Denture blank material extraction method, device and computer equipment based on visual detection

Abstract

The present invention relates to a method, device, computer equipment and storage medium for taking out denture blanks based on visual detection. The image of the denture blank is acquired by a high-precision visual system, and then high-quality image information is obtained through the image. Then, key features such as edges, corners and textures are extracted from the high-quality image. The size and shape of the denture blank are measured based on these key features to obtain measurement results. Then, a preset deep learning algorithm is used to detect surface defects of the denture blank based on the key features to obtain detection results. Combining the measurement results and the detection results, a material taking decision is generated, and a material taking instruction is generated accordingly. Finally, the material taking instruction is sent to a preset material taking device to execute a material taking operation. The method can still extract stable features in the case of insufficient lighting and extreme angles, effectively deal with small defects and complex textures in the blank, and realize efficient and accurate detection and material taking operations of the denture blank.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method, device, computer equipment and storage medium for taking out denture blanks based on visual detection. Background Art

[0002] The selection and inspection process of blanks in denture production has always been an important link affecting product quality and production efficiency. Traditional manual inspection methods rely on the operator's experience and subjective judgment, which easily leads to inconsistent and incorrect inspection results. In addition, manual inspection is inefficient and difficult to meet the needs of modern efficient production. With the development of visual systems and automation technology, it has become possible to use image processing technology to realize automated inspection and material removal of blanks, which has brought new hope to denture production.

[0003] However, the existing automated inspection systems based on visual inspection still face some challenges in practical applications. For example, during the image acquisition process, the image quality varies greatly under different angles and lighting conditions, which may lead to unstable feature extraction. Although image preprocessing technology can improve image quality, it still has certain limitations when dealing with complex backgrounds and various interference factors. In addition, the accuracy and robustness of existing feature extraction and defect detection algorithms need to be improved when dealing with small defects and complex textures in blanks.

[0004] Finally, although automated robotic arms and equipment have shown certain advantages in material picking decisions and operations, how to ensure correct material picking after efficient and accurate detection is still an urgent problem to be solved. Overall, current research and applications need to be further optimized and improved in terms of image processing algorithms, detection accuracy, and the coordination of automated material picking systems to meet the high quality and high efficiency requirements of denture production. Summary of the invention

[0005] The main purpose of the present invention is to provide a method, device, computer equipment and storage medium for taking denture blanks based on visual inspection, so as to solve the problems of unstable feature extraction and non-robustness to small defects and complex textures in the blanks in the existing automated inspection system based on visual inspection.

[0006] To achieve the above-mentioned objectives, the present invention provides a method for picking up denture blanks based on visual inspection, comprising: using a high-precision visual system to collect images of denture blanks to obtain a set of denture blank images; performing feature extraction on the set of denture blank images to obtain key features of the denture blank, wherein the key features include at least edges, corners and textures; based on the key features of the denture blank, measuring and processing the size and shape of the denture blank to obtain measurement results of the denture blank; based on the key features of the denture blank, using a preset deep learning algorithm to detect surface defects of the denture blank to obtain detection results of the denture blank; based on the measurement results and detection results of the denture blank, generating a picking decision for the denture blank, and generating a picking instruction according to the picking decision; sending the picking instruction to a preset picking device, so that the picking device performs a picking operation based on the picking instruction.

[0007] Optionally, after using a high-precision visual system to capture images of denture blanks and obtain a set of denture blank images, the denture blank sampling method further includes: based on a preset salt and pepper noise removal algorithm, determining the noise pixels in the denture blank images, and replacing the noise pixels with the median of the field pixel values ​​of the noise pixels to obtain a denoised denture blank image; performing histogram equalization processing on the denoised denture blank images to redistribute the grayscale values ​​of the denture blank images to obtain the denture blank images with uniform grayscale distribution; combining linear transformation and gamma correction to perform brightness correction on the denture blank images with uniform grayscale distribution to obtain the final denture blank images, wherein the final denture blank images are used to execute the step of "extracting features from the denture blank image set to obtain key features of the denture blank".

[0008] Optionally, feature extraction is performed on the denture blank image set to obtain key features of the denture blank, including: performing graphic transformation processing on the images in the denture blank image set to obtain a transformed denture blank image set, wherein the graphic transformation processing at least includes: flipping, rotation and scaling processing; extracting edge features from the transformed denture blank image set based on a preset denture ingredient feature processing model to generate an edge feature map; and identifying corner features in the transformed denture blank image set based on the denture ingredient feature processing model to generate a corner feature map; and extracting texture features from the transformed denture blank image set using a Gabor filter in combination with the denture ingredient feature processing model to generate a texture feature map; feeding the edge feature map, corner feature map and texture feature map into a preset denture ingredient feature fusion model, and fusing the edge feature map, corner feature map and texture feature map using a pooling layer in the denture ingredient feature fusion model to obtain key features of the denture blank.

[0009] Optionally, the size and shape of the denture blank are measured and processed based on the key features of the denture blank to obtain measurement results of the denture blank, including: using a preset denture blank contour extraction model to extract the contour of the denture blank based on the edge feature map in the key features of the denture blank, and performing polygon fitting on the contour of the denture blank to obtain a first blank contour curve; and smoothing the first blank contour curve to obtain a second blank contour curve; using a preset denture blank focus matching model to perform corner point matching in the denture blank image set based on the focus feature map in the key features of the denture blank to obtain a matching result set of the denture blank image set; using three-dimensional reconstruction technology, combined with the second blank contour curve and the matching result set, to generate measurement results of the denture blank, wherein the measurement results at least include: the three-dimensional shape and size of the denture blank.

[0010] Optionally, based on the key features of the denture blank, a preset deep learning algorithm is used to detect surface defects of the denture blank to obtain detection results of the denture blank, including: using a preset deep learning algorithm to learn the potential connection between the key features and defects, and using a preset optimizer and loss function to train a deep learning model, so that the deep learning model focuses on the surface defects of the denture blank; in the process of training the deep learning model, monitoring the performance of the deep learning model on the denture blank image set, adjusting the deep learning model hyperparameters to prevent overfitting of the deep learning model hyperparameters, and obtaining a trained deep learning model; using the trained deep learning model to perform surface defect detection on the denture blank image, and outputting the detection result as a vector, wherein the vector is used to describe the type, location, and size of the defect, to obtain the detection result of the denture blank.

[0011] Optionally, based on the measurement results and detection results of the denture blank, a material picking decision for the denture blank is generated, and a material picking instruction is generated based on the material picking decision, and it also includes: judging whether the denture blank meets the preset measurement standard based on the measurement result of the denture blank; judging whether the denture blank meets the detection standard based on the detection result of the denture blank; generating a material picking decision for the denture blank based on the judgment result, wherein the material picking decision includes at least: a target denture blank to be picked up; determining the position coordinates of the target denture blank based on the denture blank image set; and generating a material picking instruction based on the position coordinates of the target denture blank, wherein the material picking instruction includes at least: a gripping force and the end movement trajectory of the robotic arm.

[0012] Optionally, the material picking instruction is sent to a preset material picking device, so that the material picking device performs a material picking operation based on the material picking instruction, and also includes: formatting the material picking instruction into JSON format to obtain a formatted material picking instruction; verifying whether the parameters of the formatted material picking instruction are within a preset range; setting a network communication interface for communication with the material picking device based on the type of material picking device; and sending the formatted material picking instruction to the material picking device based on the communication interface.

[0013] The present invention also provides a denture blank picking device based on visual detection, comprising: an acquisition unit, which uses a high-precision visual system to collect images of denture blanks to obtain a denture blank image set; a feature extraction unit, which performs feature extraction on the denture blank image set to obtain key features of the denture blank, wherein the key features at least include: edges, corners and textures; a measurement unit, which measures and processes the size and shape of the denture blank based on the key features of the denture blank to obtain a measurement result of the denture blank; a defect detection unit, which uses a preset deep learning algorithm to detect surface defects of the denture blank based on the key features of the denture blank to obtain a detection result of the denture blank; an instruction generation unit, which generates a picking decision for the denture blank based on the measurement result and detection result of the denture blank, and generates a picking instruction according to the picking decision; an instruction execution unit, which sends the picking instruction to a preset picking device to instruct the picking device to perform a picking operation based on the picking instruction.

[0014] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0015] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0016] The present invention provides a method, device, computer equipment and storage medium for taking out denture blanks based on visual inspection. Clear denture blank image data is collected by using a high-precision visual system. These data provide a basis for subsequent feature extraction, measurement and defect detection. Robust key features that are insensitive to lighting and angle are obtained from the denture blank image. These features include edge features, corner features and texture features. In order to effectively measure the size and shape of the denture blank and the surface defects of the denture blank under poor lighting conditions, the present invention adopts a preset deep learning algorithm. The algorithm can effectively identify and retain the most representative and predictive features even in insufficient light. The learned corner feature map, texture feature map and edge feature map are subjected to dimensionality reduction processing and fusion to obtain key features that are robust to lighting, angle and complex texture. These key features provide a reliable data basis for subsequent size measurement, defect detection and material taking decisions of denture blanks. Then, the preset 3D reconstruction technology is used to restore the 3D shape and size of the denture blank, and the preset deep learning algorithm is used to learn the potential connection between the key features and defects, and the deep learning model is constructed and trained to make the deep learning model focus on the surface defects of the denture blank. Finally, based on the 3D shape and size of the denture blank and the defect detection results, the denture blank picking decision is generated. According to the picking decision, the picking instruction is generated, and the picking instruction is sent to the preset picking device, so that the picking device executes the picking instruction and picking operation.

[0017] In general, the present invention obtains high-quality images through image processing algorithms, and uses deep learning algorithms to obtain stable features that are insensitive to lighting and angles from high-quality images, and effectively fuses these features. The fused features are used for shape restoration and surface defect detection of denture blanks, thereby improving the accuracy of shape and size restoration of denture blanks and the precision of defect detection.

[0018] In summary, the present invention can still extract stable features from the image under conditions of insufficient lighting and extreme angles, effectively deal with small defects and complex textures in the blank, and achieve efficient and accurate detection and material extraction operations for denture blanks. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the steps of a method for taking out denture blanks based on visual detection according to the present invention;

[0020] Figure 2 It is a structural block diagram of a denture blank material taking device based on visual detection of the present invention;

[0021] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0022] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0024] Reference Figure 1 The embodiment of the present invention provides a method for taking out denture blanks based on visual detection, comprising:

[0025] S1: Use a high-precision visual system to collect images of denture blanks to obtain a set of denture blank images;

[0026] S2: extracting features from the denture blank image set to obtain key features of the denture blank, wherein the key features include: edges, corners and textures;

[0027] S3: measuring and processing the size and shape of the denture blank based on the key features of the denture blank to obtain a measurement result of the denture blank;

[0028] S4: Based on the key features of the denture blank, a preset deep learning algorithm is used to detect surface defects of the denture blank to obtain a detection result of the denture blank;

[0029] S5: Based on the measurement result and the detection result of the denture blank, a material taking decision of the denture blank is generated, and a material taking instruction is generated according to the material taking decision;

[0030] S6: Send the material picking instruction to a preset material picking device, so that the material picking device performs a material picking operation based on the material picking instruction.

[0031] The present invention provides a denture blank material extraction method based on visual detection, which aims to solve the problem that image feature extraction is unstable under extreme angles and insufficient light, and the existing methods cannot effectively deal with subtle defects and complex textures. The present invention obtains robust key features that are insensitive to lighting and angles from the denture blank image, and these features include edge features, corner features and texture features. In order to effectively measure the size and shape of the denture blank and the surface defects of the denture blank under poor lighting conditions, the present invention adopts a preset deep learning algorithm. The algorithm can effectively identify and retain the most representative and predictive features even under insufficient light conditions. The learned corner feature map, texture feature map and edge feature map are subjected to dimensionality reduction processing and fusion to obtain key features that are robust to lighting, angles and complex textures. These key features provide a reliable data basis for subsequent size measurement, defect detection and material extraction decisions of denture blanks. Then, the preset 3D reconstruction technology is used to restore the 3D shape and size of the denture blank, and the preset deep learning algorithm is used to learn the potential connection between the key features and defects, and the deep learning model is constructed and trained to make the deep learning model focus on the surface defects of the denture blank. Finally, based on the 3D shape and size of the denture blank and the defect detection results, the denture blank picking decision is generated. According to the picking decision, the picking instruction is generated, and the picking instruction is sent to the preset picking device, so that the picking device executes the picking instruction and picking operation.

[0032] In general, the present invention obtains high-quality images through image processing algorithms, and uses deep learning algorithms to obtain stable features that are insensitive to lighting and angles from high-quality images, and effectively fuses these features. The fused features are used for shape restoration and surface defect detection of denture blanks, thereby improving the accuracy of shape and size restoration of denture blanks and the precision of defect detection.

[0033] In summary, the present invention can still extract stable features from the image under conditions of insufficient lighting and extreme angles, effectively deal with small defects and complex textures in the blank, and achieve efficient and accurate detection and material extraction operations for denture blanks.

[0034] In one example, after using a high-precision visual system to capture images of denture blanks and obtain a set of denture blank images, the denture blank extraction method also includes: based on a preset salt and pepper noise removal algorithm, determining the noise pixels in the denture blank image, and replacing the noise pixels with the median of the field pixel values ​​of the noise pixels to obtain a denoised denture blank image; performing histogram equalization processing on the denoised denture blank image to redistribute the grayscale values ​​of the denture blank image to obtain the denture blank image with uniform grayscale distribution; combining linear transformation and gamma correction to perform brightness correction on the denture blank image with uniform grayscale distribution to obtain the final denture blank image, wherein the final denture blank image is used to execute the step of "extracting features from the denture blank image set to obtain key features of the denture blank".

[0035] In this example, after obtaining the set of denture blank images, detailed image processing steps were performed to ensure high-quality image information for subsequent feature extraction and detection. First, based on the preset salt and pepper noise removal algorithm, the image is scanned pixel by pixel to determine the noise pixels therein. Salt and pepper noise usually appears as extreme black or white spots in the image, that is, the pixel value is 0 or 255. After the noise pixel is detected, it is replaced with the median of its neighborhood pixel value. The specific method is: define a domain window of a preset size, extract all pixel values ​​in the window, sort them, and then select the sorted middle value as the new value of the current noise pixel, and replace the original value of the noise pixel with the median value, thereby completing the noise removal.

[0036] After denoising, the processed image is histogram equalized to redistribute the grayscale value of the image to make the grayscale distribution more uniform. First, the grayscale histogram of the image is calculated to count the number of pixels at each grayscale level, and then the cumulative distribution function (CDF) is calculated and normalized to the range of 0-255. The grayscale value of each pixel is replaced with the corresponding normalized CDF value. After histogram equalization, the contrast and detail clarity of the image are improved. Finally, the image is brightness corrected by combining linear transformation and gamma correction. First, a linear transformation is performed to multiply the grayscale value of each pixel of the image by a preset coefficient. If the coefficient is greater than 1, the brightness increases; if the coefficient is less than 1, the brightness decreases, ensuring that the transformed grayscale value is within the range of 0-255. Next, gamma correction is used, and a preset gamma value is selected to adjust each pixel value of the image through the gamma transformation formula. Through gamma correction, the overall brightness and contrast of the image can be improved while retaining the image details. The above preprocessing steps ensure the high quality of the denture blank image and provide a reliable basis for subsequent feature extraction, dimension measurement and defect detection.

[0037] In summary, through the above image processing steps, it is ensured that noise is removed from the denture blank image set to obtain high-quality images, which are used in subsequent feature extraction, size measurement and defect detection.

[0038] In one example, feature extraction is performed on the denture blank image set to obtain key features of the denture blank, including: performing graphic transformation processing on the images in the denture blank image set to obtain a transformed denture blank image set, wherein the graphic transformation processing at least includes: flipping, rotation and scaling processing; extracting edge features from the transformed denture blank image set based on a preset denture ingredient feature processing model to generate an edge feature map; and identifying corner features in the transformed denture blank image set based on the denture ingredient feature processing model to generate a corner feature map; and extracting texture features from the transformed denture blank image set using a Gabor filter in combination with the denture ingredient feature processing model to generate a texture feature map; feeding the edge feature map, corner feature map and texture feature map into a preset denture ingredient feature fusion model, and fusing the edge feature map, corner feature map and texture feature map using a pooling layer in the denture ingredient feature fusion model to obtain key features of the denture blank.

[0039] In this example, in the process of extracting features from the denture blank image set and obtaining the key features of the denture blank, the images in the denture blank image set are first subjected to geometric transformations such as flipping, rotating and scaling to generate a transformed denture blank image set. These transformation operations are intended to enhance the diversity of the image set and improve the recognition ability of the subsequent deep learning algorithm for denture blanks of different postures and proportions. Next, the edge features of the transformed denture blank image set are extracted based on the preset deep learning algorithm to generate an edge feature map. The convolutional neural network (CNN) structure of the deep learning algorithm can effectively extract edge information from the image and generate an edge feature map. Then, the corner feature recognition of the transformed denture blank image set is performed again based on the preset deep learning algorithm to generate a corner feature map. Corner feature recognition is usually achieved by detecting corners and intersections in the image, and these feature points have good stability in the geometric transformation of the image. Next, the preset deep learning algorithm is combined with the Gabor filter to extract the texture features of the transformed denture blank image set and generate a texture feature map. Gabor filter is a linear filter used to detect local directional and frequency features in images. By combining with deep learning algorithms, it can better capture complex texture information in images.

[0040] Finally, the pooling layer in the preset deep learning algorithm is used to fuse the generated edge feature map, corner feature map, and texture feature map to obtain the key features of the denture blank that are robust to changes in illumination, angles, and complex textures. The pooling layer reduces the size of the feature map by dimensionality reduction while retaining important feature information, and finally generates the fused key feature map. These key feature maps provide a reliable data basis for subsequent size measurement, defect detection, and material removal decisions, ensuring the efficiency and accuracy of the entire denture blank removal process.

[0041] In summary, through the above steps, the key features of the denture blank are obtained. The corner feature recognition of the transformed image set is performed through the deep learning algorithm convolutional neural network to generate a corner feature map. These corners and intersections have good stability in geometric transformation, which helps to more accurately describe the geometric structure of the image. Combining the deep learning algorithm with the Gabor filter, the local directional features and frequency features in the image are extracted to generate a texture feature map, which captures the complex texture information in the image, making the feature extraction more comprehensive. The edge feature map, corner feature map and texture feature map are reduced in dimension and fused through the pooling layer to obtain a key feature map that is robust to illumination changes, angle changes and complex textures. The pooling layer not only reduces the size of the feature map, but also retains important feature information, which improves the robustness and applicability of the feature map. The key feature maps generated by these processing steps provide a reliable data basis for subsequent size measurement, defect detection and material removal decisions, ensuring the efficiency and accuracy of the entire denture blank material removal process, and improving the quality and efficiency of denture production. These operations improve the processing quality and feature stability of the denture blank image through systematic image preprocessing and feature extraction steps, ensuring the efficiency, accuracy and automation of the entire detection and material extraction process, which is also the key to the present invention.

[0042] In one example, the size and shape of the denture blank are measured and processed based on the key features of the denture blank to obtain the measurement results of the denture blank, including: using a preset denture blank contour extraction model to extract the contour of the denture blank based on the edge feature map in the key features of the denture blank, and performing polygon fitting on the contour of the denture blank to obtain a first blank contour curve; and smoothing the first blank contour curve to obtain a second blank contour curve; using a preset denture blank focus matching model to perform corner point matching in the denture blank image set based on the focus feature map in the key features of the denture blank to obtain a matching result set of the denture blank image set; using three-dimensional reconstruction technology, combined with the second blank contour curve and the matching result set, to generate the measurement results of the denture blank, wherein the measurement results at least include: the three-dimensional shape and size of the denture blank.

[0043] In this example, the size and shape of the denture blank are measured based on its key features. First, a preset contour extraction algorithm is used to extract the contour of the denture blank from the edge feature map, and a polygon fitting is performed on it to obtain the first blank contour curve. This step generates a preliminary polygonal contour by detecting the edge lines in the image to represent the approximate shape of the blank. Next, the preset curve fitting algorithm is used to smooth the first blank contour curve to obtain the second blank contour curve. The curve fitting algorithm removes noise and irregular points to make the contour curve smoother and more accurate, reflecting the actual shape of the blank. Then, a preset deep learning algorithm is used to input the extracted key features into the deep learning model to match the corner points in the denture blank image set to obtain a matching result set. The deep learning algorithm finds the corresponding matching points by comparing the feature points in the image to ensure that the key points in images at different angles and positions can accurately correspond.

[0044] For the matching result set, combined with the camera calibration parameters, the preset 3D reconstruction technology is used to restore the 3D shape and size of the denture blank. The camera calibration parameters provide geometric information when the image is taken. The 3D reconstruction technology uses this information to reconstruct the 3D model and size of the blank by calculating the position of the points in the image in 3D space. Finally, the measurement results of the denture blank will be used to execute the step of generating the material removal decision. Based on the measurement results and detection results, the system will generate a material removal decision, and generate specific material removal instructions according to the material removal decision to guide the automated equipment to perform precise material removal operations. In this way, the entire process achieves efficient and accurate measurement and automatic material removal of denture blanks through precise image processing and deep learning algorithms.

[0045] In summary, this example obtains the measurement results of the denture blank through a series of detailed steps. These measurement results are used to generate material removal decisions and instructions to ensure that the automated equipment can accurately remove the material. The entire process improves the efficiency and accuracy of inspection and material removal, reduces manual intervention, improves production quality and efficiency, and meets the needs of modern denture production for high-precision and efficient automation.

[0046] In one example, based on the key features of the denture blank, a preset deep learning algorithm is used to detect surface defects of the denture blank to obtain a detection result of the denture blank, including: using a preset deep learning algorithm to learn the potential connection between the key features and defects, and using a preset optimizer and loss function to train a deep learning model so that the deep learning model focuses on the surface defects of the denture blank; in the process of training the deep learning model, monitoring the performance of the deep learning model on the denture blank image set, adjusting the deep learning model hyperparameters to prevent overfitting of the deep learning model hyperparameters, and obtaining a trained deep learning model; using the trained deep learning model to perform surface defect detection on the denture blank image, and outputting the detection result as a vector, wherein the vector is used to describe the type, location, and size of the defect, to obtain a detection result of the denture blank.

[0047] In this example, the process of surface defect detection based on the key features of denture blanks first uses a preset deep learning algorithm to learn the potential connection between key features and defects, and uses a preset optimizer and loss function to train the deep learning model so that the model can focus on the surface defects of denture blanks. In the process of training the deep learning model, the performance of the model on the denture blank image set is continuously monitored, and the hyperparameters of the model are adjusted to prevent the model from overfitting, thereby ensuring the generalization ability of the model. After repeated training and optimization, a trained deep learning model is finally obtained. Using the trained deep learning model, the surface defect detection of denture blank images is performed. The model analyzes each image and identifies the surface defects in it. The detection results are output in the form of vectors, which describe the type, location, size and other information of the defects in detail. These detection results provide important data support for subsequent steps. Based on the detection results and the previously measured dimensional data, the system will generate a material picking decision, and generate specific material picking instructions based on the material picking decision to guide the automated equipment to perform precise material picking operations. In this way, the entire process achieves efficient and accurate surface defect detection and automatic material removal of denture blanks, improving the quality and efficiency of denture production and ensuring product reliability and consistency.

[0048] In short, the key features and surface defects of denture blanks are accurately detected through the preset deep learning algorithm, which improves the quality and efficiency of denture production. The deep learning model can learn the potential connection between key features and defects, and is effectively trained through optimizers and loss functions, so that the model can focus on detecting surface defects and accurately identify even small defects.

[0049] In one example, based on the measurement results and detection results of the denture blank, a material picking decision for the denture blank is generated, and a material picking instruction is generated according to the material picking decision, and it also includes: judging whether the denture blank meets the preset measurement standard according to the measurement result of the denture blank; judging whether the denture blank meets the detection standard according to the detection result of the denture blank; generating a material picking decision for the denture blank according to the judgment result, wherein the material picking decision includes at least: a target denture blank to be picked up; determining the position coordinates of the target denture blank based on the denture blank image set; and generating a material picking instruction based on the position coordinates of the target denture blank, wherein the material picking instruction includes at least: a gripping force and a movement trajectory of the end of the robotic arm.

[0050] In this example, a material picking decision and a material picking instruction are generated based on the measurement and inspection results of the denture blank. The measurement results are used to determine whether the denture blank meets the size and shape standards, in order to ensure that the basic physical properties of the blank meet the requirements. Next, the inspection results are used to determine whether there are defects on the surface of the denture blank and whether these defects are within an acceptable range to determine whether the denture blank is usable. Based on the measurement and inspection results, as well as the actual production needs and quality control standards, a decision is made as to whether the denture blank should be picked up. Specifically, if the blank meets the standards in size and shape and has no serious defects, the material picking operation can be performed; if the blank does not meet the standards or has unacceptable defects, the material picking operation will not be performed. Based on these decisions, detailed material picking instructions are generated. The material picking instructions include the position coordinates of the blank on the workbench or storage area, as well as the gripping force and moving speed that need to be considered during the material picking process. These parameters ensure that the picking equipment can accurately locate and grab the denture blank to avoid damage to the blank during the material picking process. Finally, the generated material picking instruction is sent to the preset material picking device, so that the material picking device performs the material picking operation based on the material picking instruction.

[0051] In short, accurate measurement and inspection results are used to generate material picking decisions, ensuring that the denture blanks meet the standards, reducing defects, and improving material picking accuracy and production efficiency. Automated material picking instructions optimize gripping force and moving speed, reduce damage, and improve overall production quality. The entire material picking process is efficient, accurate, and automated, improving production efficiency and product quality.

[0052] In one example, the material picking instruction is sent to a preset material picking device, so that the material picking device performs a material picking operation based on the material picking instruction, and also includes: formatting the material picking instruction into JSON format to obtain a formatted material picking instruction; verifying whether the parameters of the formatted material picking instruction are within a preset range; setting a network communication interface for communication with the material picking device based on the type of material picking device; and sending the formatted material picking instruction to the material picking device based on the communication interface.

[0053] In this example, the material picking instruction is sent to the preset material picking device and the material picking operation is performed. The material picking instruction needs to be formatted into JSON format to obtain the formatted material picking instruction. JSON format is a lightweight data exchange format that is easy for machine parsing and generation, ensuring that the structure of the instruction is clear and easy to transmit. Next, the formatted material picking instruction is verified to check whether the parameters are within a reasonable range, such as whether the coordinates, gripping force, and moving speed parameters meet the operating specifications of the material picking device to ensure that the execution of the instruction will not cause equipment failure or operational errors. According to the type of material picking device, the corresponding network communication interface is set to achieve communication between the material picking device and the control system. Different types of material picking devices may use different communication protocols and interfaces, so they need to be configured according to the specific device. Finally, based on the set communication interface, the formatted and verified material picking instruction is sent to the material picking device. In this way, the material picking device can receive accurate and formatted material picking instructions and perform specific material picking operations based on these instructions.

[0054] In summary, the entire process ensures the correct transmission and execution of material picking instructions, improves the reliability and automation level of the material picking process, and thus improves production efficiency and product quality.

[0055] Reference Figure 2 The embodiment of the present invention provides a denture blank material taking device based on visual detection, comprising:

[0056] An acquisition unit 1 uses a high-precision visual system to collect images of the denture blank to obtain a set of denture blank images;

[0057] A feature extraction unit 2 performs feature extraction on the denture blank image set to obtain key features of the denture blank, wherein the key features include at least: edges, corners and textures;

[0058] A measuring unit 3 measures the size and shape of the denture blank based on the key features of the denture blank to obtain a measurement result of the denture blank;

[0059] A defect detection unit 4 detects surface defects of the denture blank using a preset deep learning algorithm based on key features of the denture blank to obtain a detection result of the denture blank;

[0060] An instruction generating unit 5 generates a material taking decision of the denture blank based on the measurement result and the detection result of the denture blank, and generates a material taking instruction according to the material taking decision;

[0061] The instruction execution unit 6 sends the material fetching instruction to a preset material fetching device, so as to enable the material fetching device to execute a material fetching operation based on the material fetching instruction.

[0062] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0063] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0064] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0065] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0066] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0067] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0068] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for taking out denture blanks based on visual inspection, characterized in that: The following steps are involved: A high-precision visual system is used to collect images of denture blanks to obtain a set of denture blank images; Extracting features from the denture blank image set to obtain key features of the denture blank, wherein the key features include at least: edges, corners and textures; Based on the key features of the denture blank, the size and shape of the denture blank are measured to obtain a measurement result of the denture blank; Based on the key features of the denture blank, a preset deep learning algorithm is used to detect surface defects of the denture blank to obtain a detection result of the denture blank; Based on the measurement results and the detection results of the denture blank, a material taking decision of the denture blank is generated, and a material taking instruction is generated according to the material taking decision; Sending the material fetching instruction to a preset material fetching device, so that the material fetching device performs a material fetching operation based on the material fetching instruction; Among them, based on the key features of the denture blank, a preset deep learning algorithm is used to detect the surface defects of the denture blank to obtain the detection result of the denture blank, including: using a preset deep learning algorithm to learn the potential connection between the key features and defects, and using a preset optimizer and loss function to train a deep learning model, so that the deep learning model pays attention to the surface defects of the denture blank; in the process of training the deep learning model, the performance of the deep learning model on the denture blank image set is monitored, and the deep learning model hyperparameters are adjusted to prevent the deep learning model hyperparameters from overfitting, so as to obtain a trained deep learning model; using the trained deep learning model to perform surface defect detection on the denture blank image, and outputting the detection result as a vector, wherein the vector is used to describe the type, position, and size of the defect, so as to obtain the detection result of the denture blank; Wherein, based on the measurement result and the detection result of the denture blank, a material picking decision of the denture blank is generated, and a material picking instruction is generated according to the material picking decision, and it also includes: judging whether the denture blank meets the preset measurement standard according to the measurement result of the denture blank; judging whether the denture blank meets the detection standard according to the detection result of the denture blank; generating a material picking decision of the denture blank according to the judgment result, wherein the material picking decision at least includes: a target denture blank to be picked up; determining the position coordinates of the target denture blank based on the denture blank image set; and generating a material picking instruction based on the position coordinates of the target denture blank, wherein the material picking instruction at least includes: a gripping force and a movement trajectory of the end of the robot arm; Among them, feature extraction is performed on the denture blank image set to obtain the key features of the denture blank, including: performing graphic transformation processing on the images in the denture blank image set to obtain a transformed denture blank image set, wherein the graphic transformation processing at least includes: flipping, rotation and scaling processing; extracting edge features in the transformed denture blank image set based on a preset denture ingredient feature processing model to generate an edge feature map; identifying corner features in the transformed denture blank image set based on the denture ingredient feature processing model to generate a corner feature map; using a Gabor filter in combination with the denture ingredient feature processing model to extract texture features in the transformed denture blank image set to generate a texture feature map; feeding the edge feature map, corner feature map and texture feature map into a preset denture ingredient feature fusion model, and using the pooling layer in the denture ingredient feature fusion model to fuse the edge feature map, corner feature map and texture feature map to obtain the key features of the denture blank.

2. The method for taking out denture blanks according to claim 1, characterized in that: After the image of the denture blank is collected by using a high-precision visual system to obtain a set of denture blank images, the denture blank material collection method further includes: Based on a preset salt and pepper noise removal algorithm, the noise pixels in the denture blank image are determined, and the noise pixels are replaced with the median of the field pixel values ​​of the noise pixels to obtain a denoised denture blank image; Performing histogram equalization processing on the denoised denture blank image to redistribute the grayscale value of the denture blank image to obtain the denture blank image with uniform grayscale distribution; Combined with linear transformation and gamma correction, brightness correction is performed on the denture blank image with uniform grayscale distribution to obtain the final denture blank image, wherein the final denture blank image is used to execute the step of "extracting features from the denture blank image set to obtain key features of the denture blank".

3. The method for taking out denture blanks according to claim 1, characterized in that: The size and shape of the denture blank are measured based on the key features of the denture blank to obtain the measurement result of the denture blank, including: Using a preset denture blank contour extraction model, extracting the contour of the denture blank based on an edge feature map in a key feature of the denture blank, and performing polygon fitting on the contour of the denture blank to obtain a first blank contour curve; and performing smoothing processing on the first blank contour curve to obtain a second blank contour curve; Using a preset denture blank focus matching model, based on a focus feature map in a key feature of the denture blank, performing corner point matching in the denture blank image set, and obtaining a matching result set of the denture blank image set; The three-dimensional reconstruction technology is used to combine the second blank contour curve and the matching result set to generate the measurement result of the denture blank, wherein the measurement result at least includes: the three-dimensional shape and size of the denture blank.

4. The method for taking out denture blanks according to claim 1, characterized in that: The method further includes sending the material fetching instruction to a preset material fetching device, so that the material fetching device performs a material fetching operation based on the material fetching instruction: Formatting the material fetching instruction into JSON format to obtain a formatted material fetching instruction; Verifying whether the parameters of the formatted material fetching instruction are within a preset range; According to the type of the reclaiming equipment, a network communication interface for communication with the reclaiming equipment is provided; The formatted material fetching instruction is sent to the material fetching equipment based on the communication interface.

5. A denture blank material taking device based on visual detection, characterized in that: include: An acquisition unit uses a high-precision visual system to collect images of denture blanks to obtain a set of denture blank images; A feature extraction unit is used to extract features from the denture blank image set to obtain key features of the denture blank, wherein the key features include at least: edges, corners and textures; A measuring unit, which measures the size and shape of the denture blank based on the key features of the denture blank to obtain a measurement result of the denture blank; A defect detection unit, based on the key features of the denture blank, uses a preset deep learning algorithm to detect surface defects of the denture blank to obtain a detection result of the denture blank; An instruction generating unit generates a material taking decision of the denture blank based on the measurement result and the detection result of the denture blank, and generates a material taking instruction according to the material taking decision; An instruction execution unit sends the material fetching instruction to a preset material fetching device, so that the material fetching device executes a material fetching operation based on the material fetching instruction; Among them, based on the key features of the denture blank, a preset deep learning algorithm is used to detect the surface defects of the denture blank to obtain the detection result of the denture blank, including: using a preset deep learning algorithm to learn the potential connection between the key features and defects, and using a preset optimizer and loss function to train a deep learning model, so that the deep learning model pays attention to the surface defects of the denture blank; in the process of training the deep learning model, the performance of the deep learning model on the denture blank image set is monitored, and the deep learning model hyperparameters are adjusted to prevent the deep learning model hyperparameters from overfitting, so as to obtain a trained deep learning model; using the trained deep learning model to perform surface defect detection on the denture blank image, and outputting the detection result as a vector, wherein the vector is used to describe the type, position, and size of the defect, so as to obtain the detection result of the denture blank; Wherein, based on the measurement result and the detection result of the denture blank, a material picking decision of the denture blank is generated, and a material picking instruction is generated according to the material picking decision, and it also includes: judging whether the denture blank meets the preset measurement standard according to the measurement result of the denture blank; judging whether the denture blank meets the detection standard according to the detection result of the denture blank; generating a material picking decision of the denture blank according to the judgment result, wherein the material picking decision at least includes: a target denture blank to be picked up; determining the position coordinates of the target denture blank based on the denture blank image set; and generating a material picking instruction based on the position coordinates of the target denture blank, wherein the material picking instruction at least includes: a gripping force and a movement trajectory of the end of the robot arm; Among them, feature extraction is performed on the denture blank image set to obtain the key features of the denture blank, including: performing graphic transformation processing on the images in the denture blank image set to obtain a transformed denture blank image set, wherein the graphic transformation processing at least includes: flipping, rotation and scaling processing; extracting edge features in the transformed denture blank image set based on a preset denture ingredient feature processing model to generate an edge feature map; identifying corner features in the transformed denture blank image set based on the denture ingredient feature processing model to generate a corner feature map; using a Gabor filter in combination with the denture ingredient feature processing model to extract texture features in the transformed denture blank image set to generate a texture feature map; feeding the edge feature map, corner feature map and texture feature map into a preset denture ingredient feature fusion model, and using the pooling layer in the denture ingredient feature fusion model to fuse the edge feature map, corner feature map and texture feature map to obtain the key features of the denture blank.

6. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images