High-precision numerical control machine tool workpiece positioning system based on machine vision

Abstract

The invention discloses a high-precision numerical control machine tool workpiece positioning system based on machine vision, and aims to improve the machining precision and the production efficiency of a numerical control machine tool through an automatic and high-precision vision technology. The system comprises an image acquisition and processing unit, a feature extraction and recognition unit, a three-dimensional reconstruction positioning unit, an algorithm optimization unit and a user interface. The image acquisition unit is responsible for acquiring a workpiece image, the feature extraction unit extracts workpiece features by using an image processing algorithm, the three-dimensional reconstruction unit reconstructs a three-dimensional model from a two-dimensional image and determines a position, and the algorithm optimization unit adopts a parallel computing method to improve the processing speed and precision. And the user interface provides real-time image display and operation instruction input. According to the system, through automatic image acquisition, efficient feature extraction, accurate three-dimensional reconstruction and parallel computing optimization, rapid and accurate positioning of workpieces is achieved, the system is suitable for variable machining environments and workpiece types, and the machining precision and production efficiency of a numerical control machine tool are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of numerical control machining, and in particular to a high-precision numerical control machine tool workpiece positioning system based on machine vision. Background Art

[0002] In modern manufacturing, the accuracy and efficiency of CNC machine tools have a decisive impact on product quality and production costs. Traditional workpiece positioning methods rely on mechanical fixtures and operator visual inspection, which have the following limitations:

[0003] Inefficiency: The manual positioning and inspection process is time-consuming and fatigue-prone, resulting in limited production efficiency.

[0004] Limited precision: The accuracy of human eye recognition and mechanical fixtures cannot meet the needs of high-precision processing.

[0005] Lack of flexibility: For the changing shapes and sizes of workpieces, traditional methods require frequent adjustments to fixtures and settings, which lacks flexibility.

[0006] Cost issues: Specialized fixtures and equipment are expensive and difficult to adapt to rapidly changing production needs.

[0007] In order to overcome these challenges, machine vision technology has been introduced into the workpiece positioning system of CNC machine tools. Machine vision can provide automated, high-precision detection and analysis capabilities, but it still faces the following technical difficulties in practical applications:

[0008] Image processing speed: In real-time or near real-time application scenarios, quickly processing large amounts of image data is a major challenge.

[0009] Stability in complex environments: The diversity of workpieces and changes in the processing environment place higher demands on the stability and robustness of the vision system.

[0010] Optimization of computing resources: High-precision vision processing algorithms are often computationally intensive and require a lot of computing resources. Summary of the invention

[0011] The purpose of the present invention is to provide a high-precision CNC machine tool workpiece positioning system based on machine vision.

[0012] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0013] The high-precision CNC machine tool workpiece positioning system based on machine vision of the present invention comprises an image acquisition and processing unit, a feature extraction and recognition unit, a three-dimensional reconstruction and positioning unit, an algorithm optimization unit and a user interface, wherein the data output end of the image acquisition and processing unit is connected to the feature extraction and recognition unit, the data output end of the feature extraction and recognition unit is connected to the data input end of the three-dimensional reconstruction and positioning unit, the data output end of the three-dimensional reconstruction and positioning unit is connected to the algorithm optimization unit, and the control signal transmission end of the algorithm optimization unit is connected to the user interface.

[0014] The image acquisition and processing unit obtains the image of the workpiece and captures sufficient details according to the characteristics of the workpiece and the processing environment. The image processing includes graying, filtering and denoising, and edge enhancement. The feature extraction and recognition unit extracts the edge, corner point, and texture features of the workpiece through an image processing algorithm. The three-dimensional reconstruction and positioning unit reconstructs the three-dimensional model of the workpiece from the two-dimensional image and determines its position in the machine tool coordinate system. The algorithm optimization unit optimizes the algorithm using a parallel computing method. The user interface displays the real-time image of the workpiece, the positioning result, and the machine tool status, and provides necessary operating instructions.

[0015] The positioning method of the high-precision CNC machine tool workpiece positioning system based on machine vision of the present invention comprises the following steps:

[0016] S1: Image acquisition and processing unit

[0017] S1.1: Select industrial cameras and lenses to ensure that they can adapt to the characteristics of the workpiece and the processing environment and capture sufficient details;

[0018] S1.2: Set the camera exposure time and gain to obtain the best image quality;

[0019] S1.3: Perform image preprocessing, including grayscale, filtering, denoising, and edge enhancement, to improve the accuracy of subsequent processing;

[0020] S2: Feature extraction and recognition unit

[0021] S2.1: Apply image processing algorithms, including Canny edge detection and Hough transform, to extract the edges, corners and texture features of the workpiece;

[0022] S2.2: Use machine learning or deep learning models to classify and identify the extracted features to determine the type and characteristic features of the artifact;

[0023] S3: 3D reconstruction positioning unit

[0024] S3.1: Reconstruct a 3D model of the workpiece from a 2D image using stereo vision, structured light, or laser triangulation techniques;

[0025] S3.2: Determine the precise position and orientation of the workpiece in the machine tool coordinate system through camera calibration and stereo matching algorithm;

[0026] S4: Algorithm Optimization Unit

[0027] S4.1: Optimize 3D reconstruction and feature recognition algorithms to improve processing speed and accuracy;

[0028] S4.2: Use parallel computing methods, such as GPU acceleration, to handle large amounts of image data and computing tasks;

[0029] S5: User Interface

[0030] S5.1: Design an intuitive user interface that displays real-time images of the workpiece, positioning results, and machine tool status;

[0031] S5.2: Provides necessary operating instructions to allow the user to input parameters, start the positioning process and adjust system settings.

[0032] The parallel computing method comprises:

[0033] Data parallelism: image data is divided into multiple small blocks, and each processing unit processes one data block to improve processing speed;

[0034] Task parallelism: decompose the image processing task into multiple subtasks and execute these subtasks simultaneously on multiple processing units;

[0035] Spatial parallelism: Use loops and recursive structures in image processing algorithms to parallelize tasks;

[0036] GPU acceleration: Use the parallel processing capabilities of the GPU to process images at high speed;

[0037] Optimization algorithm: Design efficient parallel algorithms by combining data parallelism, task parallelism and space parallelism with specific image processing tasks;

[0038] Code implementation: When implementing parallel computing, OpenCV is combined with CUDA to utilize CUDA's parallel computing capabilities to accelerate image processing tasks.

[0039] The beneficial effects of the present invention are:

[0040] The present invention is a high-precision CNC machine tool workpiece positioning system based on machine vision. Compared with the prior art, the present invention significantly improves positioning accuracy and efficiency through the following technical innovations:

[0041] Automated image acquisition: High-resolution industrial cameras and automated control strategies are used to ensure high-quality workpiece images in a variety of processing environments.

[0042] Efficient feature extraction: Utilize advanced image processing algorithms and deep learning models to quickly and accurately extract key features of workpieces.

[0043] Accurate 3D reconstruction: Combining stereo vision and structured light technology, it can achieve accurate reconstruction from 2D images to 3D models and accurately locate the position of the workpiece in the machine tool coordinate system.

[0044] Parallel computing optimization: Through parallel computing methods, such as GPU acceleration, the speed of image processing and feature recognition is significantly improved to meet the needs of real-time processing.

[0045] User-friendly interface: Design an intuitive user interface to simplify the operation process, improve user experience, and reduce operation difficulty. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a system structure principle diagram of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0048] like Figure 1 As shown: The high-precision CNC machine tool workpiece positioning system based on machine vision of the present invention includes an image acquisition and processing unit, a feature extraction and recognition unit, a three-dimensional reconstruction and positioning unit, an algorithm optimization unit and a user interface, the data output end of the image acquisition and processing unit is connected to the feature extraction and recognition unit, the data output end of the feature extraction and recognition unit is connected to the data input end of the three-dimensional reconstruction and positioning unit, the data output end of the three-dimensional reconstruction and positioning unit is connected to the algorithm optimization unit, and the control signal transmission end of the algorithm optimization unit is connected to the user interface.

[0049] The image acquisition and processing unit obtains the image of the workpiece and captures sufficient details according to the characteristics of the workpiece and the processing environment. The image processing includes graying, filtering and denoising, and edge enhancement. The feature extraction and recognition unit extracts the edge, corner point, and texture features of the workpiece through an image processing algorithm. The three-dimensional reconstruction and positioning unit reconstructs the three-dimensional model of the workpiece from the two-dimensional image and determines its position in the machine tool coordinate system. The algorithm optimization unit optimizes the algorithm using a parallel computing method. The user interface displays the real-time image of the workpiece, the positioning result, and the machine tool status, and provides necessary operating instructions.

[0050] Image acquisition and preprocessing: First, you need to select a suitable industrial camera and lens to acquire images of the workpiece. Depending on the characteristics of the workpiece and the processing environment, you may need to use a high-resolution camera to capture enough details. Image preprocessing steps may include grayscale, filtering and denoising, edge enhancement, etc. to improve the accuracy of subsequent image analysis.

[0051] Feature extraction and recognition: Extract key features of workpieces such as edges, corners, textures, etc. through image processing algorithms. Traditional image processing techniques such as Canny edge detection, Hough transform, etc., or deep learning methods such as convolutional neural networks (CNN) can be used to improve the accuracy of feature recognition.

[0052] 3D reconstruction and positioning: Using technologies such as stereo vision, structured light or laser triangulation, the 3D model of the workpiece is reconstructed from the 2D image and its position in the machine tool coordinate system is determined. This may involve technologies such as camera calibration, stereo matching, and point cloud processing.

[0053] Algorithm optimization: In order to improve the processing speed and accuracy of the system, the algorithm needs to be optimized. This may include using faster image processing libraries, optimizing algorithm flow, and adopting parallel computing methods.

[0054] System integration and testing: Integrate the above technologies into the control system of the CNC machine tool and conduct sufficient testing to ensure the stability and reliability of the system. This may involve communicating with the machine tool's control system, processing image data in real time, adjusting the machine tool's movement, etc.

[0055] User interface design: In order to facilitate operators to use, it is necessary to design an intuitive user interface to display the real-time image of the workpiece, positioning results and machine tool status, and provide necessary operating instructions.

[0056] The positioning method of the high-precision CNC machine tool workpiece positioning system based on machine vision of the present invention comprises the following steps:

[0057] S1: Image acquisition and processing unit

[0058] S1.1: Select industrial cameras and lenses to ensure that they can adapt to the characteristics of the workpiece and the processing environment and capture sufficient details;

[0059] S1.2: Set the camera exposure time and gain to obtain the best image quality;

[0060] S1.3: Perform image preprocessing, including grayscale, filtering, denoising, and edge enhancement, to improve the accuracy of subsequent processing;

[0061] S2: Feature extraction and recognition unit

[0062] S2.1: Apply image processing algorithms, including Canny edge detection and Hough transform, to extract the edges, corners and texture features of the workpiece;

[0063] S2.2: Use machine learning or deep learning models to classify and identify the extracted features to determine the type and characteristic features of the artifact;

[0064] S3: 3D reconstruction positioning unit

[0065] S3.1: Reconstruct a 3D model of the workpiece from a 2D image using stereo vision, structured light, or laser triangulation techniques;

[0066] S3.2: Determine the precise position and orientation of the workpiece in the machine tool coordinate system through camera calibration and stereo matching algorithm;

[0067] S4: Algorithm Optimization Unit

[0068] S4.1: Optimize 3D reconstruction and feature recognition algorithms to improve processing speed and accuracy;

[0069] S4.2: Use parallel computing methods, such as GPU acceleration, to handle large amounts of image data and computing tasks;

[0070] S5: User Interface

[0071] S5.1: Design an intuitive user interface that displays real-time images of the workpiece, positioning results, and machine tool status;

[0072] S5.2: Provides necessary operating instructions to allow the user to input parameters, start the positioning process and adjust system settings.

[0073] The parallel computing method comprises:

[0074] Data parallelism: Divide the image data into multiple small blocks, and each processing unit processes one data block to improve the processing speed; for example, using the gaussianBlur function in the OpenCV library, Gaussian filtering can be performed on the image in parallel.

[0075] Task parallelism: Decompose the image processing task into multiple subtasks and execute these subtasks simultaneously on multiple processing units; this method is suitable for tasks with large computational complexity and small data volume, and can be implemented through multi-threading or distributed computing.

[0076] Spatial parallelism: Utilize the loop and recursive structures in image processing algorithms to parallelize tasks; for example, operations such as image transformation and image compression can be processed simultaneously on multiple processing units through spatial parallelism.

[0077] GPU acceleration: Use the parallel processing capabilities of the GPU to process images at high speed; the GPU has thousands of parallel processing cores and is suitable for large-scale parallel computing, such as the training and reasoning of deep learning models.

[0078] Optimization algorithm: Through data parallelism, task parallelism and space parallelism, combined with specific image processing tasks, efficient parallel algorithms can be designed; parallel computing algorithm optimization is the key to improving performance. Efficient parallel algorithms can be designed through data parallelism, task parallelism and space parallelism, combined with specific image processing tasks.

[0079] Code implementation: When implementing parallel computing, OpenCV is combined with CUDA to use CUDA's parallel computing capabilities to accelerate image processing tasks. For example, kernel functions written using CUDA can perform operations such as image filtering and edge detection in parallel.

[0080] Example

[0081] System component selection and configuration

[0082] Image acquisition and processing unit: Select a high-resolution industrial camera with an appropriate lens to ensure that the details of the workpiece can be captured. The configuration of the camera and lens should be adjusted according to the size, shape and processing environment of the workpiece. Set the camera's exposure time and gain to obtain a clear and not overexposed image.

[0083] Feature extraction and recognition unit: Integrates image processing algorithms, such as Canny edge detection and Hough transform, to extract artifact features. Uses deep learning frameworks (such as TensorFlow or PyTorch) to train convolutional neural network (CNN) models for feature classification and recognition.

[0084] 3D reconstruction positioning unit: uses structured light or laser triangulation technology to reconstruct the 3D model of the workpiece from the 2D image. Perform camera calibration to ensure the accuracy of the reconstructed 3D model.

[0085] Algorithm Optimization Unit: Use parallel computing frameworks (such as CUDA or OpenCL) to optimize image processing and feature recognition algorithms. Use GPUs to accelerate computationally intensive tasks such as image filtering and feature extraction.

[0086] User Interface: Design an intuitive user interface that displays a real-time image of the workpiece, positioning results, and machine status. Provide operating instructions that allow the user to enter parameters, start the positioning process, and adjust system settings.

[0087] System implementation steps

[0088] Step 1: Image acquisition and preprocessing

[0089] Use an industrial camera to capture the workpiece image, and perform grayscale, filter denoising and edge enhancement on the image.

[0090] Step 2: Feature extraction and recognition

[0091] Apply the Canny algorithm to detect edges in the image. Use the Hough transform to detect lines and circles in the image. Input the extracted features into the trained CNN model for feature classification and recognition.

[0092] Step 3: 3D reconstruction and positioning

[0093] Use structured light technology to obtain the depth information of the workpiece. Combine the depth information and the two-dimensional image to reconstruct the three-dimensional model of the workpiece. Determine the position of the workpiece in the machine tool coordinate system through camera calibration and stereo matching algorithm.

[0094] Step 4: Algorithm Optimization

[0095] Parallelize image processing algorithms to increase processing speed. Use GPU to accelerate the inference process of deep learning models.

[0096] Step 5: System Integration and Testing

[0097] Integrate image acquisition, feature extraction, 3D reconstruction and algorithm optimization units into the control system of CNC machine tools. Perform system tests to ensure that all components work together and the positioning results are accurate.

[0098] Step 6: User Interface Design

[0099] Develop a user-friendly interface that displays a real-time image of the workpiece and the positioning results. Implement functionality for the user to input parameters and start the positioning process.

[0100] 3. Specific implementation of parallel computing methods

[0101] Data parallelism: The image is divided into multiple small blocks, and each GPU thread processes a small block for filtering and edge detection.

[0102] Task parallelism: The feature extraction task is decomposed into multiple subtasks, such as edge detection, corner detection, and texture analysis, and executed in parallel on multiple CPU cores.

[0103] Spatial parallelism: Implementing image transformation and compression algorithms on the GPU, taking advantage of its multiple processing units to process different areas of the image simultaneously.

[0104] GPU acceleration: Use CUDA to write kernel functions and perform parallel processing on images, such as parallel execution of Gaussian filtering and Sobel edge detection.

[0105] Optimize algorithms: Design efficient parallel algorithms for specific image processing tasks, such as a parallel version of the Fast Fourier Transform (FFT) for frequency domain analysis.

[0106] Code implementation: Use OpenCV and CUDA to parallelize image processing algorithms. Use CUDA streams and events to manage parallel tasks on the GPU to ensure efficient memory access and processing.

[0107] Through the above embodiments, a high-precision CNC machine tool workpiece positioning system can be constructed, which can quickly and accurately position the workpiece and improve processing efficiency and precision.

[0108] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. All technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. A high-precision CNC machine tool workpiece positioning system based on machine vision, characterized in that: It includes an image acquisition and processing unit, a feature extraction and recognition unit, a three-dimensional reconstruction and positioning unit, an algorithm optimization unit and a user interface. The data output end of the image acquisition and processing unit is connected to the feature extraction and recognition unit, the data output end of the feature extraction and recognition unit is connected to the data input end of the three-dimensional reconstruction and positioning unit, the data output end of the three-dimensional reconstruction and positioning unit is connected to the algorithm optimization unit, and the control signal transmission end of the algorithm optimization unit is connected to the user interface.

2. The high-precision CNC machine tool workpiece positioning system based on machine vision according to claim 1, characterized in that: The image acquisition and processing unit acquires the image of the workpiece and captures sufficient details according to the characteristics of the workpiece and the processing environment. The image processing includes graying, filtering and denoising, and edge enhancement. The feature extraction and recognition unit extracts the edge, corner point, and texture features of the workpiece through an image processing algorithm. The three-dimensional reconstruction and positioning unit reconstructs the three-dimensional model of the workpiece from the two-dimensional image and determines its position in the machine tool coordinate system; the algorithm optimization unit optimizes the algorithm using a parallel computing method; the user interface displays the real-time image of the workpiece, the positioning result and the machine tool status, and provides necessary operating instructions.

3. A positioning method for a high-precision CNC machine tool workpiece positioning system based on machine vision as claimed in claim 2, characterized in that: The following steps are involved: S1: Image acquisition and processing unit S1.1: Select industrial cameras and lenses to ensure that they can adapt to the characteristics of the workpiece and the processing environment and capture sufficient details; S1.2: Set the camera exposure time and gain to obtain the best image quality; S1.3: Perform image preprocessing, including grayscale, filtering, denoising, and edge enhancement, to improve the accuracy of subsequent processing; S2: Feature extraction and recognition unit S2.1: Apply image processing algorithms, including Canny edge detection and Hough transform, to extract the edges, corners and texture features of the workpiece; S2.2: Use machine learning or deep learning models to classify and identify the extracted features to determine the type and characteristic features of the artifact; S3: 3D reconstruction positioning unit S3.1: Reconstruct a 3D model of the workpiece from a 2D image using stereo vision, structured light, or laser triangulation techniques; S3.2: Determine the precise position and orientation of the workpiece in the machine tool coordinate system through camera calibration and stereo matching algorithm; S4: Algorithm Optimization Unit S4.1: Optimize 3D reconstruction and feature recognition algorithms to improve processing speed and accuracy; S4.2: Use parallel computing methods, such as GPU acceleration, to process large amounts of image data and computing tasks; S5: User Interface S5.1: Design an intuitive user interface that displays real-time images of the workpiece, positioning results, and machine tool status; S5.2: Provides necessary operating instructions to allow the user to input parameters, start the positioning process and adjust system settings.

4. The positioning method of the high-precision CNC machine tool workpiece positioning system based on machine vision according to claim 3 is characterized in that: The parallel computing method comprises: Data parallelism: image data is divided into multiple small blocks, and each processing unit processes one data block to improve processing speed; Task parallelism: decompose the image processing task into multiple subtasks and execute these subtasks simultaneously on multiple processing units; Spatial parallelism: Use loops and recursive structures in image processing algorithms to parallelize tasks; GPU acceleration: Use the parallel processing capabilities of the GPU to process images at high speed; Optimization algorithm: Design efficient parallel algorithms by combining data parallelism, task parallelism and space parallelism with specific image processing tasks; Code implementation: When implementing parallel computing, OpenCV is combined with CUDA to utilize CUDA's parallel computing capabilities to accelerate image processing tasks.

Images