Titanium rod surface wear morphology detection method and system based on multispectral image fusion

Abstract

The invention provides a titanium rod surface wear morphology detection method and system based on multispectral image fusion, and relates to the technical field of image processing.The method comprises the steps that a spectrograph is connected, a target titanium rod is clamped, and multispectral image collection is executed by controlling a rotating table; setting a fusion dimension, taking multi-scale enhancement and confidence perception as processing targets, developing an image fusion device, and deploying the image fusion device on a detection platform; and based on the synchronization timestamp constraint, importing the multispectral image into a detection platform, executing adaptive fusion processing according to an image fusion device to determine a fused image, and performing wear positioning and background assimilation processing as a wear morphology detection result. According to the method and the device, the technical problem that the detection accuracy of the surface wear morphology of the titanium rod is influenced by information loss or noise introduction caused by direct fusion due to heterogeneity of the multispectral image is solved, and the detection accuracy of the surface wear morphology of the titanium rod is improved by developing the image fusion device to perform fusion processing on the multispectral image.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and particularly to a method and system for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion. Background Art

[0002] With the development of imaging technology, multi-spectral image technology has gradually become an emerging method in surface defect detection. By combining image data from different spectral bands (such as visible light, infrared, ultraviolet, etc.), multi-spectral image technology can comprehensively capture changes in different physical characteristics of the surface. Although the introduction of multi-spectral images provides new possibilities for the surface detection of titanium rods, due to the fact that multi-spectral images are formed by light of different wavelengths, there is significant spectral heterogeneity, and each wavelength image may have different characteristics, such as resolution, contrast, and noise level. When these images are directly fused, it will lead to the loss of key features or the introduction of noise, affecting subsequent image analysis and processing, such as the detection of the surface wear morphology of titanium rods, resulting in insufficient detection accuracy and efficiency.

[0003] In summary, in the prior art, there is a technical problem that due to the heterogeneity of multi-spectral images, direct fusion will lead to information loss or the introduction of noise, thereby affecting the accuracy of detecting the surface wear morphology of titanium rods. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion, so as to solve the technical problem in the prior art that due to the heterogeneity of multi-spectral images, direct fusion will lead to information loss or the introduction of noise, thereby affecting the accuracy of detecting the surface wear morphology of titanium rods.

[0005] In view of the above problems, this application provides a method and system for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion.

[0006] In a first aspect, the present application provides a method for detecting the wear morphology of a titanium rod surface based on multi-spectral image fusion. The method for detecting the wear morphology of a titanium rod surface based on multi-spectral image fusion is implemented through a detection system for the wear morphology of a titanium rod surface based on multi-spectral image fusion. Among them, the method for detecting the wear morphology of a titanium rod surface based on multi-spectral image fusion includes: connecting a spectrometer, clamping a target titanium rod and controlling a rotating table to perform multi-spectral image acquisition under multi-view constraints; setting a fusion dimension, taking multi-scale enhancement and confidence perception as processing objectives, developing an image fusion device and embedding it in the detection platform. Among them, the fusion dimension at least includes a data dimension, a feature dimension and a decision dimension. Introduce cross-modal attention into the feature dimension, and the processing objectives are homogeneous enhancement and confidence weighting of common elements at multiple scales, and directional enhancement and confidence verification of complementary elements; based on synchronous timestamp constraints, import the multi-spectral images into the detection platform, perform adaptive fusion processing according to the image fusion device to determine a fusion image, and perform wear localization and background assimilation processing based on corner window recognition as the wear morphology detection result, and generate a pop-up window display on the detection platform.

[0007] Optionally, determine a fusion method according to the fusion dimension; respectively construct fusion layers for the fusion methods corresponding to each fusion dimension, perform fusion layer cascading and construct a lateral interaction channel to determine the image fusion device.

[0008] Optionally, receive a fusion requirement, where the fusion requirement includes at least one fusion dimension; according to the fusion requirement, activate the fusion layer of the image fusion device and perform processing of common elements and complementary elements based on the fusion dimension in parallel to determine the fusion image.

[0009] Optionally, in the parallel execution of the processing of common elements and complementary elements based on the fusion dimension, the homogeneous enhancement and confidence weighting of common elements at multiple scales are performed, including: for any fusion dimension, identify the common elements of the multi-spectral image and perform differential scale processing based on homogeneity to determine a first enhancement result, where the differential scale is determined based on the difference between the homogeneity standard and the image standard; for the first enhancement result, determine N fusion groups through multi-image mapping; traverse the N fusion groups and perform confidence-weighted fusion processing to determine the common element fusion result.

[0010] Optionally, for each fusion group, determine the fusion weight of the features within the group through confidence calibration and perform fusion processing with the fusion weight as a constraint.

[0011] Optionally, in the parallel execution of the processing based on the common elements and complementary elements in the fusion dimension, the directional enhancement and confidence verification of the complementary elements are performed, including: introducing an element recognition baseline, where the element recognition baseline is a quality standard; enhancing the complementary elements directionally according to the element recognition baseline to determine a second enhancement result; and performing confidence verification on the second enhancement result to determine the complementary element fusion result.

[0012] Optionally, according to the spatial element distribution of the image, the spatial distribution splicing of the common element fusion result and the complementary element fusion result is performed to determine the fused image.

[0013] Optionally, the fused image is identified. By performing corner point recognition, a corner point window is determined, where the change of pixels is used as the corner point, and the corner point window is a preset pixel scale; the fused image is marked with the corner point window, where the corner point window marks the non-smooth points on the surface of the titanium rod.

[0014] Optionally, according to the distribution of the marked corner point windows, the worn morphology area is framed; for the fused image, the non-worn morphology area is background-assimilated as the worn morphology distribution; the titanium rod defect database is connected, and the worn morphology distribution is matched and corresponding marked to determine the worn morphology detection result, where the titanium rod defect database is built in the detection platform.

[0015] In a second aspect, the present application also provides a titanium rod surface wear morphology detection system based on multi-spectral image fusion for performing the titanium rod surface wear morphology detection method based on multi-spectral image fusion as described in the first aspect. The titanium rod surface wear morphology detection system based on multi-spectral image fusion includes: an image acquisition module for connecting a spectrometer, clamping the target titanium rod and controlling the turntable to perform multi-spectral image acquisition under multi-view constraints; an image fusion module for setting a fusion dimension, taking multi-scale enhancement and confidence perception as the processing objectives, developing an image fusion device and embedding it in the detection platform, where the fusion dimension at least includes a data dimension, a feature dimension and a decision dimension, introducing cross-modal attention into the feature dimension, and the processing objectives are the homogeneous enhancement and confidence weighting of common elements at multiple scales and the directional enhancement and confidence verification of complementary elements; a wear morphology detection module for importing the multi-spectral image into the detection platform based on the synchronous timestamp constraint, performing adaptive fusion processing according to the image fusion device to determine the fused image, and performing wear positioning and background assimilation processing based on corner point window recognition as the wear morphology detection result, and generating a pop-up window display on the detection platform.

[0016] One or more technical solutions provided in the present application have at least the following beneficial effects:

[0017] By connecting a spectrometer, clamping the target titanium rod and controlling the rotating table, multi-spectral image acquisition under multi-view constraints is performed; setting the fusion dimension, with multi-scale enhancement and confidence perception as the processing objectives, developing an image fusion device and embedding it in the detection platform. Among them, the fusion dimension at least includes the data dimension, the feature dimension, and the decision dimension. Cross-modal attention is introduced into the feature dimension, and the processing objectives are homogeneous enhancement and confidence weighting at the same scale of common elements, and directional enhancement and confidence verification of complementary elements; based on the synchronous timestamp constraint, the multi-spectral images are imported into the detection platform, and the adaptive fusion processing is performed according to the image fusion device to determine the fused image, and wear localization and background assimilation processing based on corner window recognition are carried out as the wear morphology detection result, and a pop-up window is displayed on the detection platform. That is to say, through multi-spectral image acquisition, the surface of the titanium rod is scanned and analyzed from multiple dimensions, and an image fusion device is developed to set multiple dimension branches. Guided by requirements, the corresponding dimension branches are triggered to perform targeted processing fusion and spatial splicing of common elements and directional elements, so as to improve the accuracy and reliability of the surface wear morphology detection result.

[0018] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are given below. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description. Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0020] Figure 1 It is a schematic flow chart of the method for detecting the surface wear morphology of a titanium rod based on multi-spectral image fusion in the present application;

[0021] Figure 2 It is a schematic structural diagram of the system for detecting the surface wear morphology of a titanium rod based on multi-spectral image fusion in the present application.

[0022] Description of the reference numerals: Image acquisition module 11, Image fusion module 12, Wear morphology detection module 13. Detailed Description of the Embodiments

[0023] The present application provides a method and system for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion, which solves the technical problem in the prior art that due to the heterogeneity of multi-spectral images, direct fusion will lead to information loss or noise introduction, thus affecting the accuracy of detecting the surface wear morphology of titanium rods. Through multi-spectral image acquisition, the surface of the titanium rod is scanned and analyzed from multiple dimensions. An image fusion device is developed to set multiple dimensional branches. Guided by requirements, the corresponding dimensional branches are triggered to perform targeted processing fusion and spatial splicing of common elements and directional elements, thereby improving the accuracy and reliability of the detection results of the surface wear morphology.

[0024] Next, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application. Additionally, it should be noted that for the sake of description, only the parts related to the present application are shown in the drawings rather than all of them.

[0025] Embodiment 1. Please refer to the attached Figure 1 , the present application provides a method for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion. Among them, the method for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion is executed by a system for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion. The method for detecting the surface wear morphology of titanium rods based on multi-spectral image fusion specifically includes the following steps:

[0026] S100: Connect the spectrometer, clamp the target titanium rod and control the turntable to perform multi-spectral image acquisition under multi-view constraints.

[0027] Specifically, the target titanium rod is fixed on the test platform through a clamping device to ensure that the titanium rod remains stable during the acquisition process and avoid image distortion caused by external interference. The turntable is usually driven by a precision motor and is used to rotate the target object (here the titanium rod) to different angles. Usually, the rotation angle is set to rotate 10° every 5 seconds to obtain surface information from different angles. The more angles can provide more perspective information. The spectrometer is installed in a suitable position and connected to the detection platform to ensure that image information can be correctly captured. Lights of different bands (such as visible light, infrared, and ultraviolet) are selected, and images of multiple bands are simultaneously obtained through the spectrometer. Images of each band can provide different levels of surface information. For example, visible light can help capture the surface morphology, while infrared images can show the heat distribution generated by friction.

[0028] A spectrometer is a device that can measure the intensity of light or electromagnetic radiation at different wavelengths. It is widely used in the acquisition of multi-spectral images and works in different bands according to different application scenarios, such as ultraviolet, visible light, infrared, etc., for acquiring multi-spectral images of the surface of the target titanium rod and obtaining reflection or emission information in different bands.

[0029] During the image acquisition process, by controlling the rotation state, images of the titanium rod surface at multiple angles are obtained to overcome wear or defects that cannot be observed at a single angle. At each viewing angle, the spectrometer will acquire images of the titanium rod surface in different bands, that is, image data of each angle is obtained simultaneously in multiple different spectral bands.

[0030] Exemplarily, the wear condition of a certain target titanium rod surface is relatively complex. There are not only minor wear on the surface but also local cracks. A spectrometer equipped with infrared and visible light bands is used. The rotation angle of the turntable is set to collect once every 15°. Image data of 4 bands are used during the acquisition process. There are a total of 20 angles, and image data of 4 bands are collected at each angle, resulting in at least 100 images in total. The infrared band image shows the temperature distribution, revealing the uneven distribution of heat generated by friction on the titanium rod surface; the visible light image shows the microscopic features of the surface wear area, revealing the depth and morphology of the surface wear; the ultraviolet band image shows the surface micro-cracks; the near-infrared band image shows the local temperature rise area caused by increased friction.

[0031] Through multi-view and multi-band image acquisition, a comprehensive understanding of the wear morphology of the titanium rod surface is achieved, ensuring that no minor wear or cracks are missed. Through multi-view acquisition, information loss caused by the limitation of a single viewing angle is avoided, especially in complex or irregular wear areas, ensuring the reliability of the detection results.

[0032] S200: Set the fusion dimension. With multi-scale enhancement and confidence perception as the processing objectives, develop an image fusion device and embed it in the detection platform. Among them, the fusion dimension at least includes the data dimension, the feature dimension, and the decision dimension. Introduce cross-modal attention into the feature dimension, and the processing objectives are homogeneous enhancement and confidence weighting at different scales of common elements, and directional enhancement and confidence verification of complementary elements.

[0033] Furthermore, S200 of this application includes:

[0034] According to the fusion dimension, determine the fusion method; respectively construct fusion layers for the fusion methods corresponding to each fusion dimension, perform fusion layer cascading and construct a lateral interaction channel to determine the image fusion device.

[0035] Specifically, multiple fusion dimensions are set to ensure effective fusion of information at different levels. The multiple fusion dimensions refer to different data levels or perspectives considered during the fusion of multiple spectral images, including the data dimension, feature dimension, and decision dimension. Each dimension corresponds to a different processing level in the image fusion process, ensuring that the fused image can retain more key information for accurate analysis. The data dimension involves the fusion of raw image data, integrating data from different bands. Through the fusion of the data dimension, data from different bands are spliced or combined to form a comprehensive multi-band data set. The feature dimension is fused based on features extracted from the image (such as edges, textures, morphology, etc.) to distinguish different types of surface wear or defects. The decision dimension determines the detection decision based on the fused feature data, such as whether there is wear, the location and degree of wear.

[0036] Select appropriate fusion methods according to different fusion dimensions. The fusion method refers to the specific fusion method used under each fusion dimension, including weighted average, max pooling, splicing fusion, contrast enhancement, etc. The data dimension fusion mainly targets image data from different bands or perspectives, and the corresponding fusion methods include the maximum value method, minimum value method, weighted average fusion, etc. The maximum value method selects the maximum value at each pixel position in each image to enhance certain features of the image, such as edges and textures. The minimum value method selects the minimum value of all images for each pixel to retain the darker areas in the image and can remove noise under certain conditions. Weighted average fusion assigns different weights to images of different bands and fuses them through weighted averaging. For each pixel, the pixel values of each source image (different spectra, perspectives, etc.) are weighted and averaged. For example, infrared images may be more informative in certain wear areas, while visible light images are clearer in details, and different weights can be given to different bands.

[0037] The fusion of the feature dimension focuses on features extracted from the image, such as edges, textures, morphology, etc., and the corresponding fusion methods include feature splicing, max pooling, etc. Feature splicing combines features (such as textures, edges, etc.) extracted from each band or perspective to generate a comprehensive feature vector. Max pooling refers to selecting the strongest feature value among the image features of multiple bands or perspectives to retain the most representative features, which is suitable for wear areas with obvious contrast.

[0038] Decision dimension fusion mainly focuses on the process of making a final decision based on different features, and the corresponding fusion methods include voting mechanisms, weighted decision-making, etc. The voting mechanism is based on multiple models or multiple dimensions (such as features in different bands), and the voting mechanism is used to determine the final result. For example, if a worn area is detected in all three bands, the voting mechanism is used to confirm the existence of wear in that area. Weighted decision-making is based on different dimensions (such as the confidence levels of different bands), and the decision outputs of each dimension are calculated by weighted summation.

[0039] According to the fusion method corresponding to each dimension, an independent fusion layer is constructed. The purpose of each fusion layer is to process and fuse image information for a specific dimension. For the fusion layer of data dimensions, the input layer involves multiple channels and receives multi-source data (images from different perspectives, spectral bands, etc.); the processing layer processes the input images according to the selected fusion method; the output layer outputs the processed data in the form of an image. Using the weighted average method, a dynamic weight layer is added to the fusion layer, and this layer adjusts the fusion weights of different data sources according to image quality, signal-to-noise ratio, or other metrics. In the processing layer, through the weighted average algorithm, the input images are first normalized, the weighted coefficients of each image are calculated, and then pixel-level weighted summation is performed.

[0040] For the fusion layer of feature dimensions, a corresponding layer is designed. Through the feature concatenation layer, multiple input feature maps are concatenated along the feature dimension (i.e., the channel dimension) to generate a new feature map, that is, all input feature maps are concatenated in the channel dimension to form a larger feature map. The concatenated feature map usually contains all features from different sources, and the concatenated feature map can be input into the convolutional layer for further feature extraction. Through the weighted summation layer, according to the weights of each feature map, the weights are applied to each input feature map, and then they are weighted and summed. The max pooling layer is used to pool each input feature map, and the pooled feature maps are merged as needed. The pooling operation usually reduces the spatial dimension and retains the most significant features. Finally, the fused feature map is output through the output layer. In addition, when constructing the feature fusion layer, it is also necessary to consider how to optimize the expression ability of the output feature map. By adding convolutional layers, different features are further extracted and fused. Adding a batch normalization layer helps to stabilize the training process and reduce the problems of gradient vanishing or gradient explosion. Finally, the constructed feature fusion layer outputs a fused feature map, which contains comprehensive information from multiple input feature maps.

[0041] Since the fusion task of feature dimensions is relatively complex, especially when dealing with modal differences, the representation of features may be different in different input modes. In order to enhance the effect of feature fusion, a cross-modal attention mechanism is introduced to allow the network to adaptively select important features and eliminate irrelevant features, thereby improving the quality of feature fusion. The cross-modal attention mechanism can selectively enhance the features of important modalities based on the feature importance of each modality by calculating the attention weights, thereby improving the fusion effect. For example, in the detection of wear morphology on the surface of titanium rods, infrared images may be able to better capture surface temperature changes, while visible light images can provide clearer information. Therefore, the weight of important modalities can be increased through the cross-modal attention mechanism.

[0042] There are many options for the fusion of decision dimensions. The specific method to be selected usually depends on the application scenario and specific needs. The voting method is suitable for classification tasks. There are usually two methods: majority voting and weighted voting. The majority voting method refers to selecting the category with the most prediction results as the final decision among the predictions of multiple models. The weighted voting rule assigns different weights to the prediction results of each model. The votes of models with larger weights are more influential, reducing the impact of errors of individual models and improving stability. If you choose to use the voting method, create a voting mechanism for the output of each input model or task. Collect the outputs of all models and then perform majority voting or weighted voting. For weighted voting, dynamically assign weights by learning the confidence of each model, which is completed through simple counting or weighted accumulation operations, and finally output the voting results.

[0043] When the prediction results of multiple models are close and the differences are not large, weighted voting can be used to adjust the influence of each model in the final decision according to its confidence. Weighted voting can not only improve the stability of the overall decision, but also reduce the impact of a certain accidental erroneous model on the decision. Especially when multiple models perform well, weighted voting can smooth and optimize the overall results. When the prediction differences between models are more obvious, majority voting is usually more effective. Because in this case, the judgments of multiple models that are more consistent are usually more credible, while the judgments of a very small number of models may have large deviations. The majority voting mechanism can effectively avoid the impact of extreme predictions by letting the choice of the majority model determine the final result. In particular, when some models may be overfitted or misidentified, the influence of minority voting will be effectively suppressed.

[0044] Weighted summation fuses the outputs of different models. The output of each model is multiplied by a weight and then summed to get the final output. Based on the model's performance on the training set, in certain specific tasks, or obtained through user-side input, a weight is assigned to each decision source (i.e., the output of each model), and the output of each model is multiplied by the corresponding weight and summed to finally output a fused decision result.

[0045] The fusion hierarchical cascade connects fusion layers of different dimensions (such as data, feature, and decision fusion layers) in sequence. That is, to combine these fusion layers into a comprehensive image fusion device, it is necessary to cascade these three fusion layers. By connecting the output ends of the fusion layers of different dimensions together, a composite fusion structure is formed. The lateral interaction channels establish channels for information transfer between multiple dimensions, allowing different dimensions (such as data dimension, feature dimension, and decision dimension) to communicate and share information with each other, enhancing the connection between each level, so that the output of each layer not only depends on the input of the current dimension, but can also reflect the information of other dimensions, thereby improving the overall fusion effect.

[0046] Through the lateral interaction channels, the fusion results from the data dimension can interact with the fusion results from the feature dimension and the decision dimension, improving the fusion effect of each layer. For example, the data dimension is adjusted by the features extracted by the feature dimension, the feature dimension optimizes the feature extraction strategy according to the output of the decision dimension, and the decision dimension makes more accurate decisions using the information from the data and feature dimensions, ultimately improving the overall fusion effect. When implementing these interactions, methods such as skip connections or residual connections are used to make the information flow between different layers smoother, avoiding information loss or bottlenecks. After the optimization of the cascade and the lateral interaction channels, the final output will be a multi-dimensional and multi-level fused image representation that can simultaneously contain the information of the data, feature, and decision layers.

[0047] Through the design of the hierarchical cascade and the lateral interaction channels, an image fusion device is obtained, which combines all the information of the data dimension, the feature dimension, and the decision dimension, and can generate high-quality fused images. It can not only efficiently fuse data from different sources, but also integrate information through the hierarchical connection and interaction channels, making the image fusion device have stronger expression ability and decision-making ability.

[0048] Taking multi-scale enhancement and confidence perception as processing objectives, the image information is enhanced at different scales. Especially when detecting the wear on the surface of the titanium rod, it is necessary to capture wear features of different sizes and shapes. Confidence perception adjusts the fusion weights according to the reliability of different regions in the image.

[0049] For common elements, the processing objectives are homogeneous enhancement and confidence weighting. Common elements refer to similar or consistent features such as shape, texture, and structure that can be found in different perspectives and different modalities (e.g., visible light and infrared images). Homogeneous enhancement of common elements means uniformly improving the performance of features at multiple scales, while confidence weighting is to weight these features according to the importance or reliability of different regions, and determine their contribution to the final fusion result based on the importance of each region, ensuring that high-confidence regions dominate in the fusion. For different regions, confidence estimation is performed based on clarity, contrast, and texture information. Edge regions, regions with rich texture, etc. are considered to have higher confidence and will thus be given greater weights. In other words, multi-scale enhancement refers to processing images at multiple scales to enhance the performance of specific features in the image. Homogeneous enhancement means uniformly enhancing the same type of features in the image at multiple scales, making the performance of these common elements more consistent and clear at different scales.

[0050] For complementary elements, the processing objectives are directional enhancement and confidence verification. Complementary elements refer to features that contain complementary information and provide different information in images of different perspectives or different modalities. Different from common elements, complementary elements usually cannot be comprehensively obtained through a single modality or a single perspective, providing multiple information levels for image analysis. Directional enhancement is to adopt different processing strategies according to the characteristics of each piece of information to directionally enhance certain regions or features in the image, such as enhancing features highly related to the worn area in the image. Confidence verification is to verify the reliability of each region after fusion to ensure the reliability of the final fusion result. Confidence verification of image quality (such as clarity, contrast, etc.) is to evaluate image quality indicators, determine the imaging quality of each region, and thus estimate the confidence score. For example, calculate the clarity, contrast, etc. of the region, and normalize multiple indicators to the same dimension (such as 0-1) through the maximum-minimum normalization method, assign corresponding weights according to the importance of each indicator, and perform weighted calculation based on the weights and multiple indicators to obtain the confidence score. Usually, the weights are obtained based on the importance of each indicator in the wear detection of the titanium rod surface, combined with historical experience. If the regional confidence score is high, the region is considered reliable, while for regions with low confidence, additional verification is required, such as multiple samplings or image comparisons, to confirm whether wear has actually occurred in this region.

[0051] Embed the developed image fuser into the detection platform. Whenever new image data is collected by the detection platform, immediately perform fusion processing on the images among the data dimension, feature dimension, and decision dimension, and generate the final fused image. By setting the fusion dimension, trigger the corresponding dimension branch according to different requirements, and perform targeted processing fusion and spatial stitching of the common elements and directional elements to improve the image fusion result and ensure that the final detection result is more accurate.

[0052] S300: Based on the synchronous timestamp constraint, import the multi-spectral image into the detection platform, determine the fused image according to the adaptive fusion processing performed by the image fuser, perform wear localization and background assimilation processing based on corner window recognition, and use it as the wear morphology detection result, and generate a pop-up window display on the detection platform.

[0053] Specifically, the synchronous timestamp constraint means that when collecting multi-spectral images, ensure that all image acquisition devices (such as spectrometers, cameras, sensors, etc.) collect data at the same time point (timestamp), avoid image data inconsistency or misalignment problems caused by time differences, and ensure the correct spatial alignment of multi-view or multi-spectral images. Import the multi-spectral images with the same timestamp into the detection platform, and through the image fuser in the detection platform, automatically select the appropriate fusion layer to activate according to the characteristics and fusion requirements of the images, and perform the processing of common elements and complementary elements under the fusion dimension in parallel. According to the element distribution in the space, perform spatial distribution stitching on the fused results of the common elements and the fused results of the complementary elements to determine the fused image.

[0054] After obtaining the fused image, perform corner window recognition to identify the pixels with significant changes in the image, that is, the defective or worn areas on the surface of the titanium rod. After identifying the corner window, perform background assimilation processing to highlight the worn area. Replace the pixels in the background area with a blank background to highlight the worn morphology area. Finally, according to the worn morphology area, match the corresponding worn morphology distribution in the titanium rod defect database to form the final detection result, including the worn morphology, defect type, location, etc. For example, it is identified that there are scratches and microcracks on the surface of the titanium rod, and these areas are highlighted through background assimilation processing. A pop-up window will appear on the detection platform, including information such as the presence of scratches (X-axis = 100mm, Y-axis = 50mm) and microcracks (X-axis = 150mm, Y-axis = 75mm) on the surface of the titanium rod.

[0055] Through the synchronous timestamp constraint, ensure the temporal synchronization of image data from different sources. Through corner window recognition and background assimilation processing, accurately identify the worn morphology on the surface of the titanium rod, and by highlighting the worn area, improve the visibility and accuracy of the detection result.

[0056] Furthermore, S300 of this application includes:

[0057] Receive a fusion requirement, where the fusion requirement includes at least one fusion dimension; according to the fusion requirement, activate the fusion layer of the image fusion device, and perform processing based on common elements and complementary elements under the fusion dimension in parallel to determine the fused image.

[0058] Specifically, receive the specific requirements for image fusion processing, including the dimensions to be fused, the fusion target, etc. According to the received fusion requirement, determine the required fusion dimension, and correspondingly activate the fusion layers in the image fusion device, including the data fusion layer, the feature fusion layer, and the decision fusion layer. Each fusion layer processes different fusion dimensions. For example, if the requirement includes enhancing the worn area and highlighting the crack feature, identify the data dimension (processing different spectral data) and the feature dimension (processing surface wear and cracks).

[0059] After activating the corresponding fusion layer, perform processing on the common elements and complementary elements under the fusion dimension in parallel. The common elements refer to the features or information shared between different modalities or data sources, while the complementary elements refer to the features with a complementary relationship between different modalities. Stitch the fusion results of the common elements and the fusion results of the complementary elements according to the spatial element distribution of the image to determine the fused image.

[0060] Exemplarily, assume that when performing surface defect detection on a titanium rod, 2 different spectral images are used: one shows the global wear feature, and the other highlights the crack area. Receive a fusion requirement to enhance the worn area and highlight the crack feature; according to the requirement, activate the fusion layers of the data dimension and the feature dimension. The data dimension is responsible for processing different spectral information, and the feature dimension is responsible for highlighting the wear and crack features; perform processing in parallel to enhance the worn area (common elements) and the crack area (complementary elements) respectively; after parallel processing, fuse the two into a final image. As a result, the final image clearly shows the surface wear and cracks of the titanium rod, and at the same time fuses the feature information under different spectra.

[0061] The fused image can comprehensively display all key features (such as wear and cracks), help analyze the state of the titanium rod surface, significantly improve the image processing effect, and accurately identify and locate various defects.

[0062] Furthermore, the present application further includes the following steps:

[0063] In the parallel processing of the common elements and complementary elements under the fusion dimension, perform homogeneous enhancement and confidence weighting of the common elements at multiple scales, including:

[0064] For any fusion dimension, identify the common elements of the multispectral image and perform differential scale processing based on homogeneity to determine the first enhancement result, where the differential scale is determined based on the difference between the homogeneity standard and the image standard; for the first enhancement result, determine N fusion groups through multi-image mapping; traverse the N fusion groups and perform confidence-weighted fusion processing to determine the fused result of the common elements.

[0065] For each fusion group, determine the fusion weights of the features within each group through confidence calibration, and perform fusion processing with the fusion weights as the constraint.

[0066] Specifically, for performing homogeneous enhancement and confidence weighting of common elements at multiple scales: For each fusion dimension, identify the common elements in the multispectral image, that is, the similar or consistent parts in different spectral image modalities, such as surface texture or crack features. Common elements are usually some basic structural features in the image, which have similar manifestations in different image modalities. Through differential scale processing, perform homogeneous enhancement of these common elements at different scales, and perform image enhancement processing in regions with similar features to ensure the improvement of image quality in similar regions without introducing too much noise or irrelevant information.

[0067] The homogeneity standard refers to determining the processing method through similar or consistent regions in the image, used to identify the scales of similar regions in the image, and perform different scale processing on these regions to extract features at different levels. The image standard is the feature expression of the image itself at different scales. Identify the regions with similar features in the image according to the homogeneity standard, process the image at different scales, compare the differences between the image at a larger scale (macro level) and a smaller scale (micro level), and obtain the difference, that is, the difference or change amount calculated between different scales. Differential scale processing is a multi-scale image processing method that compares and processes image features at different scales (such as large scale, small scale), analyzes the changes in the image, and extracts important information.

[0068] By processing the differences of the image at different scales, obtain a preliminary image enhancement result, that is, the first enhancement result, which represents the state of the common elements after multi-scale enhancement and is the preliminary image enhancement effect obtained after differential scale processing. Through multi-image mapping, map the first enhancement result with other image information to form N fusion groups. Through multi-image mapping, features at the same position (such as cracks, worn areas) can be aligned to avoid information loss or incorrect fusion. For example, there are two images from different perspectives, showing the same part at different angles. Through multi-image mapping, align the two images so that features at the same position in each image (such as cracks) can be aligned.

[0069] After completing multi-image mapping, multiple images are obtained, where each image may contain different spectral information, different perspective information, or information at different scales. Through multi-image mapping, N fusion groups are determined, and each group contains information of the same position or the same feature from different images, aiming to obtain richer and more accurate image content by fusing these image features. A fusion group refers to a subset of images combined from multiple images or different sources during the image fusion process.

[0070] For each fusion group, through confidence calibration, appropriate fusion weights are assigned to the features within each fusion group, which reflects the credibility of each information source during the fusion process. Confidence calibration determines the weight of each feature in the final fusion according to the reliability of each feature. By evaluating the features within each fusion group, the confidence of each feature is calculated, and then the weights of each feature are adjusted according to these confidences. Information sources with higher confidence will be assigned higher weights, while information sources with lower confidence will be assigned lower weights. According to the fusion weights corresponding to each fusion group, each fusion group is fused to obtain the fusion result of the common elements. The features within each fusion group will be weighted and fused according to their corresponding fusion weights to ensure that high-weight features have a greater impact on the final fusion result and ensure that the most important and reliable features dominate in the final image.

[0071] Exemplarily, assume that three images are collected, respectively from different spectral bands (infrared, ultraviolet, and visible light). Image 1 is from visible light, the image quality evaluation clarity is 0.8, Image 2 is from infrared light, the image quality evaluation clarity is 0.7, Image 3 is from ultraviolet light, and the image quality evaluation is 0.6. The common elements of the three images are the worn areas, cracks, micro-cracks, etc. on the surface of the titanium rod. According to the difference between the image standard and the homogeneous standard, the differential scale is calculated. The edge contrast of Image 1 is 1.2, the homogeneous standard is 0.8, and the differential scale is 1.2 - 0.8 = 0.4. The edge contrast of Image 2 is 0.9, the homogeneous standard is 0.7, and the differential scale is 0.9 - 0.7 = 0.2. The edge contrast of Image 2 is 0.7, the homogeneous standard is 0.6, and the differential scale is 0.7 - 0.6 = 0.1. According to the size of the differential scale, the enhancement effect of Image 1 is the strongest, Image 2 is the second, and Image 3 is the weakest. According to the different features of the three images, multi-image mapping is performed. By observing the features of different images, the following three fusion groups are determined: Group 1 is the fusion of Image 1 and Image 2, focusing on enhancing the display of the surface worn area and cracks; Group 2 is the fusion of Image 2 and Image 3, highlighting the crack and micro-crack areas; Group 3 is the fusion of Image 1 and Image 3, strengthening the display of the overall texture and micro-cracks on the surface of the titanium rod.

[0072] According to the clarity scores of each image, the fusion weights of the images in each fusion group are calculated. The clarity of Image 1 in Group 1 is 0.8, and the clarity of Image 2 is 0.7. By calculating the sum of the clarity scores as 1.5, the weight of Image 1 is calculated as approximately 0.53 (0.8 / 1.5), and the weight of Image 2 is 0.47. Image 1 and Image 2 are weighted and fused. Since the weight of Image 1 is higher, the details of the worn areas of Image 1 are enhanced in the final image; in Group 2, the clarity of Image 2 is 0.70, and the clarity of Image 3 is 0.60. The calculated weight of Image 2 is 0.54, and the weight of Image 3 is 0.46. Image 2 and Image 3 are weighted and fused, and the crack details of Image 2 are better retained, and the display of micro-cracks in Image 3 is enhanced; in Group 3, the clarity of Image 1 is 0.80, and the clarity of Image 3 is 0.60. The calculated weight of Image 1 is 0.57, and the weight of Image 3 is 0.43. After Image 1 and Image 3 are fused, the details of the overall texture and micro-cracks are clearer.

[0073] Image fusion is performed through differential scale processing and confidence calibration, and the weights are determined according to the confidence of the features at each stage. Finally, the fused image is obtained, significantly reducing the influence of noise and enhancing the effective information in the image, thereby achieving more accurate surface wear detection.

[0074] Furthermore, this application also includes the following steps:

[0075] During the parallel execution of the processing of the common elements and complementary elements under the fusion dimension, the directional enhancement and confidence verification of the complementary elements are performed, including:

[0076] Introduce an element recognition baseline, where the element recognition baseline is a quality standard; according to the element recognition baseline as the orientation, perform enhancement processing on the complementary elements to determine the second enhancement result; for the second enhancement result, perform confidence verification to determine the complementary element fusion result.

[0077] Specifically, introduce an element recognition baseline, that is, set a quality standard to guide the enhancement processing of complementary elements, which defines the quality and manifestation forms that certain features (such as cracks, scratches, surface unevenness, etc.) in the image should possess. For example, it is required that the width of the crack is at least 5 pixels or more, the contrast of the crack reaches 20% of the average contrast of the image, the width of the scratch should be at least 3 pixels of the image, and the noise in the background area should be controlled within ±10% of the standard deviation of the image.

[0078] Based on the determined element recognition baseline, the complementary elements in the image are directionally enhanced. According to the requirements of the element recognition baseline, different complementary features are amplified or strengthened to facilitate better extraction of these complementary elements from the image. The features that appear in different spectral channels in the multispectral image are enhanced, especially those regions that can provide supplementary information, and the parts with complementary information in the image are strengthened. For example, assume that a thermal defect area is identified in the infrared image and a scratch area is identified in the visible light image. According to the predefined quality criteria (such as crack width, scratch contrast, etc.), the thermal defect and the scratch are amplified.

[0079] The second enhancement result is the image result obtained after enhancing the complementary elements, which makes the valuable complementary information more prominent than the original image. After the second enhancement result is processed, confidence verification is performed. Check whether the enhanced area meets the expectations and whether this enhancement can provide real and effective feature information. Confidence verification detects whether the enhanced area exceeds the normal range and whether unreliable information is introduced due to noise or processing errors. For example, if the crack area is over-enhanced resulting in an unnatural boundary, or too many false signals are introduced during the enhancement process, these unqualified areas need to be excluded through the confidence verification algorithm. After directional enhancement and confidence verification, a reliable complementary element fusion result is obtained, which integrates the complementary information of multiple spectral channels and has been verified to ensure its quality. For example, a threshold is set according to the enhancement degree of each complementary element to determine whether to include it in the final fusion result. If the confidence level of the enhanced crack feature is greater than 80%, then this crack feature can be considered valid and enter the final fusion process.

[0080] Through directional enhancement, the complementary features in different channels can be highlighted, making small wear or cracks more clearly visible in the fusion image. By confidence verification, noise and irrelevant information are removed to ensure that the final image reaches a high level in terms of quality and information accuracy.

[0081] Furthermore, the present application further includes the following steps:

[0082] According to the spatial element distribution of the image, the spatial distribution splicing of the common element fusion result and the complementary element fusion result is performed to determine the fusion image.

[0083] Specifically, according to the spatial element distribution of the image, that is, the spatial distribution pattern of each feature (such as surface cracks, wear areas, etc.) in the image, including the distribution method and density. The spatial distribution splicing of the common element fusion result and the complementary element fusion result is performed to ensure that these features can be correctly aligned in space. The fusion results of the common elements and the complementary elements each represent different feature regions in the image. First, it is necessary to determine which parts are common regions and which parts are complementary regions.

[0084] Common elements generally represent the features that are prevalent in the image (such as overall wear or surface roughness), while complementary elements represent special or difficult-to-identify areas (such as cracks or micro-defects). Through spatial stitching, the information of common elements and complementary elements is combined to obtain an image that comprehensively reflects all features, ensuring that all important information in the image can be displayed while avoiding information loss or mis-stitching. If the boundary between common elements and complementary elements is clear, the two are directly stitched together without obvious transition problems. If the boundary between common elements and complementary elements is blurred or there is a certain overlap, a weighted method is used to ensure smooth transition between the two during the stitching process, such as using Gaussian weighting or boundary blurring technology to achieve smooth transition at the stitching point and avoid abrupt boundaries.

[0085] Spatial distribution stitching refers to combining the features of two or more images according to their spatial positions. In other words, the information of different regions in multiple images is combined into one image to form an overall image. After spatial stitching and processing, a fused image is obtained, which synthesizes the information of common elements and complementary elements and reflects the state of the target object (such as the surface of a titanium rod) more completely and accurately.

[0086] Through the spatial stitching of common elements and complementary elements, an image showing multiple features such as wear and cracks on the surface of the titanium rod is formed, identifying small cracks and wear areas and reducing the possibility of false detection and missed detection.

[0087] Furthermore, the present application further includes the following steps:

[0088] Identify the fused image. By performing corner recognition, determine the corner window. Herein, the change of pixels is used as the corner, and the corner window is a preset pixel scale. Use the corner window to identify the fused image, where the corner window marks the non-smooth points on the surface of the titanium rod.

[0089] According to the distribution of the identified corner windows, frame the wear morphology area. For the fused image, perform background assimilation on the non-wear morphology area as the wear morphology distribution. Connect to the titanium rod defect database, match and correspondingly mark the wear morphology distribution to determine it as the wear morphology detection result, where the titanium rod defect database is built into the detection platform.

[0090] Specifically, corner recognition is performed on the fused image to detect significant feature points in the image, which usually appear as regions with large variations in the image. Corners are a special type of image feature that represent the most significant points of change in the image, typically occurring at the intersection of edges in two different directions or in strong contrast regions in the image. The surface of the titanium rod should be smooth under normal circumstances, meaning there are no obvious scratches, cracks, or other defects on its surface. Therefore, when certain regions in the image exhibit significant pixel changes, it usually means there are flaws or defects in these regions.

[0091] Use corner recognition methods to locate the regions of significant change in the image. Through corner detection algorithms (such as Harris corner detection, Shi-Tomasi corner detection, etc.), identify the points with drastic pixel changes in the image, which usually appear at the edges or corners of the image.

[0092] After the corners are recognized, set a window with a preset pixel scale according to the position of each corner. The corner window is a fixed-size area defined around each corner, used to capture the local features around the corner for analyzing the detailed changes in the image. For example, if the pixel scale is set to 5x5, the window size for each corner is 5x5 pixels. Further analyze the pixel value changes within these windows to check for surface unevenness, cracks, etc. If the position of a certain corner in the image is (100, 150) and the pixel scale is 5x5, then the range of this corner window will be the area from (98, 148) to (102, 152).

[0093] According to the distribution of the corner windows, connect the corners into a rectangular or polygonal framework, which will represent the positions of the flaw regions. Based on the settings of the corner windows, mark the regions on the surface of the titanium rod where flaws may exist, such as cracks, wear, or depressions. By identifying multiple corner windows and marking them on the image, form the framework of the flaw regions. Finally, all the flaw regions located by corner recognition will be shown on the image of the titanium rod surface.

[0094] Through corner recognition and window calibration, obtain the regions on the surface of the titanium rod where defects may exist, and frame the regions corresponding to these corners, namely the wear or flaw points. The wear morphology region refers to the surface damage region on the titanium rod surface caused by factors such as friction and pressure, usually showing scratches, depressions, cracks, or uneven surface structures.

[0095] Perform background assimilation on the non-worn topography regions of the fused image, change the pixel values of these regions to a unified background color (such as white or transparent), and distinguish the non-worn regions from the worn regions, thereby enhancing the contrast of the worn topography regions. In the processed image, the non-worn regions may be set to a white or transparent background, while the worn regions retain their original texture and color, making the worn topography more prominent and easier to observe.

[0096] Compare and match the framed worn topography regions with the known defects in the titanium rod defect database, determine whether the worn regions conform to the known defect patterns, and make markings. If similar features are detected, extract the corresponding defect types from the database and make markings to obtain the worn topography detection results, including information such as the specific location and defect type of the worn regions. The worn topography detection results are displayed in graphical or text form on the detection platform to understand the state of the titanium rod surface. The titanium rod defect database is a dataset built into the detection platform, which records the characteristic information of different types of titanium rod defects and is used to compare, match, and mark the worn topography on the titanium rod surface to help judge and identify specific defect types.

[0097] By framing the worn topography regions and performing background assimilation, the worn regions are extracted, avoiding the influence of interference factors, thereby improving the accuracy of wear detection. The background assimilation process can effectively highlight the worn regions, significantly enhancing the contrast between the worn regions and the background, which helps to automatically identify and mark the defect types in the worn topography regions and improves the detection efficiency.

[0098] In summary, the method for detecting the worn topography on the titanium rod surface based on multi-spectral image fusion provided by this application has the following beneficial effects:

[0099] By connecting a spectrometer, clamping the target titanium rod and controlling the rotating table, multi-spectral image acquisition under multi-view constraints is performed; setting the fusion dimension, with multi-scale enhancement and confidence perception as the processing objectives, developing an image fusion device and embedding it in the detection platform. Among them, the fusion dimension at least includes the data dimension, the feature dimension and the decision dimension. Cross-modal attention is introduced into the feature dimension, and the processing objectives are homogeneous enhancement and confidence weighting of common elements at multiple scales, and directional enhancement and confidence verification of complementary elements; based on the synchronous timestamp constraint, the multi-spectral images are imported into the detection platform, and the adaptive fusion process is performed according to the image fusion device to determine the fused image, and wear positioning and background assimilation processing based on corner window recognition are carried out. As the wear morphology detection result, a pop-up window is displayed on the detection platform. That is to say, through multi-spectral image acquisition, the surface of the titanium rod is scanned and analyzed from multiple dimensions, and an image fusion device is developed to set multiple dimension branches. Guided by the requirements, the corresponding dimension branches are triggered to perform targeted processing fusion and spatial splicing of common elements and directional elements, so as to improve the accuracy and reliability of the surface wear morphology detection result.

[0100] Embodiment 2. Based on the same inventive concept as the method for detecting the surface wear morphology of a titanium rod based on multi-spectral image fusion in the foregoing Embodiment 1, the present application also provides a system for detecting the surface wear morphology of a titanium rod based on multi-spectral image fusion. Please refer to the attached Figure 2 , the system for detecting the surface wear morphology of a titanium rod based on multi-spectral image fusion includes:

[0101] An image acquisition module 11, which is used to connect a spectrometer, clamp the target titanium rod and control the rotating table to perform multi-spectral image acquisition under multi-view constraints; an image fusion module 12, which is used to set the fusion dimension, with multi-scale enhancement and confidence perception as the processing objectives, develop an image fusion device and embed it in the detection platform. Among them, the fusion dimension at least includes the data dimension, the feature dimension and the decision dimension. Cross-modal attention is introduced into the feature dimension, and the processing objectives are homogeneous enhancement and confidence weighting of common elements at multiple scales, and directional enhancement and confidence verification of complementary elements; a wear morphology detection module 13, which is used to import the multi-spectral images into the detection platform based on the synchronous timestamp constraint, perform adaptive fusion processing according to the image fusion device to determine the fused image, and perform wear positioning and background assimilation processing based on corner window recognition. As the wear morphology detection result, a pop-up window is displayed on the detection platform.

[0102] Furthermore, the image fusion module 12 in the system for detecting the surface wear morphology of a titanium rod based on multi-spectral image fusion is also used for:

[0103] Determine the fusion method according to the fusion dimension; construct a fusion layer for each fusion method corresponding to the fusion dimension, perform cascading of fusion levels and construct a lateral interaction channel to determine the image fusion device.

[0104] Furthermore, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0105] Receive a fusion requirement, where the fusion requirement includes at least one fusion dimension; according to the fusion requirement, activate the fusion layer of the image fusion device, and perform processing based on common elements and complementary elements under the fusion dimension in parallel to determine the fused image.

[0106] Furthermore, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0107] In the parallel execution of processing based on common elements and complementary elements under the fusion dimension, perform homogeneous enhancement and confidence weighting of common elements at multiple scales, including:

[0108] For any fusion dimension, identify the common elements of the multi-spectral images and perform differential scale processing based on homogeneity to determine a first enhancement result, where the differential scale is determined based on the difference between the homogeneity standard and the image standard; for the first enhancement result, determine N fusion groups through multi-image mapping; traverse the N fusion groups and perform confidence weighting fusion processing to determine the common element fusion result.

[0109] Furthermore, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0110] For each fusion group, determine the fusion weight of the features within the group through confidence calibration, and perform fusion processing with the fusion weight as a constraint.

[0111] Furthermore, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0112] In the parallel execution of processing based on common elements and complementary elements under the fusion dimension, perform directional enhancement and confidence verification of complementary elements, including:

[0113] Introduce an element recognition baseline, where the element recognition baseline is a quality standard; perform enhancement processing on the complementary elements in the direction based on the element recognition baseline to determine a second enhancement result; for the second enhancement result, perform confidence verification to determine the complementary element fusion result.

[0114] Further, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0115] According to the distribution of image space elements, perform spatial distribution splicing on the common element fusion result and the complementary element fusion result to determine the fused image.

[0116] Further, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0117] Identify the fused image. By performing corner point recognition, determine a corner point window, where the change in pixels is used as the corner point, and the corner point window is a preset pixel scale; use the corner point window to mark the fused image, where the corner point window marks the non-smooth points on the surface of the titanium rod.

[0118] Further, the wear morphology detection module 13 in the titanium rod surface wear morphology detection system based on multi-spectral image fusion is further configured to:

[0119] Frame the wear morphology area according to the distribution of the marked corner point windows; for the fused image, perform background assimilation on the non-wear morphology area as the wear morphology distribution; connect to the titanium rod defect database, match and mark the wear morphology distribution to determine the wear morphology detection result, where the titanium rod defect database is built into the detection platform.

[0120] The various embodiments in this specification are described in a progressive manner, and the key point of each embodiment is to illustrate the differences from other embodiments. The Figure 1 The method and specific examples of the titanium rod surface wear morphology detection based on multi-spectral image fusion in the first embodiment are equally applicable to the titanium rod surface wear morphology detection system based on multi-spectral image fusion in this embodiment. Through the detailed description of the titanium rod surface wear morphology detection method based on multi-spectral image fusion above, those skilled in the art can clearly know the titanium rod surface wear morphology detection system based on multi-spectral image fusion in this embodiment. Therefore, for the sake of simplicity of the specification, it will not be elaborated here.

[0121] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0122] Obviously, those skilled in the art can also make several improvements and modifications to this application without departing from the principle of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A titanium rod surface wear morphology detection method based on multispectral image fusion, characterized in that: include: Connect the spectrometer, clamp the target titanium rod and perform multi-spectral image acquisition under multi-view constraints by controlling the rotating stage; Set the fusion dimension, take multi-scale enhancement and confidence perception as the processing goals, develop an image fusion device and embed it in the detection platform, wherein the fusion dimension at least includes data dimension, feature dimension and decision dimension, introduce cross-modal attention into the feature dimension, and the processing goals are multi-scale homogeneous enhancement and confidence weighting of common elements, and directional enhancement and confidence verification of complementary elements; Based on the synchronization timestamp constraint, the multispectral image is imported into the detection platform, and the image fusion device performs adaptive fusion processing to determine the fused image, and wear positioning and background assimilation processing based on corner window recognition are performed. As the wear morphology detection result, a pop-up window display is generated on the detection platform.

2. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 1 is characterized in that: Develop image fuser, including: Determine a fusion method according to the fusion dimension; A fusion layer is constructed for each fusion method corresponding to each fusion dimension, the fusion layer is cascaded and a lateral interaction channel is constructed to determine the image fuser.

3. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 1 is characterized in that: The method further comprises: performing adaptive fusion processing according to the image fusion device to determine a fused image, comprising: receiving a fusion requirement, wherein the fusion requirement includes at least one fusion dimension; According to the fusion requirement, the fusion layer of the image fuser is activated, and the processing based on the common elements and the complementary elements under the fusion dimension is performed in parallel to determine the fused image.

4. The titanium rod surface wear morphology detection method based on multispectral image fusion as claimed in claim 3 is characterized in that: In parallel, the processing of common elements and complementary elements based on the fusion dimension is performed, and the multi-scale homogeneity enhancement and confidence weighting of common elements are performed, including: For any fusion dimension, identifying common elements of the multispectral image and performing differential scale processing based on homogeneity to determine a first enhancement result, wherein the differential scale is determined based on a difference between a homogeneity standard and an image standard; For the first enhancement result, determining N fusion groups through multi-image mapping; The N fusion groups are traversed, confidence weighted fusion processing is performed, and a common element fusion result is determined.

5. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 4 is characterized in that: For each fusion group, the fusion weight of the features in each group is determined through confidence checking, and the fusion processing is performed with the fusion weight as a constraint.

6. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 4 is characterized in that: In parallel, the processing of common elements and complementary elements based on the fusion dimension is performed, and the directional enhancement and confidence verification of complementary elements are performed, including: Introducing an element identification baseline, wherein the element identification baseline is a quality standard; According to the element recognition baseline being directional, the complementary element is enhanced to determine a second enhancement result; A confidence verification is performed on the second enhancement result to determine a complementary element fusion result.

7. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 6 is characterized in that: According to the distribution of image space elements, the common element fusion result and the complementary element fusion result are spatially distributed and spliced ​​to determine the fused image.

8. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 7 is characterized in that: Wear location based on corner window recognition, including: Identify the fused image, and determine a corner point window by performing corner point identification, wherein the corner point is determined by the trend of pixels, and the corner point window is a preset pixel scale; The fused image is marked with the corner point window, wherein the corner point window marks the non-smooth points on the surface of the titanium rod.

9. The titanium rod surface wear morphology detection method based on multispectral image fusion according to claim 8 is characterized in that: After the fused image is marked, the following steps are included: According to the distribution of the corner point windows marked, the wear morphology area is framed; For the fused image, background assimilation is performed on the non-wear morphology area to obtain wear morphology distribution; Connect to the titanium rod defect database, match and mark the wear morphology distribution, and determine it as the wear morphology detection result, wherein the titanium rod defect database is built into the detection platform.

10. Titanium rod surface wear morphology detection system based on multi-spectral image fusion, characterized in that: The steps for implementing the titanium rod surface wear morphology detection method based on multispectral image fusion as described in any one of claims 1 to 9, the titanium rod surface wear morphology detection system based on multispectral image fusion comprises: An image acquisition module is used to connect to the spectrometer, clamp the target titanium rod and perform multi-spectral image acquisition under multi-view constraints by controlling the rotating stage; An image fusion module is used to set the fusion dimension, take multi-scale enhancement and confidence perception as the processing goals, develop an image fuser and embed it in the detection platform, wherein the fusion dimension at least includes the data dimension, the feature dimension and the decision dimension, introduces cross-modal attention into the feature dimension, and the processing goals are the homogeneous enhancement and confidence weighting of common elements at multiple scales, and the directional enhancement and confidence verification of complementary elements; The wear morphology detection module is used to import the multispectral image into the detection platform based on the synchronization timestamp constraint, perform adaptive fusion processing according to the image fusion device to determine the fused image, perform wear positioning and background assimilation processing based on corner point window recognition, and generate a pop-up window display on the detection platform as the wear morphology detection result.

Images