Titanium alloy product defect detection method and device based on photoacoustic image fusion

Abstract

The invention provides a titanium alloy product defect detection method and device based on photoacoustic image fusion. The technical problem that internal and surface defects of a titanium alloy product cannot be accurately recognized in the prior art is solved. Comprising the following steps: performing coordinate mapping on a surface defect detection optical image of the titanium alloy product and internal defect detection acoustic signal information of the titanium alloy product; obtaining a surface defect detection optical image of the titanium alloy product in the photoacoustic detection information fusion background field plane rectangular coordinate system C and a correct position of an internal defect detection acoustic signal of the titanium alloy product, and then making a defect schematic diagram to obtain a photoacoustic image fusion defect detection result of the titanium alloy product; the inference formula of coordinate mapping is # imgabs0 #; # imgabs 1 #; # imgabs < 2 >; and 3 # imgabs3 #. The method can be widely applied to the technical field of defect detection.

Description

Technical Field

[0001] The present application belongs to the technical field of defect detection, and more specifically, to a method and device for titanium alloy product defect detection based on photoacoustic image fusion. Background Art

[0002] Titanium alloy additive manufacturing technology, also known as titanium alloy 3D printing, has the advantages of rapid net forming of complex components and high material utilization. It is one of the most representative and most concerned disruptive technologies in the manufacturing industry, and plays an important role in aerospace, medical care, equipment and instruments and other fields.

[0003] Titanium alloy products manufactured by titanium alloy additive manufacturing technology will be affected by external environments such as high temperature, high pressure, high load, and strong electromagnetic fields during production and service, which may easily cause defects on the surface or inside of titanium alloy products. If the damage cannot be identified in time and effective measures cannot be taken, the harsh conditions will cause titanium alloy products to fail during use, which may easily lead to major safety accidents, cause irreparable economic losses, and may even endanger human life.

[0004] The current titanium alloy product defect detection method can only identify surface defects or sub-surface defects separately, and cannot obtain a comprehensive and accurate defect detection result, which makes the quality of titanium alloy products unguaranteed, resulting in great safety problems for products using titanium alloy products, thus causing major safety accidents. Therefore, it is urgent to provide a titanium alloy product defect detection method and device based on photoacoustic image fusion to solve the above technical problems. Summary of the invention

[0005] The purpose of the embodiments of the present application is to provide a titanium alloy product defect detection method and device based on photoacoustic image fusion to solve the existing technical problem that the internal and surface defects of titanium alloy products cannot be accurately identified.

[0006] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application provides a titanium alloy product defect detection method based on photoacoustic image fusion, comprising the following steps: Coordinate mapping is performed on the optical image of the surface defect detection of the titanium alloy product and the acoustic signal information of the internal defect detection of the titanium alloy product to obtain the correct location of the optical image of the surface defect detection of the titanium alloy product and the acoustic signal information of the internal defect detection of the titanium alloy product in the rectangular coordinate system C of the photoacoustic detection information fusion background field plane, and then a defect schematic diagram is made to obtain the photoacoustic image fusion defect detection result of the titanium alloy product; The inference formula for coordinate mapping is: ; ; ; ; In the formula, and are the horizontal and vertical coordinates of the center point of the optical image of the surface defect detection of titanium alloy products in the plane rectangular coordinate system A, and are the overall length and width of the titanium alloy product reflected in the plane rectangular coordinate system A, and are the horizontal and vertical coordinates of the center point of the internal defect detection acoustic signal information of the titanium alloy product in the plane rectangular coordinate system B, and They are the overall length and width of the titanium alloy product reflected in the plane rectangular coordinate system B. and They are the actual length and width of the titanium alloy product. and are the horizontal and vertical coordinates of the center point of the optical image of the surface defect detection of the titanium alloy product in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion, and They are respectively the horizontal and vertical coordinates of the center point of the acoustic signal information of internal defect detection of titanium alloy products in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion.

[0007] Preferably, the specific steps of obtaining the acoustic signal information of the internal defect detection of the titanium alloy product are: obtaining the original acoustic signal of the titanium alloy product and performing preprocessing to obtain the processed acoustic signal, running the imaging algorithm to obtain the internal defect detection ultrasonic imaging, and then performing post-processing to obtain the acoustic signal information of the internal defect detection of the titanium alloy product; The ultrasonic imaging of internal defect detection is a burial depth map of the defect position. The burial depth of the non-defective position is 0, and the formula for the burial depth of the defective position is: ; In the formula, is the propagation speed of ultrasonic waves in titanium alloy products, It is the moment when the peak of the echo on the surface of the titanium alloy product is located. It is the moment when the peak of the internal defect echo of the titanium alloy product is located. is the buried depth of internal defects.

[0008] Preferably, the specific steps of obtaining the optical image for surface defect detection of the titanium alloy product are: obtaining the original optical image of the titanium alloy product, performing preprocessing to obtain a high-quality image to be detected, and then inputting it into the improved YOLOv10n detection model to obtain the optical image for surface defect detection of the titanium alloy product; The preprocessing of the original optical image of titanium alloy products includes image histogram equalization, image median filtering and image sharpening; The calculation formula of the mapping function in image histogram equalization is: ; ; In the formula, Indicates the gray value is The probability of a pixel appearing in the original optical image of the titanium alloy product, Indicates that the gray value does not exceed The cumulative probability of the pixel is Represents the grayscale value after mapping, Indicates the range of grayscale values.

[0009] Preferably, the improved YOLOv10n detection model includes a lightweight C2f_Dual module; The lightweight C2f_Dual module includes and Convolution kernel and convolution filter using group convolution technique.

[0010] Preferably, the preprocessing of the original acoustic signal of the titanium alloy product includes signal envelope extraction and signal smoothing filtering; The formula for signal smoothing filtering is: ; In the formula, is the original sound signal of titanium alloy products. is the smoothing filter window size, an odd number, is the acoustic signal after smoothing filtering, j and i are the labels of the signal sequence.

[0011] Preferably, the analytical signal formula constructed by signal envelope extraction is: ; In the formula, is the original sound signal of titanium alloy products. for The acoustic signal obtained after Hilbert transform is Is an imaginary unit.

[0012] Preferably, the specific formula for obtaining the original optical image of the titanium alloy product is: ; ; ; In the formula, is the horizontal field of view, is the sensor width, is the focal length of the camera, is the actual field of view width, is the object distance, is the image width, To obtain the actual pixel size of the original optical image of the titanium alloy product.

[0013] Preferably, the calculation formula for image median filtering is: ; In the formula, is the original optical image after median filtering. The gray value of the position, The original optical image is The gray value of the position, A two-dimensional template.

[0014] Preferably, the formula for image sharpening is: ; In the formula, The original optical image of the titanium alloy product after sharpening The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is Grayscale value of the position.

[0015] A second aspect of the embodiment of the present application provides a titanium alloy product defect detection device based on photoacoustic image fusion, comprising: a camera, a bracket, a storage table, an ultrasonic board, a single crystal straight probe, a gantry structure, a programmable motion controller, a motor, an encoder, a host computer and a water tank; a camera is arranged on the top of the bracket, and a storage table is arranged on the bottom of the bracket; the camera and the ultrasonic board are respectively connected to the host computer, the ultrasonic board, the water tank, the encoder, the programmable motion controller and the motor are respectively connected to the gantry structure, the ultrasonic board is connected to the single crystal straight probe, and the motor and the encoder are relatively connected through the gantry structure; A camera for obtaining original optical images of titanium alloy products; Programmable motion controller for scanning motion on a gantry structure using motor-controlled encoders and single-element straight probes; Ultrasonic board, used to send excitation pulses and store the original sound signal of titanium alloy products; The single crystal straight probe is used to convert the excitation pulse into ultrasonic waves; receive the ultrasonic waves that pass through the interface between the water in the water tank and the titanium alloy product placed in the water tank, that is, the original acoustic signal of the titanium alloy product; The host computer is used to perform photoacoustic image fusion defect detection on titanium alloy products by utilizing the original optical image of the titanium alloy product and the original acoustic signal of the titanium alloy product.

[0016] The titanium alloy product defect detection method and device based on photoacoustic image fusion provided by the present invention have the following beneficial effects: The present invention uses a modal signal combined with photoacoustics to detect internal defects and surface defects of titanium alloy products, uses a coordinate mapping method and makes a specific schematic diagram to mark the defect positions inside and on the surface of the defects, and such a detection method can simultaneously detect high-risk areas with internal and surface defects, and the detection range is more comprehensive, and the obtained photoacoustic image fusion defect detection results of titanium alloy products are more reliable. The present invention uses an ultrasonic board, a single crystal straight probe, a programmable motion controller, a motor and a water tank to cooperate with each other to collect the original sound signal of the titanium alloy product, wherein the ultrasonic wave generated by the vibration passes through the interface between the water in the water tank and the titanium alloy product placed in the water tank, reaches the defect area, and is reflected to the single crystal straight probe. The original sound signals of various positions of the titanium alloy product are collected in sequence by this self-generated and self-received method, and such a collection method can collect sound signals in all directions to the maximum extent, and the collection range is wider and the detection results are more accurate.

[0017] The present invention uses image histogram equalization, image median filtering, image sharpening and an improved YOLOv10n detection model to detect surface defects, which can make the image clearer, the algorithm lighter, and the obtained detection results more accurate. Envelope extraction, smoothing filtering and imaging algorithms are used to detect the size, type, depth and center position coordinates of internal defects, which not only reduces the interference of cluttered noise, but also improves the quality of acoustic signals, provides a guarantee for subsequent coordinate mapping, and ensures the integrity of the titanium alloy product defect detection method based on photoacoustic image fusion of the present invention, so that the detection range is more comprehensive and the results obtained are more accurate.

[0018] In summary, the titanium alloy product defect detection method and device based on photoacoustic image fusion provided by the present invention can detect defect locations more comprehensively and produce more accurate detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0020] Figure 1 A schematic diagram of the overall process of a titanium alloy product defect detection method based on photoacoustic image fusion provided in one embodiment of the present application; Figure 2 A schematic diagram of the overall process of an improved YOLOv10n model provided in an embodiment of the present application; Figure 3 A schematic structural diagram of a titanium alloy product defect detection device based on photoacoustic image fusion provided in one embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application more clearly understood, the present application 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 application and are not used to limit the present application.

[0022] The first aspect of this embodiment provides a titanium alloy product defect detection method based on photoacoustic image fusion, which obtains the original optical image of the titanium alloy product through a camera, performs preprocessing, and then inputs it into the improved YOLOv10n detection model to obtain the surface defect detection optical image of the titanium alloy product; obtains the original acoustic signal of the titanium alloy product by scanning, performs preprocessing and imaging algorithm, and then performs post-processing to obtain the internal defect detection acoustic signal information of the titanium alloy product; coordinate mapping the surface defect detection optical image of the titanium alloy product and the internal defect detection acoustic signal information of the titanium alloy product is performed to obtain the correct placement of the surface defect detection optical image of the titanium alloy product and the internal defect detection acoustic signal information of the titanium alloy product in the rectangular coordinate system C of the photoacoustic detection information fusion background field plane, and then makes a defect schematic diagram to obtain the photoacoustic image fusion defect detection result of the titanium alloy product. This modal signal that combines optical signals and acoustic signals can obtain more accurate photoacoustic image fusion defect detection results of titanium alloy products. The order of the above steps is not required and can be adjusted according to the specific implementation process.

[0023] See also Figure 1 , is a schematic diagram of the overall process of a titanium alloy product defect detection method based on photoacoustic image fusion provided in an embodiment of the present application, including: S1: Obtain the original optical image of the titanium alloy product, perform preprocessing to obtain a high-quality image to be inspected, and then input it into the improved YOLOv10n detection model to obtain an optical image for surface defect detection of the titanium alloy product.

[0024] The specific steps include: 1. Obtain the original optical image of titanium alloy products: According to the size of the titanium alloy product to be inspected, the object distance and focal length of the camera are reasonably adjusted, and favorable lighting conditions are set. The camera then takes pictures to obtain the original optical image of the titanium alloy product. Specifically, the parameters of the camera sensor size, focal length, object distance and output image size are obtained. From these parameters, the actual pixel size of the original optical image of the titanium alloy product can be derived. The specific formula is as follows: ; ; ; In the formula, is the horizontal field of view, is the sensor width, is the focal length of the camera, is the actual field of view width, is the object distance, is the image width, To obtain the actual pixel size of the original optical image of the titanium alloy product.

[0025] If the required parameters of the camera are incomplete, a calibration ruler can be used to estimate the actual pixel size of the original optical image of the titanium alloy product. The calculation formula is: ; in is the actual length of the calibration area, is the pixel width of the calibration area, The calibration estimation result of the actual pixel size of the original optical image collected for titanium alloy products.

[0026] 2. The preprocessing of the original optical image of the titanium alloy product includes image histogram equalization, image median filtering and image sharpening. The preprocessing of the original optical image of the titanium alloy product obtains a high-quality image to be inspected with good contrast and low noise. Among them, the order of the steps in the preprocessing of the original optical image of the titanium alloy product is not required and can be adjusted according to the specific implementation process.

[0027] (1) Image histogram equalization is achieved by redistributing the grayscale values ​​of the image. Based on the histogram of the image, it redistributes the grayscale values ​​within the entire grayscale range so that the areas that were originally too dark or too bright will be properly adjusted, thereby improving the quality of the image.

[0028] In an optional embodiment, a histogram of the original optical image is firstly counted, and then the frequency of the grayscale value of each original optical image is obtained from the histogram of the original optical image, and the frequency of the grayscale value of each original optical image is further analyzed to obtain a cumulative distribution function, and finally a new mapping of the grayscale value of the original optical image is completed, that is, the grayscale of the original optical image is redistributed.

[0029] Among them, the calculation formula of the mapping function in image histogram equalization is: ; ; In the formula, Indicates the gray value is The probability of a pixel appearing in the original optical image of the titanium alloy product, Indicates that the gray value does not exceed The cumulative probability of the pixel is Represents the grayscale value after mapping, Indicates the range of grayscale values.

[0030] (2) Image median filtering is a nonlinear filtering technology based on sorting statistics theory. It can filter out isolated noise points and improve image quality.

[0031] Specifically, the calculation formula for image median filtering is: ; In the formula, Indicates the original optical image after median filtering. The gray value of the position, Represents the original optical image in The gray value of the position, For a two-dimensional template, you can choose , and .

[0032] (3) Image sharpening specifically refers to the Laplace gradient sharpening of an image. Image sharpening can highlight areas where the grayscale values ​​between pixels change dramatically, helping to enhance the edges and details in the image, thereby improving image quality.

[0033] Specifically, the standard Laplace operator formula for image Laplace gradient sharpening is: ; In the formula, The original optical image of the titanium alloy product after sharpening is shown in The gray value of the position, The original optical image of the titanium alloy product is shown in The gray value of the position, The original optical image of the titanium alloy product is shown in The gray value of the position, The original optical image of the titanium alloy product is shown in The gray value of the position, The original optical image of the titanium alloy product is shown in The gray value of the position, The original optical image of the titanium alloy product is shown in Grayscale value of the position.

[0034] 3. The high-quality image to be inspected is input into the improved YOLOv10n detection model. After the convolutional neural network processes the high-quality image to be inspected layer by layer, the optical image for surface defect detection of titanium alloy products can be obtained accurately and efficiently.

[0035] For details, please refer to Figure 2 The high-quality images to be inspected are input into the improved YOLOv10n detection model for further processing. The improved YOLOv10n detection model is divided into three parts: Backbone layer, Neck layer and Head layer. The high-quality images to be inspected are processed layer by layer to realize the positioning and identification of surface defect targets of titanium alloy products.

[0036] (1) The Backbone layer is the part of the improved YOLOv10n detection model used to extract high-quality features of the image to be detected.

[0037] ① The high-quality image to be detected passes through two layers of kernel size The convolution layer (Conv) outputs the convolutional high-quality image to be detected to the C2f_Dual (DualConv lightweight dual convolution design) module.

[0038] ② After the C2f_Dual module includes 3×3 and 1×1 convolution kernels, and uses group convolution technology to efficiently arrange convolution filters, it is used to extract image features of different levels and abstraction levels in high-quality images to be detected, which not only reduces the computational cost and the number of parameters, but is also lighter than the original C2f module.

[0039] ③ The image features extracted by the C2f_Dual module are processed by a layer with a kernel size of The convolutional layer (Conv).

[0040] ④ Output the convolutional image features to the C2f_Dual module, and output two branches after processing. One branch waits for the large-scale image features of the Concat (feature concatenation) module in the Neck layer; the other branch passes through the SCDown (lightweight downsampling method) module to reduce the computational cost and number of parameters while ensuring the detection accuracy of the improved YOLOv10n detection model.

[0041] ⑤After sampling by the SCDown module, the sampling result is output to the C2f_Dual module. After processing, two branches are output, one waiting for the mid-scale image features of the Concat (feature concatenation) module in the Neck layer, and the other is sampled by the SCDown module.

[0042] ⑥ After sampling, the sampling results are output to the C2f_Dual module for image feature extraction.

[0043] ⑦ The proposed image features are sent to the SPPF (Fast Spatial Pyramid Pooling) module for pooling. The present invention uses multiple small pooling layers instead of a single large core pooling operation to further enrich feature information and reduce computational overhead.

[0044] ⑧ Output the pooled image features to the PSA (partial self-attention mechanism) module to improve the detection performance without significantly increasing the computational cost. After processing, two branches are output, one waiting for the small-scale image features of the Concat (feature concatenation) module in the Neck layer, and the other waiting for the UpSample (upper sample operation) module in the Neck layer.

[0045] In summary, the Backbone layer outputs image features of large, medium and small scales. Its structure mainly consists of 3 convolutional layers (Conv), 4 C2f_Dual modules, 2 SCDown modules, 1 SPPF module and 1 PSA module.

[0046] (2) The Neck layer is used to fuse image features. It takes the large, medium, and small scale image features output by the Backbone layer as input, performs top-down and bottom-up feature concatenation on the large, medium, and small scale image features, and outputs optimized image features of large, medium, and small scales.

[0047] The structure of the Neck layer mainly consists of 1 convolutional layer (Conv), 4 Concat (feature splicing) modules, 3 C2f_Dual modules, 1 C2fCIB (inverted bottleneck structure from convolution to feature splicing) module, 1 SCDown module and 2 Upsample modules.

[0048] (3) The Head layer is used to predict the target. The optimized image features of the large, medium and small scales output by the Neck layer are connected to the detection head v10Detect, respectively, to obtain the positioning frames and categories of the large, medium and small scale target objects, thereby realizing the positioning and recognition functions of the target. The optical image of the surface defect detection of the titanium alloy product is obtained.

[0049] Compared with the original YOLOv10n detection model, the improved YOLOv10n detection model of the present invention adds a lightweight C2f_Dual module to replace the original C2f module. and The convolution kernels of simultaneously process the same input feature map channels, fully combining and The advantages of the convolution kernel optimize the processing of information and the extraction of features. On the other hand, the efficient group convolution is used to efficiently arrange the convolution filters, which reduces the computational cost and the number of parameters, thereby successfully reducing the number of parameters and the amount of computation of the C2f module, and achieving the overall lightweight of the YOLOv10n detection model.

[0050] Specifically, the number of parameters of the YOLOv10n model before and after the improvement is 2.7 million and 2.5 million, a decrease of 7.4%. The amount of computation of the YOLO model before and after the improvement is 8.4 GFLOPs and 7.7 GFLOPs, a decrease of 8.3%.

[0051] This conclusion is drawn from the comparative experiment of the improved YOLOv10n detection model under the self-constructed titanium alloy surface defect dataset. The workstation operating system used in the experiment of the present invention is Window10, the GPU is NVIDIA RTX A5000, and the deep learning framework is Pytorch2.0.1+CUDA11.7. The dataset used for training the improved YOLOv10n detection model is a self-constructed titanium alloy product surface defect dataset, with a total of 2,800 images, divided into training set, validation set and test set in a ratio of 8:1:1, and the defect type is 1, i.e. scratches.

[0052] The training hyperparameters of the improved YOLOv10n detection model are set as follows: the number of iterations is 200, the learning rate momentum is 0.937, the weight decay coefficient is 0.0005, and the optimization algorithm is the stochastic gradient descent algorithm. The comparative experiments show that the improved YOLOv10n detection model has improved in terms of lightweight compared to the original YOLOv10n detection model.

[0053] S2: Acquire the original acoustic signal of the titanium alloy product and perform preprocessing to obtain the acoustic signal, then run the imaging algorithm to obtain the internal defect detection ultrasonic imaging, and then perform post-processing to obtain the internal defect detection acoustic signal information of the titanium alloy product.

[0054] The specific steps include: 1. Obtain the original sound signal of titanium alloy products: According to the size of the titanium alloy product to be tested, the various parameters of the water immersion ultrasonic testing system are reasonably adjusted, including the center frequency of the single crystal straight probe, the excitation voltage of the ultrasonic board, the scanning range, the step length and the water injection volume.

[0055] Afterwards, the scanning task is performed to obtain the original sound signal for internal defect detection of titanium alloy products.

[0056] Specifically, in an optional embodiment, the single crystal straight probe can be designed to have a fixed focal length, without special focusing, the excitation voltage of the ultrasonic board can be selected to be 100V, 200V or 400V, the scanning path is "S" shaped, and the scanning speed setting range is , the minimum stepping accuracy is set to 1mm.

[0057] After the single crystal straight probe is stimulated by the pulse emitted by the ultrasonic board, it vibrates to generate ultrasonic waves. These ultrasonic waves pass through the interface between the water in the water tank and the titanium alloy product placed in the water tank, reach the defect area, and then reflect back to the single crystal straight probe. The single crystal straight probe collects the original sound signals at various positions of the titanium alloy product in sequence by self-transmission and self-reception. Its time domain expression is: ; In the formula, | f(t) | is the absolute value of the amplitude of the original sound signal of the titanium alloy product received over time, is the amplitude of the original sound signal of the titanium alloy product, t is the ultrasonic propagation time, and the ultrasonic wave is reflected when the internal defect is detected. is the peak moment of the original acoustic signal of the titanium alloy product. is the working center frequency of the single crystal straight probe.

[0058] 2. Preprocess the original acoustic signal of the titanium alloy product, including signal envelope extraction and signal smoothing filtering, to obtain a processed acoustic signal with obvious defect echo and low noise.

[0059] (1) Envelope extraction is performed on the original acoustic signal of the titanium alloy product, which mainly involves Hilbert transform.

[0060] Specifically, in an optional embodiment, all positive frequency components are phase-shifted , all negative frequency components are phase shifted In order to obtain the signal envelope, it is necessary to construct an analytical signal, the formula is: ; In the formula, is the original sound signal of titanium alloy products. for The acoustic signal obtained after Hilbert transform is is an imaginary unit. Among them, the absolute value of the analytical signal This is the desired envelope signal.

[0061] (2) Envelope signal Smoothing filtering can eliminate clutter and noise interference as much as possible and improve signal quality.

[0062] Specifically, in an optional embodiment, the formula for smoothing filtering is: ; In the formula, is the original sound signal of titanium alloy products. is the smoothing filter window size, usually an odd number, is the acoustic signal after smoothing filtering, j and i are the labels of the signal sequence.

[0063] 3. Run the imaging algorithm: According to the amplitude and occurrence time of the internal defect echo of the product and the echo of the upper and lower surfaces of the product in the acoustic signal, a detection gate with a reasonable height and appropriate width is set, and the imaging algorithm is run to obtain the ultrasonic imaging of internal defect detection, that is, the burial depth map of the defect location.

[0064] Specifically, in an optional embodiment, two gates are set, namely gate A and gate B. Gate A is responsible for monitoring the surface defect echo of the titanium alloy product, and gate B is responsible for monitoring the internal defect echo of the titanium alloy product. The height and width of the gates need to be set according to manual experience and specific task conditions.

[0065] Internal defect ultrasonic imaging can be understood as a burial depth map of the defect location. The burial depth of a non-defective location is 0, and the calculation formula for the burial depth of a defective location is as follows: ; In the formula, is the propagation speed of ultrasonic waves in titanium alloy, which is 6100m / s. It is the moment when the peak of the echo on the surface of the titanium alloy product is located. It is the moment when the peak of the internal defect echo of the titanium alloy product is located. is the buried depth of internal defects.

[0066] 4. Post-processing of ultrasonic imaging for internal defect detection: Feature extraction is performed based on the shape and grayscale differences of the defect target in the ultrasonic imaging of internal defect detection, and the internal defect detection acoustic signal information of the titanium alloy product is further efficiently obtained, including the defect size, type, depth and center position coordinates.

[0067] In ultrasonic imaging for internal defect detection, there is a grayscale difference between the defect area and the background. In order to simplify the ultrasonic imaging data for internal defect detection and facilitate further analysis and identification, image binarization is introduced. The calculation formula is: ; In the formula, For the original image The gray value at is the gray value after binarization.

[0068] Find and mark the connected areas, i.e., defect areas, in the binary image, and further extract the defect feature information, including defect size, type, and center position coordinates. Associate the internal defect detection ultrasonic imaging with the defect feature information to obtain detailed internal defect detection acoustic signal information of titanium alloy products, i.e., defect size, type, depth, and center position coordinates.

[0069] S3: The optical image for surface defect detection of titanium alloy products obtained by S1 and the acoustic signal information for internal defect detection of titanium alloy products obtained by S2 are subjected to coordinate mapping and interactive programming to obtain the correct positioning of the optical image for surface defect detection of titanium alloy products and the acoustic signal information for internal defect detection of titanium alloy products in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion, and then a defect schematic diagram is made to obtain the photoacoustic image fusion defect detection result of the titanium alloy product.

[0070] The specific steps include: 1. Coordinate mapping: The optical image of surface defect detection of titanium alloy products is vertically projected into plane rectangular coordinate system A, and then the acoustic signal information of internal defect detection of titanium alloy products is vertically projected into plane rectangular coordinate system B. According to the theory of coordinate mapping, the optical image of surface defect detection of titanium alloy products and the acoustic signal information of internal defect detection of titanium alloy products in plane rectangular coordinate systems A and B are uniformly mapped into plane rectangular coordinate system C, that is, the photoacoustic detection information fusion background field.

[0071] During the coordinate transformation process, the relative position of the optical image of surface defect detection of the titanium alloy product to the entire titanium alloy product remains unchanged, thereby obtaining the correct positioning of the optical image of surface defect detection of each titanium alloy product in the photoacoustic detection information fusion background field and the correct positioning of the internal defect detection acoustic signal information of each titanium alloy product.

[0072] Specifically, the reasoning formula for mapping the center point of the surface defect detection optical image of the titanium alloy product in the plane rectangular coordinate system A to the plane rectangular coordinate system C of the photoacoustic detection information fusion background field, and the center point of the internal defect detection acoustic signal information of the titanium alloy product in the plane rectangular coordinate system B to the plane rectangular coordinate system C of the photoacoustic detection information fusion background field is: ; ; ; ; In the formula, and are the horizontal and vertical coordinates of the center point of the optical image of the surface defect detection of titanium alloy products in the plane rectangular coordinate system A, and They are the overall length and width of the titanium alloy product reflected in the plane rectangular coordinate system A. and are the horizontal and vertical coordinates of the center point of the internal defect detection acoustic signal information of the titanium alloy product in the plane rectangular coordinate system B, and are the overall length and width of the titanium alloy product reflected in the plane rectangular coordinate system B, and They are the actual length and width of the titanium alloy product. and are the horizontal and vertical coordinates of the center point of the optical image of the surface defect detection of the titanium alloy product in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion, and They are respectively the horizontal and vertical coordinates of the center point of the acoustic signal information of internal defect detection of titanium alloy products in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion.

[0073] 2. Draw a defect schematic diagram, that is, the photoacoustic image fusion defect detection results of titanium alloy products include photoacoustic detection information fusion results and decision-level detection results.

[0074] According to the correct placement of the optical image of surface defect detection of titanium alloy products and the acoustic signal information of internal defect detection of titanium alloy products in the rectangular coordinate system C of the background field plane of photoacoustic detection information fusion, a defect schematic diagram is made to obtain the photoacoustic image fusion defect detection result of titanium alloy products.

[0075] Specifically, in an optional embodiment, an interactive program is designed for the photoacoustic image fusion defect detection results of titanium alloy products to facilitate use by staff. The program automatically issues a decision-level detection result to assist subsequent repair and maintenance work.

[0076] The photoacoustic image fusion defect detection results of titanium alloy products are visual and interactive. The staff can view the defect information in the defect diagram by clicking the mouse. The decision-level detection results automatically issued by the program record the detection data of each defect in detail, and comprehensively evaluate the danger level of the titanium alloy product, including three levels of "low", "medium" and "high", and give reasonable opinions to assist in the subsequent repair and maintenance of titanium alloy products.

[0077] A second aspect of this embodiment provides a titanium alloy product defect detection device based on photoacoustic image fusion.

[0078] Figure 3 A schematic structural diagram of a titanium alloy product defect detection device based on photoacoustic image fusion proposed in one embodiment of the present invention is shown; Please refer to Figure 3 , a titanium alloy product defect detection device based on photoacoustic image fusion proposed in an embodiment of the present invention comprises: Camera 1, bracket 2, storage table 3, ultrasonic board 4, single crystal straight probe 5, gantry structure 6, programmable motion controller 7, motor 8, encoder 9, host computer 10 and water tank 11; camera 1 is arranged on the top of bracket 2, storage table 3 is arranged on the bottom of bracket 2, camera 1 and ultrasonic board 4 are respectively connected to host computer 10, ultrasonic board 4, water tank 11, programmable motion controller 7 and motor 8 are respectively connected to gantry structure 6, and ultrasonic board 4 is connected to single crystal straight probe 5.

[0079] The camera 1 arranged on the top of the bracket 2 is used to obtain the original optical image of the titanium alloy product placed on the storage table 3, and transmit it to the upper computer 10 to use a titanium alloy product defect detection method based on photoacoustic image fusion of the present invention to preprocess the original optical image, including image histogram equalization, image median filtering and image sharpening. The high-quality image to be detected obtained by preprocessing the original optical image is input into the improved YOLOv10n detection model to obtain the surface defect detection optical image of the titanium alloy product.

[0080] The programmable motion controller 7 controls the motion motor 8 to rotate to realize the scanning of the single crystal straight probe 5. The motor 8 drives the encoder 9 to rotate synchronously. The encoder 9 sends a synchronous excitation pulse by rotating to control the ultrasonic board 4 to synchronously emit the excitation pulse. The single crystal straight probe 5 is stimulated by the excitation pulse to emit an ultrasonic wave. The ultrasonic wave penetrates the water in the water tank 11 and the titanium alloy product placed in the water tank 11 in turn. The reflected ultrasonic wave is transmitted back to the single crystal straight probe 5. The ultrasonic echo is converted into an electrical signal and stored in the ultrasonic board 4, that is, the original sound signal of the titanium alloy product. It is then transmitted to the host computer 10 to perform defect detection using a titanium alloy product defect detection method based on photoacoustic image fusion according to the present invention.

[0081] The host computer 10 performs preprocessing on the original sound signal, including envelope extraction and smoothing filtering. The present invention uses a titanium alloy product defect detection method based on photoacoustic image fusion to run an imaging algorithm and post-process it to obtain internal defect detection sound signal information of the titanium alloy product.

[0082] The titanium alloy product defect detection method based on photoacoustic image fusion of the present invention is used to coordinately map the surface defect detection optical image of the titanium alloy product and the internal defect detection acoustic signal information of the titanium alloy product, and map them in the photoacoustic detection information fusion background field, so as to obtain the correct placement of the surface defect detection optical image of the titanium alloy product and the internal defect detection acoustic signal information of the titanium alloy product in the plane rectangular coordinate system C of the photoacoustic detection information fusion background field, and then make a defect schematic diagram to obtain the photoacoustic image fusion defect detection result of the titanium alloy product. At this point, the defect detection of the titanium alloy product is completed accurately and comprehensively.

[0083] The present invention provides a titanium alloy product defect detection method and device based on photoacoustic image fusion. The present invention uses a modal signal combined with photoacoustic to detect internal defects and surface defects of titanium alloy products, and uses a coordinate mapping method and makes a specific schematic diagram to mark the defect positions inside and on the surface of the defects. Such a detection method can simultaneously detect high-risk areas with internal and surface defects, and the detection range is more comprehensive. The obtained photoacoustic image fusion defect detection results of titanium alloy products are more reliable. The present invention uses an ultrasonic board, a single crystal straight probe, a programmable motion controller, a motor and a water tank to cooperate with each other to collect the original sound signal of the titanium alloy product, wherein the ultrasonic wave generated by the vibration passes through the interface between the water in the water tank and the titanium alloy product placed in the water tank, reaches the defect area, and is reflected to the single crystal straight probe. The original sound signals of various positions of the titanium alloy product are collected in sequence in this self-generated and self-received manner. Such a collection method can collect sound signals in all directions to the maximum extent, and the collection range is wider and the detection results are more accurate.

[0084] The present invention uses image histogram equalization, image median filtering, image sharpening and an improved YOLOv10n detection model to detect surface defects, which can make the image clearer, the algorithm lighter, and the obtained detection results more accurate. Envelope extraction, smoothing filtering and imaging algorithms are used to detect the size, type, depth and center position coordinates of internal defects, which not only reduces the interference of cluttered noise, but also improves the quality of acoustic signals, provides a guarantee for subsequent coordinate mapping, and ensures the integrity of the titanium alloy product defect detection method based on photoacoustic image fusion of the present invention, so that the detection range is more comprehensive and the results obtained are more accurate.

[0085] In summary, the titanium alloy product defect detection method and device based on photoacoustic image fusion provided by the present invention can detect defect locations more comprehensively and produce more accurate detection results.

[0086] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0087] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A titanium alloy product defect detection method based on photoacoustic image fusion, characterized in that: include: Coordinate mapping is performed on the optical image of the surface defect detection of the titanium alloy product and the acoustic signal information of the internal defect detection of the titanium alloy product to obtain the correct location of the optical image of the surface defect detection of the titanium alloy product and the acoustic signal information of the internal defect detection of the titanium alloy product in the rectangular coordinate system C of the photoacoustic detection information fusion background field plane, and then a defect schematic diagram is made to obtain the photoacoustic image fusion defect detection result of the titanium alloy product; The inference formula of the coordinate mapping is: ； ； ； ； In the formula, and are the horizontal and vertical coordinates of the center point of the optical image of the surface defect detection of titanium alloy products in the plane rectangular coordinate system A, and are the overall length and width of the titanium alloy product reflected in the plane rectangular coordinate system A, and are the horizontal and vertical coordinates of the center point of the internal defect detection acoustic signal information of the titanium alloy product in the plane rectangular coordinate system B, and are the overall length and width of the titanium alloy product reflected in the plane rectangular coordinate system B, and They are the actual length and width of the titanium alloy product. and are the horizontal and vertical coordinates of the center point of the optical image of the surface defect detection of the titanium alloy product in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion, and They are respectively the horizontal and vertical coordinates of the center point of the acoustic signal information of internal defect detection of titanium alloy products in the rectangular coordinate system C of the background field plane of the photoacoustic detection information fusion.

2. A titanium alloy product defect detection method based on photoacoustic image fusion as claimed in claim 1, characterized in that: The specific steps of obtaining the acoustic signal information of the internal defect detection of the titanium alloy product are: obtaining the original acoustic signal of the titanium alloy product and performing preprocessing to obtain the processed acoustic signal, running the imaging algorithm to obtain the internal defect detection ultrasonic imaging, and then performing post-processing to obtain the acoustic signal information of the internal defect detection of the titanium alloy product; The internal defect detection ultrasonic imaging is a burial depth map of the defect position, the burial depth of the non-defective position is 0, and the formula for the burial depth of the defective position is: ； In the formula, is the propagation speed of ultrasonic waves in titanium alloy products, It is the moment when the peak of the echo on the surface of the titanium alloy product is located. It is the moment when the peak of the internal defect echo of the titanium alloy product is located. is the buried depth of internal defects.

3. The titanium alloy product defect detection method based on photoacoustic image fusion according to claim 1, characterized in that: The specific steps of obtaining the optical image for surface defect detection of the titanium alloy product are: obtaining the original optical image of the titanium alloy product, performing preprocessing to obtain a high-quality image to be detected, and then inputting it into the improved YOLOv10n detection model to obtain the optical image for surface defect detection of the titanium alloy product; The preprocessing of the original optical image of the titanium alloy product includes image histogram equalization, image median filtering and image sharpening; The mapping function calculation formula in the image histogram equalization is: ； ； In the formula, Indicates the gray value is The probability of a pixel appearing in the original optical image of the titanium alloy product, Indicates that the gray value does not exceed The cumulative probability of the pixel is Represents the gray value after mapping, Indicates the range of grayscale values.

4. A titanium alloy product defect detection method based on photoacoustic image fusion as claimed in claim 3, characterized in that: The improved YOLOv10n detection model includes a lightweight C2f_Dual module; The lightweight C2f_Dual module includes and Convolution kernel and convolution filter using group convolution technique.

5. The titanium alloy product defect detection method based on photoacoustic image fusion according to claim 2, characterized in that: The preprocessing of the original sound signal of the titanium alloy product includes signal envelope extraction and signal smoothing filtering; The formula for the signal smoothing filter is: ； In the formula, is the original sound signal of titanium alloy products. is the smoothing filter window size, an odd number, is the acoustic signal after smoothing filtering, j and i are the labels of the signal sequence.

6. A titanium alloy product defect detection method based on photoacoustic image fusion as claimed in claim 5, characterized in that: The analytical signal formula constructed by the signal envelope extraction is: ； In the formula, is the original sound signal of titanium alloy products. for The acoustic signal obtained after Hilbert transform is Is an imaginary unit.

7. The titanium alloy product defect detection method based on photoacoustic image fusion according to claim 3, characterized in that: The specific formula for obtaining the original optical image of the titanium alloy product is: ； ； ； In the formula, is the horizontal field of view, is the sensor width, is the focal length of the camera, is the actual field of view width, is the object distance, is the image width, To obtain the actual pixel size of the original optical image of the titanium alloy product.

8. The titanium alloy product defect detection method based on photoacoustic image fusion according to claim 3, characterized in that: The calculation formula of the image median filtering is: ； In the formula, is the original optical image after median filtering. The gray value of the position, The original optical image is The gray value of the position, A two-dimensional template.

9. The titanium alloy product defect detection method based on photoacoustic image fusion according to claim 3, characterized in that: The image sharpening formula is: ； In the formula, The original optical image of the titanium alloy product after sharpening The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is The gray value of the position, The original optical image of the titanium alloy product is Grayscale value of the position.

10. A titanium alloy product defect detection device based on photoacoustic image fusion, characterized in that: include: A camera, a bracket, a storage table, an ultrasonic board, a single crystal straight probe, a gantry structure, a programmable motion controller, a motor, an encoder, a host computer and a water tank; the camera is arranged on the top of the bracket, and the storage table is arranged on the bottom of the bracket; the camera and the ultrasonic board are respectively connected to the host computer, the ultrasonic board, the water tank, the encoder, the programmable motion controller and the motor are respectively connected to the gantry structure, the ultrasonic board is connected to the single crystal straight probe, and the motor and the encoder are relatively connected through the gantry structure; The camera is used to obtain the original optical image of the titanium alloy product; The programmable motion controller is used to control the encoder and the single crystal straight probe to perform scanning motion on the gantry frame through the motor; The ultrasonic board is used to send excitation pulses and store original sound signals of titanium alloy products; The single crystal straight probe is used to convert the excitation pulse into ultrasonic waves; receive the ultrasonic waves that pass through the water in the water tank and the interface of the titanium alloy product placed in the water tank, that is, the original acoustic signal of the titanium alloy product; The host computer is used to perform photoacoustic image fusion defect detection on the titanium alloy product by utilizing the original optical image of the titanium alloy product and the original acoustic signal of the titanium alloy product.

Images