: Compare the extracted feature data with the standard feature data of the corresponding process in the standard workpiece database 31, and calculate the similarity, such as... Figure 5 As shown, it specifically includes:

[0092] S61: For edge features, the improved Hausdorff distance is used to calculate similarity. The improved Hausdorff distance can effectively reduce the influence of outliers and improve the accuracy of edge matching.

[0093] S62: For texture features and color features, cosine similarity is used to calculate similarity. Cosine similarity can effectively measure the angle between two vectors and is not sensitive to the length of the vectors.

[0094] S63: For shape and size features, Euclidean distance is used to calculate similarity;

[0095] S64: For temperature features, the similarity is calculated using the absolute difference.

[0096] S65: The similarity of the above different features is fused using a weighted fusion method to obtain a comprehensive similarity. The weights can be set according to the importance of different features. For example, for workpieces with high dimensional accuracy requirements, the weight of dimensional features can be increased.

[0097] S7: Determine whether the processing quality of this process is qualified based on the overall similarity. If the overall similarity is greater than the preset qualification threshold, it is judged as qualified; otherwise, it is judged as unqualified.

[0098] If the product is not up to standard, a defect report is generated and displayed through the human-machine interaction unit 33. The defect report includes the defect type (such as dimensional deviation, surface scratches, burrs, missing material, etc.), defect location, defect size, and defect severity. At the same time, the industrial control computer 3 sends a stop command to the control system of the Swiss-type lathe 1 through the control system interface to prevent the continued processing of defective products.

[0099] S8: Check whether the previous workpiece has been completely cut off and left the processing station, such as... Figure 6 As shown, it specifically includes:

[0100] S81: During the system debugging phase, a background image of the processing station when there is no workpiece is collected as a reference image;

[0101] S82: After the current process is completed, acquire real-time images of the processing station;

[0102] S83: Perform differential processing on the real-time image and the reference image to obtain a differential image;

[0103] S84: Binarize the difference image, use the Otsu algorithm to automatically determine the binarization threshold, and then perform morphological opening and closing operations to remove noise and small connected regions and extract the foreground region;

[0104] S85: Calculate the area of ​​the foreground region. If the area is greater than the preset threshold, it is determined that the previous workpiece has not been cut off or is still in the processing station. At this time, the industrial control computer 3 sends an alarm signal to the control system of the Swiss-type machine 1 through the control system interface of the Swiss-type machine 1 and prohibits the next workpiece from starting processing until the fault is manually eliminated.

[0105] S9: If the processing quality of this process is qualified and the previous workpiece has been cut off normally, the Swiss-type machine 1 continues to process the next process.

[0106] S10: Repeat steps S2 to S9 until all processing steps are completed. Once all steps are completed, the system generates a complete inspection report, recording the processing quality and inspection time of each workpiece.

[0107] Example 2

[0108] This embodiment adds a tool wear detection function based on embodiment 1. During the machining process, tool wear will lead to a decrease in the surface quality and dimensional accuracy of the workpiece. This invention indirectly detects the tool wear by analyzing the texture features and temperature distribution features of the workpiece surface.

[0109] When the tool wears, the surface roughness of the workpiece increases, the texture features change, and the cutting temperature rises. Therefore, in step S5, in addition to extracting the above features, the surface roughness features and cutting temperature features of the workpiece are also extracted. In step S6, these features are compared with standard features. If the similarity of the roughness features is lower than a preset threshold or the cutting temperature is higher than a preset threshold, it is determined that the tool is severely worn and needs to be replaced. The system will generate a tool wear alarm message and display it through the human-machine interaction unit 33 to remind the operator to replace the tool in time.

[0110] Finally, it should be noted that the above description is only a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be pointed out that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe, characterized in that, Includes the following steps: S1: System initialization, according to the model of the workpiece to be processed, retrieve the corresponding standard image and feature data from the standard workpiece database (31); S2: The Swiss-type lathe (1) starts processing. After completing a processing step, the control system of the Swiss-type lathe (1) sends a process completion signal to the industrial computer (3) through the control system interface of the Swiss-type lathe (1). S3: The industrial computer (3) controls the multi-dimensional vision inspection unit (2) to collect image information and three-dimensional contour information of the current workpiece from different angles; S4: Preprocess the acquired images, including image denoising, brightness normalization, geometric correction, and image enhancement; S5: Extract feature data from the preprocessed image, including edge features, texture features, shape features, and color features; S6: Compare the extracted feature data with the standard feature data of the corresponding process in the standard workpiece database (31) and calculate the similarity; S7: Based on the similarity, determine whether the processing quality of the process is qualified. If it is not qualified, generate a defect report and display it through the human-machine interaction unit (33), and send a stop command to the control system of the sliding head machine (1). S8: Detect whether the previous workpiece has been completely cut off and left the processing station. If it has not been cut off, send an alarm signal to the Swiss Army Machine (1) control system and prohibit the next workpiece from starting processing. S9: If the processing quality of this process is qualified and the previous workpiece has been cut off normally, the Swiss Army Machine (1) continues to process the next process; S10: Repeat steps S2 to S9 until all processing steps are completed.

2. The multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe according to claim 1, characterized in that, The preprocessing described in step S4 specifically includes: S41: Use median filtering algorithm to remove salt-and-pepper noise from the image; S42: The image brightness is normalized using a histogram equalization algorithm to eliminate the influence of illumination changes; S43: Based on the position information of the Swiss-type lathe spindle and the calibration parameters of the camera, perform geometric correction on the image to eliminate the influence of workpiece posture deviation; S44: The Retinex algorithm is used to enhance the image, improving its contrast and clarity.

3. The multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe according to claim 1, characterized in that, The specific methods for extracting the feature data in step S5 include: S51: The Canny edge detection algorithm is used to extract the edge features of the image; S52: Extract texture features of the image using the gray-level co-occurrence matrix, including contrast, correlation, energy, and entropy; S53: The shape features of the image are extracted using Hu invariant moments, which have invariance to translation, rotation and scaling. S54: Convert the image to the HSV color space and extract color histogram features.

4. The multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe according to claim 1, characterized in that, The comparison method in step S6 is as follows: S61: For edge features, the improved Hausdorff distance is used to calculate similarity; S62: For texture features and color features, cosine similarity is used to calculate similarity; S63: For shape features, Euclidean distance is used to calculate similarity; S64: The similarity of the different features mentioned above is fused using a weighted fusion method to obtain a comprehensive similarity.

5. A multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe according to claim 1, characterized in that, The defect report in step S7 includes the defect type, defect location, defect size, and defect severity.

6. The multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe according to claim 1, characterized in that, Step S8 is as follows: S81: Collect the background image of the processing station as a reference image; S82: After the current process is completed, acquire real-time images of the processing station; S83: Perform differential processing on the real-time image and the reference image to obtain a differential image; S84: Perform binarization and morphological operations on the difference image to extract the foreground region; S85: Calculate the area of ​​the foreground region. If the area is greater than the preset threshold, it is determined that the workpiece has not been cut off; otherwise, it is cut off.

7. A multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe according to claim 1, characterized in that, It also includes a tool wear detection step: the tool wear is indirectly detected by analyzing the texture features and temperature distribution features of the workpiece surface. If the similarity of the roughness features is lower than the preset threshold or the cutting temperature is higher than the preset threshold, it is judged that the tool wear is severe and a tool wear alarm message is generated.

8. A system for implementing the multi-dimensional real-time visual inspection method for the machining process of a Swiss-type lathe as described in any one of claims 1-7, characterized in that, include: The multi-dimensional visual inspection unit (2) includes at least three industrial cameras and corresponding light source systems, which are respectively installed above, on the side and at the end face of the machining station of the Swiss-type lathe (1) to collect image information of the workpiece from different angles; The industrial control computer (3) is connected to the multi-dimensional vision detection unit (2) for receiving and processing the collected image information, and executing image comparison algorithms and detection logic; A standard workpiece database (31) is stored in the industrial control computer (3) and is used to store standard images and feature data of different types of workpieces after each processing step. The Swiss-type lathe (1) control system interface is used to realize the communication between the industrial computer (3) and the control system of the Swiss-type lathe (1), receive processing process signals and send detection results and control instructions; The human-machine interaction unit (33) is connected to the industrial control computer (3) and is used to display the detection results, set system parameters and perform manual intervention.

9. A multi-dimensional real-time visual inspection system for the machining process of a Swiss-type lathe according to claim 8, characterized in that, The multi-dimensional visual detection unit (2) also includes: A 3D laser contour sensor is installed above the machining station to collect the three-dimensional contour information of the workpiece and achieve accurate measurement of dimensions. Infrared thermal imaging cameras are used to collect information on the surface temperature distribution of workpieces after processing, in order to help detect processing defects and tool wear.

10. A multi-dimensional real-time visual inspection system for the machining process of a Swiss-type lathe according to claim 8, characterized in that, The light source system includes: A ring light source, mounted around the lens of the industrial camera above the processing station, is used to provide uniform frontal illumination; A strip light source, mounted on one side of the industrial camera at the side of the processing station, is used to provide side lighting to highlight the surface texture and edge features of the workpiece; A coaxial light source is installed in front of the lens of the industrial camera at the end face of the processing station to eliminate reflections on the workpiece end face and improve image contrast.

Images