Transparent film abnormal layer position inference method, device, equipment and medium

Abstract

The present disclosure provides a transparent film abnormal layer position inference method, device, equipment and medium, comprising: acquiring multi-focus layer image data; acquiring multi-angle image data; acquiring multi-waveband image data; analyzing the multi-peak value response characteristics of the abnormal target under different focal layers according to the multi-focus layer image data; analyzing the multi-angle response characteristics of the abnormal target under different angles according to the multi-angle image data; analyzing the waveband difference response characteristics of the abnormal target under different wavebands according to the multi-waveband image data; determining the relative prediction probability of the abnormal target being located on the transparent film, the relative prediction probability of the abnormal target being located in the transparent film, and the relative prediction probability of the abnormal target being located under the transparent film according to the multi-peak value response characteristics, the multi-angle response characteristics and the waveband difference response characteristics; and determining the layer position inference result according to the relative prediction probability of the abnormal target being located on the transparent film, the relative prediction probability of the abnormal target being located in the transparent film, and the relative prediction probability of the abnormal target being located under the transparent film. Thus, the accuracy of abnormal layer position inference is improved.

Description

Technical Field

[0001] The embodiments of this disclosure relate to the field of image processing technology, and more specifically, to a method, apparatus, device, and medium suitable for inferring abnormal layers in a transparent film. Background Technology

[0002] In the field of transparent film inspection, anomalies may be located on the upper surface of the film, between layers within the film, at the interface beneath the film, or on the surface of the adjacent substrate. Although anomalies at different layers may appear as similar bright spots, dark spots, edges, or textures in two-dimensional images, their origins and treatment strategies differ significantly. If the layer location cannot be distinguished, relying solely on the "presence or absence of anomalies" test results is often insufficient to support actual production.

[0003] In related technologies, the existence of anomalies is mainly determined by a single image or a single imaging mode. For example, anomaly detection and layer classification can be performed directly on a single bright field image by the naked eye or a pre-trained deep learning model.

[0004] However, using existing technologies, the accuracy of stratigraphic inference is low. Summary of the Invention

[0005] The embodiments described herein provide a method, apparatus, device, and medium for inferring abnormal layers in a transparent membrane, overcoming the aforementioned problems.

[0006] Firstly, based on the content of this disclosure, a method for inferring abnormal layers in a transparent membrane is provided, including: Acquire multi-focal layer image data of anomalous targets in the same transparent film corresponding to different focal layers; acquire multi-angle image data of anomalous targets in the same transparent film corresponding to different observation angles; and acquire multi-band image data of anomalous targets in the same transparent film corresponding to different spectral bands. The multi-peak response characteristics of the anomalous target at different focal layers are analyzed based on the multi-focal layer image data; the multi-angle response characteristics of the anomalous target at different viewing angles are analyzed based on the multi-angle image data; and the band difference response characteristics of the anomalous target at different bands are analyzed based on the multi-band image data. Based on the multi-peak response characteristics, the multi-angle response characteristics, and the band difference response characteristics, the relative predicted probability of the abnormal target being located on the transparent film, the relative predicted probability of the abnormal target being located inside the transparent film, and the relative predicted probability of the abnormal target being located below the transparent film are determined. Based on the relative predicted probability that the anomalous target is located on the transparent membrane, the relative predicted probability that the anomalous target is located inside the transparent membrane, and the relative predicted probability that the anomalous target is located below the transparent membrane, the layer inference result of the anomalous target relative to the transparent membrane is determined.

[0007] Secondly, according to the present disclosure, a device for inferring abnormal layers of a transparent membrane is provided, comprising: The acquisition module is used to acquire multi-focal layer image data of anomaly targets in the same transparent film corresponding to different focal layers; acquire multi-angle image data of anomaly targets in the same transparent film corresponding to different observation angles; and acquire multi-band image data of anomaly targets in the same transparent film corresponding to different spectral bands. The analysis module is used to analyze the multi-peak response characteristics of the anomalous target at different focal layers based on the multi-focal layer image data; to analyze the multi-angle response characteristics of the anomalous target at different viewing angles based on the multi-angle image data; and to analyze the band difference response characteristics of the anomalous target at different bands based on the multi-band image data. The first determining module is used to determine, based on the multi-peak response characteristics, the multi-angle response characteristics, and the band difference response characteristics, the relative predicted probability that the abnormal target is located on the transparent film, the relative predicted probability that the abnormal target is located inside the transparent film, and the relative predicted probability that the abnormal target is located below the transparent film; The second determining module is used to determine the layer inference result of the abnormal target relative to the transparent membrane based on the relative predicted probability that the abnormal target is located on the transparent membrane, the relative predicted probability that the abnormal target is located inside the transparent membrane, and the relative predicted probability that the abnormal target is located below the transparent membrane.

[0008] Thirdly, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the transparent membrane abnormal layer inference method as described in any of the above embodiments.

[0009] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the transparent film abnormal layer inference method as described in any of the above embodiments.

[0010] The method for inferring the torsional position of anomalies in a transparent film provided in this application includes: acquiring multi-focal layer image data of anomalies in the same transparent film corresponding to different focal layers; acquiring multi-angle image data of anomalies in the same transparent film corresponding to different observation angles; and acquiring multi-band image data of anomalies in the same transparent film corresponding to different spectral bands. The method analyzes the multi-peak response characteristics of the anomaly at different focal layers based on the multi-focal layer image data; analyzes the multi-angle response characteristics of the anomaly at different observation angles based on the multi-angle image data; analyzes the band difference response characteristics of the anomaly at different spectral bands based on the multi-band image data; determines the relative prediction probability of the anomaly being located on the transparent film, the relative prediction probability of the anomaly being located inside the transparent film, and the relative prediction probability of the anomaly being located below the transparent film based on the multi-peak response characteristics, multi-angle response characteristics, and band difference response characteristics; and determines the torsional position inference result of the anomaly relative to the transparent film based on the relative prediction probability of the anomaly being located on the transparent film, the relative prediction probability of the anomaly being located inside the transparent film, and the relative prediction probability of the anomaly being located below the transparent film. In this way, by jointly processing images from three dimensions—focus layer, observation angle, and wavelength—abnormal targets at different locations on the transparent film can be accurately distinguished, effectively solving the problem of layer misjudgment in traditional single-dimensional detection and improving the accuracy and reliability of abnormal layer inference.

[0011] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. It should be understood that the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure, wherein: Figure 1 This is a flowchart illustrating a method for inferring abnormal layers in a transparent membrane, as disclosed in this publication.

[0013] Figure 2 This is a schematic diagram of the structure of a transparent membrane abnormal layer inference device provided in this disclosure.

[0014] Figure 3 This is a schematic diagram of the structure of a computer device provided in this disclosure.

[0015] It should be noted that the elements in the attached diagram are schematic and not drawn to scale. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are also within the scope of protection of this disclosure.

[0017] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this subject matter pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the specification and in the relevant art, and shall not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, the statement of “connecting” or “coupling” two or more parts together shall mean that these parts are directly joined together or joined through one or more intermediate components.

[0018] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0019] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists, A and B exist simultaneously, or B exists. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Terms such as "first" and "second" are only used to distinguish one component (or part of a component) from another component (or another part of a component).

[0020] In the description of this application, unless otherwise stated, "multiple" means two or more (including two), and similarly, "multiple groups" means two or more (including two groups).

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0022] Figure 1 This is a schematic flowchart of a method for inferring abnormal layers in a transparent membrane according to an embodiment of this disclosure. Figure 1As shown, the specific process of the method for inferring abnormal layers of transparent membranes includes: S110. Acquire multi-focal layer image data of abnormal targets in the same transparent film corresponding to different focal layers; acquire multi-angle image data of abnormal targets in the same transparent film corresponding to different observation angles; and acquire multi-band image data of abnormal targets in the same transparent film corresponding to different bands.

[0023] Multi-focal layer image data can be used to provide imaging information at different depths; multi-angle image data can be used to provide imaging information from different observation perspectives; and multi-band image data can be used to provide imaging information with different optical responses. After acquiring multi-focal layer image data, multi-angle image data, and multi-band image data, object-level registration and cropping can be performed on the same anomalous target to ensure comparability between different layers.

[0024] S120. Analyze the multi-peak response characteristics of abnormal targets at different focal layers based on multi-focal layer image data; analyze the multi-angle response characteristics of abnormal targets at different observation angles based on multi-angle image data; analyze the band difference response characteristics of abnormal targets at different bands based on multi-band image data.

[0025] Specifically, it can calculate indicators such as edge sharpness, local contrast, segmentation stability, and target area change of abnormal targets in multi-focal layer image data, and extract significant peak positions, peak widths, and peak stability features of the focal layers. Multi-peak response features may include, but are not limited to, significant peak positions, peak widths, peak offsets, and peak stability of the focal layers.

[0026] The changes in the highlight center position, shadow direction, halo radius, and edge bright band shift of anomaly targets are statistically analyzed from multiple observation angles to form multi-angle response characteristics. These multi-angle response characteristics may include, but are not limited to, highlight center drift, halo radius change, shadow direction change, edge bright band shift, and reflection intensity attenuation.

[0027] By analyzing the intensity difference, edge visibility difference, and material response difference between the anomalous target and the background in different bands, band difference response characteristics are obtained. These band difference response characteristics may include, but are not limited to, intensity difference, contrast difference, edge visibility difference, and segmentation stability difference.

[0028] In some embodiments, analyzing the multi-peak response characteristics of abnormal targets at different focal layers based on multi-focal layer image data includes: Peak detection is performed on multi-focal layer image data to obtain the significant peak positions of the abnormal target under different focal layers; based on the significant peak positions of the abnormal target under different focal layers, the peak width corresponding to each significant peak is calculated; based on the significant peak positions of the abnormal target under different focal layers and the peak width corresponding to each significant peak, the multi-peak response characteristics of the abnormal target under different focal layers are determined.

[0029] Specifically, significant peaks can be selected based on the ratio of signal peak value to local background signal at different focal layers. The average width of all significant peaks is calculated, and the offset distance of the peak position relative to the average depth position at each focal layer is statistically analyzed to extract peak width, peak offset, and peak stability features. Peak stability features can be determined based on the ratio of the number of significant peaks corresponding to abnormal targets to the total number of focal layers. The smaller the ratio, the weaker the peak stability; the larger the ratio, the stronger the peak stability.

[0030] In some embodiments, the analysis of multi-angle response characteristics of abnormal targets at different viewing angles based on multi-angle image data includes: Extract the highlight center drift, shadow direction change, halo expansion radius, and edge bright band offset of the abnormal target from multi-angle image data under different viewing angles; determine the multi-angle response characteristics of the abnormal target under different viewing angles based on the highlight center drift, shadow direction change, halo expansion radius, and edge bright band offset of the abnormal target under different viewing angles.

[0031] Among them, the highlight center drift can be calculated based on the coordinate difference of the highlight center position under different observation angles; the shadow direction change can be obtained by statistically analyzing the difference in the directional angle between adjacent angles of the shadow area; the halo expansion radius can be extracted based on the maximum distance from the outer halo contour of the abnormal target to the center under different angles; and the edge bright band offset can be calculated by the relative position difference between the bright band at the transparent film interface and the center of the abnormal target under different angles.

[0032] In some embodiments, analyzing the band difference response characteristics of abnormal targets in different bands based on multi-band image data includes: Extract the differential response information of abnormal targets and background in different bands from multi-band image data. The differential response information includes at least one of the following: intensity difference, contrast difference, poor edge visibility, and poor segmentation stability. Based on the differential response information of abnormal targets and background in different bands, determine the band differential response features of abnormal targets in different bands.

[0033] Among them, the intensity difference can be obtained by calculating the difference in the mean gray level between the region where the abnormal target is located and the background region under the same band; the contrast difference is calculated by the difference in the standard deviation of gray level between the abnormal target and the adjacent background region; the edge visibility difference is determined based on the statistical difference in the gradient magnitude of the edge of the abnormal target under different bands; and the segmentation stability difference can be calculated by the difference in the crossover ratio of the segmentation results of the abnormal target under different bands.

[0034] S130. Based on the multi-peak response characteristics, multi-angle response characteristics, and band difference response characteristics, determine the relative prediction probability of the abnormal target being located on the transparent film, the relative prediction probability of the abnormal target being located inside the transparent film, and the relative prediction probability of the abnormal target being located below the transparent film.

[0035] In some embodiments, based on multi-peak response characteristics, multi-angle response characteristics, and band difference response characteristics, the relative prediction probability of an anomalous target being located on the transparent film, the relative prediction probability of an anomalous target being located inside the transparent film, and the relative prediction probability of an anomalous target being located below the transparent film are determined, including: By inputting multi-peak response features, multi-angle response features, and band difference response features into the stratigraphic inference model, the initial prediction probabilities of the anomalous target being located on the transparent membrane, the initial prediction probabilities of the anomalous target being located inside the transparent membrane, and the initial prediction probabilities of the anomalous target being located below the transparent membrane are obtained. The initial prediction probabilities of the anomalous target being located on the transparent membrane, the initial prediction probabilities of the anomalous target being located inside the transparent membrane, and the initial prediction probabilities of the anomalous target being located below the transparent membrane are normalized to obtain the relative prediction probabilities of the anomalous target being located on the transparent membrane, the relative prediction probabilities of the anomalous target being located inside the transparent membrane, and the relative prediction probabilities of the anomalous target being located below the transparent membrane.

[0036] The hierarchical inference model can be a rule engine, a statistical model, a machine learning model, a neural network model, or a combination of the above models.

[0037] Therefore, by using the hierarchical inference model to probabilistically infer the hierarchical location of anomaly targets, it is possible to effectively combine the response characteristics of anomaly targets in different dimensions, improve the accuracy and stability of the hierarchical inference results of anomaly targets, and reduce the probability of misjudgment.

[0038] S140. Based on the relative predicted probability of the anomalous target being located on the transparent membrane, the relative predicted probability of the anomalous target being located inside the transparent membrane, and the relative predicted probability of the anomalous target being located below the transparent membrane, determine the layer inference result of the anomalous target relative to the transparent membrane.

[0039] Among them, the main layer label and confidence level can be output based on the highest probability; if the probability distribution is close, the evidence is conflicting, or the uncertainty is too high, a reshoot suggestion can be made to increase the focal layer density, angle sample or band sample.

[0040] When the probability distributions of different strata are too close or the evidence is insufficient, reverse triggering can be used to collect additional samples of the focal layer or angle to improve the stability of strata determination. The probability distribution of strata can be expressed as shown in the following formula (1).

[0041] P ( L | X )= f(z _ peak , w _ z , V _ theta , R ( θ ), E ( z , θ ), D ( λ (1) In formula (1), P ( L | X ) represents the layer probability distribution, outputting the probability and confidence interval of an anomalous target belonging to the membrane, whether it is above, inside, or below the membrane; L Indicates the layer category; X Represents the joint eigenvector; function f It can be implemented using rule engines, statistical models, machine learning models, or neural network models. z _ peak This indicates the significant peak position of the anomalous target in the focal layer sequence; the closer it is to the surface reference focal plane, the more likely it is to be an anomalous on the film; this relationship can be obtained through equipment calibration. w _ z This indicates the focal layer peak width or effective response span; a wider peak width means a deeper anomaly, more complex scattering, or a longer imaging path. V _ theta It represents the multi-angle highlight center drift vector; it can be used to measure the displacement law of the reflection center as the viewing angle changes, and is related to the layer. R ( θ ) represents the halo radius variation function, used to describe the degree of halo expansion at different angles, and can distinguish between surface adhesion and deep refraction. E ( z , θ ) represents the joint function of edge sharpness, which indicates the variation of abnormal edges under different focal layers and different angles. D ( λ ) represents the band response difference function, which is used to describe the degree of distinction between anomalies and background under different bands.

[0042] In some embodiments, the layer inference result of the anomalous target relative to the transparent membrane is determined based on the relative predicted probability of the anomalous target being located on the transparent membrane, the relative predicted probability of the anomalous target being located inside the transparent membrane, and the relative predicted probability of the anomalous target being located below the transparent membrane, including: From the relative predicted probability of the anomalous target being on the transparent membrane, the relative predicted probability of the anomalous target being inside the transparent membrane, and the relative predicted probability of the anomalous target being below the transparent membrane, the layer corresponding to the highest predicted probability is determined as the main layer; the main layer is determined as the layer where the anomalous target is located inside the transparent membrane.

[0043] If the number of layers corresponding to the maximum predicted probability is one, then that layer is determined as the main layer; if the number of layers corresponding to the maximum predicted probability is at least two, then the main layer can be selected from at least two layers by combining priority or voting methods, which can effectively reduce the probability of deviation in layer inference and ensure the rationality of the inference results.

[0044] In some embodiments, before determining the master layer as the layer in which the abnormal target is located within the transparent membrane, the method further includes: if the number of layers corresponding to the maximum prediction probability is at least two, then based on a preset layer priority, selecting the layer with higher priority from the at least two layers corresponding to the maximum prediction probability as the master layer.

[0045] The preset order of layer priority can be: inside the transparent membrane is higher than above the transparent membrane, and above the transparent membrane is higher than below the transparent membrane. The preset layer priority can be set based on the degree of impact of abnormal targets on the performance of the transparent membrane, which facilitates the priority location of anomalies with a greater impact on product quality and better meets the needs of actual testing.

[0046] In this embodiment, for contaminants on the membrane, the system observed a sharper and more stable response peak at the focal plane near the surface, while the specular drift changed more directly with the angle, resulting in an experimental output probability of 0.83 for contaminants on the membrane. For foreign objects under the membrane, the system observed a lag in peak position and an increased peak width, with significantly improved distinction from the background in a certain near-infrared band, thus outputting an output probability of 0.76 with a moderate confidence level. For inclusions within the membrane, the system found that their focal plane and angle responses were between those on and under the membrane, with moderate distinction across different bands, ultimately outputting the highest probability of inclusion within the membrane, while retaining a higher uncertainty to trigger manual verification.

[0047] In this embodiment, multi-focal layer image data of an anomalous target corresponding to different focal layers in the same transparent film are acquired; multi-angle image data of an anomalous target corresponding to different observation angles in the same transparent film are acquired; and multi-band image data of an anomalous target corresponding to different spectral bands in the same transparent film are acquired. Based on the multi-focal layer image data, the multi-peak response characteristics of the anomalous target at different focal layers are analyzed; based on the multi-angle image data, the multi-angle response characteristics of the anomalous target at different observation angles are analyzed; based on the multi-band image data, the spectral difference response characteristics of the anomalous target at different spectral bands are analyzed; based on the multi-peak response characteristics, multi-angle response characteristics, and spectral difference response characteristics, the relative prediction probability of the anomalous target being located on the transparent film, the relative prediction probability of the anomalous target being located inside the transparent film, and the relative prediction probability of the anomalous target being located below the transparent film are determined; based on the relative prediction probability of the anomalous target being located on the transparent film, the relative prediction probability of the anomalous target being located inside the transparent film, and the relative prediction probability of the anomalous target being located below the transparent film, the layer inference result of the anomalous target relative to the transparent film is determined. In this way, by jointly processing images from three dimensions—focus layer, observation angle, and wavelength—abnormal targets at different locations on the transparent film can be accurately distinguished, effectively solving the problem of layer misjudgment in traditional single-dimensional detection and improving the accuracy and reliability of abnormal layer inference.

[0048] In summary, the method described in this embodiment is applicable to anomaly detection scenarios involving transparent films, optical films, polarizers, composite films, or other translucent / semi-translucent materials. It enables relative tomographic analysis of anomalies in transparent films without requiring actual 3D measurements, effectively improving the accuracy of anomaly identification and process interpretation capabilities in transparent material detection. It outputs tomographic conclusions usable for process decision-making without constructing a true 3D model; it significantly improves the ability to distinguish between anomalies on, inside, and under the film in transparent film scenarios; it is compatible with existing industrial imaging equipment, making it highly feasible in engineering; it can be coupled with re-image mechanisms to improve the stability of identification on difficult samples; and it provides higher-value information for cleaning, rework, responsibility interface delineation, and process attribution.

[0049] Figure 2 This is a schematic diagram of a transparent membrane abnormality inference device provided in this embodiment. The transparent membrane abnormality inference device may include: The acquisition module 210 is used to acquire multi-focal layer image data of an abnormal target in the same transparent film corresponding to different focal layers; acquire multi-angle image data of an abnormal target in the same transparent film corresponding to different observation angles; and acquire multi-band image data of an abnormal target in the same transparent film corresponding to different spectral bands.

[0050] Analysis module 220 is used to analyze the multi-peak response characteristics of abnormal targets at different focal layers based on multi-focal layer image data; to analyze the multi-angle response characteristics of abnormal targets at different observation angles based on multi-angle image data; and to analyze the band difference response characteristics of abnormal targets at different bands based on multi-band image data.

[0051] The first determining module 230 is used to determine the relative predicted probability of an abnormal target being located on the transparent film, the relative predicted probability of an abnormal target being located inside the transparent film, and the relative predicted probability of an abnormal target being located below the transparent film, based on the multi-peak response characteristics, multi-angle response characteristics, and band difference response characteristics.

[0052] The second determining module 240 is used to determine the layer inference result of the abnormal target relative to the transparent membrane based on the relative predicted probability of the abnormal target being located on the transparent membrane, the relative predicted probability of the abnormal target being located inside the transparent membrane, and the relative predicted probability of the abnormal target being located below the transparent membrane.

[0053] In some embodiments, the analysis module 220 is specifically used for: Peak detection is performed on multi-focal layer image data to obtain the significant peak positions of the abnormal target under different focal layers; based on the significant peak positions of the abnormal target under different focal layers, the peak width corresponding to each significant peak is calculated; based on the significant peak positions of the abnormal target under different focal layers and the peak width corresponding to each significant peak, the multi-peak response characteristics of the abnormal target under different focal layers are determined.

[0054] In some embodiments, the analysis module 220 is specifically used for: Extract the highlight center drift, shadow direction change, halo expansion radius, and edge bright band offset of the abnormal target from multi-angle image data under different viewing angles; determine the multi-angle response characteristics of the abnormal target under different viewing angles based on the highlight center drift, shadow direction change, halo expansion radius, and edge bright band offset of the abnormal target under different viewing angles.

[0055] In some embodiments, the analysis module 220 is specifically used for: Extract the differential response information of abnormal targets and background in different bands from multi-band image data. The differential response information includes at least one of the following: intensity difference, contrast difference, poor edge visibility, and poor segmentation stability. Based on the differential response information of abnormal targets and background in different bands, determine the band differential response features of abnormal targets in different bands.

[0056] In some embodiments, the first determining module 230 is specifically used for: By inputting multi-peak response features, multi-angle response features, and band difference response features into the stratigraphic inference model, the initial prediction probabilities of the anomalous target being located on the transparent membrane, the initial prediction probabilities of the anomalous target being located inside the transparent membrane, and the initial prediction probabilities of the anomalous target being located below the transparent membrane are obtained. The initial prediction probabilities of the anomalous target being located on the transparent membrane, the initial prediction probabilities of the anomalous target being located inside the transparent membrane, and the initial prediction probabilities of the anomalous target being located below the transparent membrane are normalized to obtain the relative prediction probabilities of the anomalous target being located on the transparent membrane, the relative prediction probabilities of the anomalous target being located inside the transparent membrane, and the relative prediction probabilities of the anomalous target being located below the transparent membrane.

[0057] In some embodiments, the second determining module 240 is specifically used for: From the relative predicted probability of the anomalous target being on the transparent membrane, the relative predicted probability of the anomalous target being inside the transparent membrane, and the relative predicted probability of the anomalous target being below the transparent membrane, the layer corresponding to the highest predicted probability is determined as the main layer; the main layer is determined as the layer where the anomalous target is located inside the transparent membrane.

[0058] In some embodiments, a selection module is also included.

[0059] The selection module is used to select the layer with higher priority from the at least two layers corresponding to the maximum prediction probability as the main layer if there are at least two layers with the maximum prediction probability.

[0060] The transparent membrane abnormal layer inference device provided in this disclosure can perform the above method embodiments. Its specific implementation principle and technical effect can be found in the above method embodiments, and will not be repeated here.

[0061] This application also provides a computer device. Please refer to the following for details. Figure 3 , Figure 3 This is a basic structural block diagram of the computer device in this embodiment.

[0062] The computer device includes a memory 310 and a processor 320 that are interconnected via a system bus. It should be noted that only a computer device with memory 310 and processor 320 is shown in the figure; however, it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented alternatively. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0063] Computer devices can include desktop computers, laptops, handheld computers, and cloud servers. These devices allow for human-computer interaction with users through keyboards, mice, remote controls, touchpads, or voice-activated devices.

[0064] The memory 310 includes at least one type of readable storage medium, including non-volatile memory or volatile memory, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. RAM may include static RAM or dynamic RAM. In some embodiments, the memory 310 may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the memory 310 may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, or flash card equipped on the computer device. Of course, the memory 310 may include both internal storage units and external storage devices of the computer device. In this embodiment, the memory 310 is typically used to store the operating system and various application software installed on the computer device, such as the program code of the method described above. In addition, the memory 310 can also be used to temporarily store various types of data that have been output or will be output.

[0065] Processor 320 is typically used to perform overall operations of a computer device. In this embodiment, memory 310 is used to store program code or instructions, including computer operation instructions, and processor 320 is used to execute the program code or instructions stored in memory 310 or process data, such as program code that runs the methods described above.

[0066] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus system can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0067] Another embodiment of this application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads computer-readable program code stored in the computer-readable medium, enabling the processor to execute the functional actions specified in each step or combination of steps in the above method; and to generate means for implementing the functional actions specified in each block or combination of blocks in the block diagram.

[0068] Computer-readable media include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any suitable combination thereof, wherein the memory is used to store program code or instructions, the program code including computer operation instructions, and the processor is used to execute the program code or instructions of the above-described methods stored in the memory.

[0069] The definitions of memory and processor can be found in the description of the foregoing computer device embodiments, and will not be repeated here.

[0070] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0071] In the various embodiments of this application, the functional units or modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0072] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0073] In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" as described in this application does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims listing several means, several units of these means may be embodied by the same item of hardware. The use of "first," "second," and "third," etc., does not indicate any order and these words should be interpreted as names. Unless otherwise specified, the steps in the above embodiments should not be construed as limiting the order of execution.

[0074] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for inferring abnormal layers in a transparent membrane, characterized in that, include: Acquire multi-focal layer image data of abnormal targets in the same transparent film corresponding to different focal layers; Acquire multi-angle image data of anomalies in the same transparent membrane corresponding to different observation angles; In addition, it acquires multi-band image data of anomalous targets in the same transparent film corresponding to different bands; The multi-peak response characteristics of the anomalous target at different focal layers are analyzed based on the multi-focal layer image data; the multi-angle response characteristics of the anomalous target at different viewing angles are analyzed based on the multi-angle image data; and the band difference response characteristics of the anomalous target at different bands are analyzed based on the multi-band image data. Based on the multi-peak response characteristics, the multi-angle response characteristics, and the band difference response characteristics, the relative predicted probability of the abnormal target being located on the transparent film, the relative predicted probability of the abnormal target being located inside the transparent film, and the relative predicted probability of the abnormal target being located below the transparent film are determined. Based on the relative predicted probability that the anomalous target is located on the transparent membrane, the relative predicted probability that the anomalous target is located inside the transparent membrane, and the relative predicted probability that the anomalous target is located below the transparent membrane, the layer inference result of the anomalous target relative to the transparent membrane is determined.

2. The method according to claim 1, characterized in that, Based on the analysis of the multi-focal layer image data, the multi-peak response characteristics of the abnormal target at different focal layers are analyzed, including: Peak detection is performed on the multi-focal layer image data to obtain the significant peak positions of the abnormal target at different focal layers; Based on the significant peak positions of the abnormal target under different focal layers, calculate the peak width corresponding to each significant peak. Based on the significant peak positions of the anomalous target under different focal layers and the peak widths corresponding to each significant peak, the multi-peak response characteristics of the anomalous target under different focal layers are determined.

3. The method according to claim 1, characterized in that, Based on the analysis of the multi-angle image data, the multi-angle response characteristics of the abnormal target under different viewing angles are analyzed, including: Extract the highlight center drift, shadow direction change, halo expansion radius, and edge bright band offset of the abnormal target from the multi-angle image data under different viewing angles. Based on the amount of highlight center drift, shadow direction change, halo expansion radius, and edge bright band offset of the anomalous target under different viewing angles, the multi-angle response characteristics of the anomalous target under different viewing angles are determined.

4. The method according to claim 1, characterized in that, Based on the analysis of the multi-band image data, the band difference response characteristics of the abnormal target in different bands are analyzed, including: Extract the difference response information of the abnormal target and the background in different bands from the multi-band image data. The difference response information includes at least one of the following: intensity difference, contrast difference, poor edge visibility, and poor segmentation stability. Based on the difference in response information between the abnormal target and the background in different bands, the band difference response characteristics of the abnormal target in different bands are determined.

5. The method according to claim 1, characterized in that, Based on the multi-peak response characteristics, the multi-angle response characteristics, and the band difference response characteristics, the relative predicted probability of the anomalous target being located on the transparent film, the relative predicted probability of the anomalous target being located inside the transparent film, and the relative predicted probability of the anomalous target being located below the transparent film are determined, including: The multi-peak response features, the multi-angle response features, and the band difference response features are input into the stratigraphic inference model to obtain the initial prediction probability that the anomalous target is located on the transparent membrane, the initial prediction probability that the anomalous target is located inside the transparent membrane, and the initial prediction probability that the anomalous target is located below the transparent membrane. The initial prediction probabilities of the abnormal target being located on the transparent membrane, the initial prediction probabilities of the abnormal target being located inside the transparent membrane, and the initial prediction probabilities of the abnormal target being located below the transparent membrane are normalized to obtain the relative prediction probabilities of the abnormal target being located on the transparent membrane, the relative prediction probabilities of the abnormal target being located inside the transparent membrane, and the relative prediction probabilities of the abnormal target being located below the transparent membrane.

6. The method according to claim 1, characterized in that, Based on the relative predicted probability of the anomalous target being located on the transparent membrane, the relative predicted probability of the anomalous target being located inside the transparent membrane, and the relative predicted probability of the anomalous target being located below the transparent membrane, the layer inference result of the anomalous target relative to the transparent membrane is determined, including: From the relative predicted probability of the anomalous target being located on the transparent membrane, the relative predicted probability of the anomalous target being located inside the transparent membrane, and the relative predicted probability of the anomalous target being located below the transparent membrane, the layer corresponding to the maximum predicted probability is determined as the main layer. The primary layer is determined to be the layer in which the abnormal target is located within the transparent membrane.

7. The method according to claim 6, characterized in that, Before determining the principal layer as the layer in which the anomalous target is located within the transparent membrane, the method further includes: If there are at least two layers corresponding to the maximum predicted probability, then based on the preset layer priority, the layer with higher priority is selected as the main layer from the at least two layers corresponding to the maximum predicted probability.

8. A device for inferring abnormal layers in a transparent membrane, characterized in that, include: The acquisition module is used to acquire multi-focal layer image data of abnormal targets in the same transparent film corresponding to different focal layers; Acquire multi-angle image data of anomaly targets in the same transparent film corresponding to different observation angles; and acquire multi-band image data of anomaly targets in the same transparent film corresponding to different spectral bands; The analysis module is used to analyze the multi-peak response characteristics of the anomalous target at different focal layers based on the multi-focal layer image data; to analyze the multi-angle response characteristics of the anomalous target at different viewing angles based on the multi-angle image data; and to analyze the band difference response characteristics of the anomalous target at different bands based on the multi-band image data. The first determining module is used to determine, based on the multi-peak response characteristics, the multi-angle response characteristics, and the band difference response characteristics, the relative predicted probability that the abnormal target is located on the transparent film, the relative predicted probability that the abnormal target is located inside the transparent film, and the relative predicted probability that the abnormal target is located below the transparent film; The second determining module is used to determine the layer inference result of the abnormal target relative to the transparent membrane based on the relative predicted probability that the abnormal target is located on the transparent membrane, the relative predicted probability that the abnormal target is located inside the transparent membrane, and the relative predicted probability that the abnormal target is located below the transparent membrane.

9. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the transparent membrane abnormal layer inference method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for inferring abnormal layers of transparent films as described in any one of claims 1 to 7.

Images