Titanium alloy forming defect detection method and system

Abstract

The present application relates to the technical field of image analysis, and in particular to a titanium alloy forming defect detection method and system, the method comprising: acquiring optical image data and thermal imaging image data of a titanium alloy surface during a hot rolling process of the titanium alloy; identifying a defect region in the optical image data, and determining a suspected defect region in the optical image data in combination with the thermal imaging image data; determining a real defect probability of the suspected defect region based on temperature changes of a corresponding region of the suspected defect region in the thermal imaging image data; and determining a defect type of the suspected defect region based on the real defect probability. The present application effectively improves defect detection accuracy by simultaneously combining color features in optical images and temperature change features in thermal imaging image data for titanium alloy surface defect detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis, in particular to a titanium alloy forming defect detection method and system. BACKGROUND

[0002] Titanium alloy is widely used in high-end manufacturing fields such as aero-engine blades and aerospace structural parts due to its high strength, low density and excellent corrosion resistance. However, during the forming process of titanium alloy hot-rolled plates, surface cracks, inclusions, and scale indentation defects are prone to occur due to complex material plastic deformation, severe high-temperature oxidation, and fluctuations in process parameters. Taking an aero-engine blade as an example, the service environment has a very low tolerance for internal defects of the material. Even a 0.2mm level crack can cause catastrophic failure. Therefore, the accuracy and real-time performance of defect detection are crucial to ensure the reliability and safety of equipment.

[0003] There are various existing titanium alloy forming defect detection methods: ultrasonic detection detects internal defects through high-frequency sound wave reflection; eddy current detection identifies defects based on electromagnetic induction principles through conductivity differences; infrared thermal imaging locates defects by monitoring differences in material heat dissipation rates, for example, crack areas have abnormal temperature drop rates due to increased thermal resistance; machine vision detection analyzes surface color or texture abnormalities based on RGB images, such as dark spots of scale and linear features of cracks.

[0004] However, the existing detection methods have the following defects: single detection parameter: relying only on a single parameter such as color, temperature, or electromagnetic characteristics, which cannot exclude environmental interference (such as oxidation stains and hot rolling temperature fluctuations), resulting in a false detection rate of >10%; unable to verify synchronously: traditional step-by-step detection requires switching equipment, which cannot achieve real-time comparison of data at the same location before and after hot rolling; defect type confusion: scale indentation and internal cracks have similar visual features, and a single detection method cannot distinguish between them, requiring manual re-inspection. SUMMARY

[0005] To solve the technical problem of relying only on a single parameter such as color, temperature, or electromagnetic characteristics for titanium alloy forming defect detection, resulting in a large error rate, the purpose of the present application is to provide a titanium alloy forming defect detection method and system, and the technical solution adopted is as follows:

[0006] In a first aspect, the present application provides a titanium alloy forming defect detection method, comprising the following steps:

[0007] Obtaining optical image data and thermal imaging image data of the titanium alloy surface during the titanium alloy hot rolling process;

[0008] identify a defect region in the optical image data, and in combination with the thermal image data, determine a suspected defect region in the optical image data;

[0009] determine a real defect probability of the suspected defect region based on temperature variation of a corresponding region of the suspected defect region in the thermal image data;

[0010] determine a defect type of the suspected defect region based on the real defect probability.

[0011] In combination with the first aspect, in some possible implementation manners, determining the suspected defect region in the optical image data includes:

[0012] identify a defect region in a target frame optical image in the optical image data, and determine a color difference region;

[0013] determine a plurality of target frame thermal images of the target frame optical image in the thermal image data, the target frame thermal images corresponding to a same titanium alloy surface region as the target frame optical image;

[0014] determine a local temperature difference between a corresponding region of the color difference region and a normal region in the target frame thermal image;

[0015] when any local temperature difference is greater than a set local temperature difference threshold, determine the corresponding color difference region as a suspected defect region.

[0016] In combination with the first aspect, in some possible implementation manners, determining the color difference region includes:

[0017] convert the target frame optical image to a CIELAB color space to obtain a LAB image;

[0018] determine a pixel mean value of each channel of the LAB image;

[0019] based on a difference between a pixel value of each pixel point in the LAB image in each channel and a corresponding pixel mean value, determine a color difference value of each pixel point in the LAB image;

[0020] based on the color difference value, perform threshold segmentation on the LAB image to obtain the color difference region.

[0021] In combination with the first aspect, in some possible implementation manners, determining the local temperature difference between the corresponding region of the color difference region and the normal region in the target frame thermal image includes:

[0022] determine a reference region of a rolling region in which the corresponding region of the color difference region is located in the target frame thermal image;

[0023] regard the region in the reference region except the region corresponding to the color difference region as a reference normal region;

[0024] determine a mean value of pixel values of all pixel points in the region corresponding to the color difference region in the target frame thermal imaging image, to obtain a color difference region temperature pixel;

[0025] determine a mean value of pixel values of all pixel points in the reference normal region, to obtain a normal region temperature pixel;

[0026] determine a local temperature difference between the color difference region and the normal region based on a difference between the color difference region temperature pixel and the normal region temperature pixel.

[0027] In combination with the first aspect, in some possible implementation manners, determining the real defect probability of the suspected defect region comprises:

[0028] determining a first defect probability of the suspected defect region based on a local temperature difference size of the suspected defect region in the plurality of target frame thermal imaging images and a difference in local temperature difference in adjacent target frame thermal imaging images;

[0029] determining a second defect probability of the suspected defect region based on a difference in color difference region temperature pixel of the suspected defect region in adjacent target frame thermal imaging images and a change trend of local temperature difference in the plurality of target frame thermal imaging images;

[0030] determining the real defect probability of the suspected defect region based on the first defect probability and the second defect probability.

[0031] In combination with the first aspect, in some possible implementation manners, determining the first defect probability of the suspected defect region comprises:

[0032] determining an accumulated value of local temperature difference of the suspected defect region in the plurality of target frame thermal imaging images, to obtain a local temperature difference accumulated value;

[0033] determining a mean value of difference in local temperature difference of the suspected defect region in each two adjacent target frame thermal imaging images, to obtain a local temperature difference difference mean value;

[0034] determining the first defect probability of the suspected defect region based on the local temperature difference accumulated value and the local temperature difference difference mean value.

[0035] In combination with the first aspect, in some possible implementation manners, determining the second defect probability of the suspected defect region comprises:

[0036] determining color difference region temperature change stability based on a difference value of color difference region temperature pixel of the suspected defect region in each two adjacent target frame thermal imaging images.

[0037] determine a frame thermal image with the most abnormal local temperature difference trend based on the local temperature difference trend of the suspected defect region in the target frame thermal images;

[0038] determine a local temperature difference retention degree based on the difference between the local temperature difference of the suspected defect region in the frame thermal image to be analyzed and the local temperature difference in other target frame thermal images;

[0039] determine a second defect probability of the suspected defect region based on the color difference region temperature change stability and the local temperature difference retention degree.

[0040] In some possible implementation manners, in combination with the first aspect, the determination of the frame thermal image with the most abnormal local temperature difference trend includes:

[0041] perform polynomial fitting on the local temperature difference of the suspected defect region in the target frame thermal images to obtain a fitted local temperature difference;

[0042] determine a maximum residual error from the maximum residual error and the corresponding fitted local temperature difference;

[0043] determine the target frame thermal image corresponding to the maximum residual error as the frame thermal image with the most abnormal local temperature difference trend.

[0044] In some possible implementation manners, in combination with the first aspect, the determination of the defect type of the suspected defect region includes:

[0045] if the real defect probability is greater than the high-risk probability threshold, it is determined that there is a high-risk defect corresponding to the suspected defect region;

[0046] if the real defect probability is less than or equal to the high-risk probability threshold and greater than the low-risk probability threshold, it is determined that there is a medium-risk defect corresponding to the suspected defect region;

[0047] otherwise, it is determined that there is no risk defect corresponding to the suspected defect region.

[0048] In a second aspect, the present application further provides a titanium alloy forming defect detection device, which includes:

[0049] an image data acquisition module configured to acquire optical image data and thermal image data of a titanium alloy surface in a titanium alloy hot rolling process;

[0050] a suspected defect region acquisition module configured to identify a defect region in the optical image data, and determine a suspected defect region in the optical image data in combination with the thermal image data;

[0051] a defect probability obtaining module configured to determine a real defect probability of the suspected defect region based on a temperature change of a corresponding region of the suspected defect region in the thermal imaging image data;

[0052] a defect type determining module configured to determine a defect type of the suspected defect region based on the real defect probability.

[0053] In a third aspect, the present application further provides a titanium alloy forming defect detection system, which comprises a memory and a processor. The memory is configured to store executable computer program code, and the processor is configured to call and run the executable computer program code from the memory, so that the system executes the method in the first aspect or any possible implementation manner of the first aspect.

[0054] In a fourth aspect, the present application further provides a computer program product, which comprises computer program code. When the computer program code is run on a computer, the computer program code causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0055] In a fifth aspect, the present application further provides a computer readable storage medium, which stores computer program code. When the computer program code is run on a computer, the computer program code causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0056] The present application has the following beneficial effects: by obtaining optical image data and thermal imaging image data of a titanium alloy surface in a titanium alloy hot rolling process, identifying a defect region in the optical image data, determining a suspected defect region in the optical image data in combination with the thermal imaging image data, determining a real defect probability of the suspected defect region based on a temperature change of a corresponding region of the suspected defect region in the thermal imaging image data, and determining a defect type of the suspected defect region based on the real defect probability, the present application obtains multi-modal image data of a titanium alloy surface in a titanium alloy hot rolling process, i.e. simultaneously obtains optical image data and thermal imaging image data, and identifies a suspected defect region in the optical image data and its real defect probability in combination with color features in the optical image and temperature change features in the thermal imaging image data, so as to finally realize titanium alloy forming defect detection and effectively improve defect detection precision. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0058] Figure 1 A step flow chart of a titanium alloy forming defect detection method according to an embodiment of the present application;

[0059] Figure 2 A step flow chart of determining a suspected defect area according to an embodiment of the present application;

[0060] Figure 3 A step flow chart of determining a real defect probability of a suspected defect area according to an embodiment of the present application;

[0061] Figure 4 A structural schematic diagram of a titanium alloy forming defect detection device according to an embodiment of the present application;

[0062] Figure 5 A structural schematic diagram of a titanium alloy forming defect detection system according to an embodiment of the present application. DETAILED DESCRIPTION

[0063] In order to clearly illustrate the technical features of the present application, the following will describe the present application in detail through specific embodiments and in conjunction with the drawings.

[0064] The embodiments of the present application will be described in more detail below with reference to the drawings. Although some embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms, and should not be interpreted as being limited to the embodiments described herein, but rather these embodiments are provided to more thoroughly and completely understand the present application. It should be understood that the drawings and embodiments of the present application are only for illustrative purposes, and are not intended to limit the scope of protection of the present application.

[0065] It should be understood that each step described in the method embodiments of the present application can be executed in different order and / or in parallel. In addition, the method embodiments can include additional steps and / or omit the execution of the steps shown. The scope of the present application is not limited in this respect.

[0066] The term "comprising" and variations thereof as used in the present application are open-ended, that is, "including but not limited to". The term "based on" is "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Related definitions of other terms will be given in the following description.

[0067] It should be noted that the terms "first", "second", and the like in the present application are only used to distinguish different devices, modules or units, and are not intended to limit the order or interdependence of the functions performed by these devices, modules or units.

[0068] Although the operations or steps in the embodiments of the present application are described in a particular order in the accompanying drawings, it should not be understood as requiring the operations or steps to be performed in the particular order or in a serial order, or requiring all of the operations or steps to be performed to obtain the desired results. In the embodiments of the present application, the operations or steps can be performed in series, in parallel, or a part of the operations or steps.

[0069] At the same time, it can be understood that the data involved in the technical solutions of the present application (including but not limited to the data itself, the acquisition or use of the data) should comply with the requirements of the corresponding laws, regulations and relevant provisions. Unless otherwise defined, all technical and scientific terms used in the present application have the same meaning as understood by those skilled in the art to which the present application belongs, and all parameters or indicators in the formulas involved in the present application are normalized values that eliminate the dimension influence.

[0070] In order to solve the problem of large error rate caused by relying on only a single parameter of color, temperature or electromagnetic characteristics for titanium alloy forming defect detection, the embodiments of the present application provide a titanium alloy forming defect detection method and system, which acquires optical image data and thermal imaging image data of the surface of titanium alloy in the hot rolling process of titanium alloy, and simultaneously combines color features in the optical image and temperature change features in the thermal imaging image data to detect surface defects of titanium alloy, effectively improving the defect detection precision.

[0071] In the following, a titanium alloy forming defect detection method and system provided by the embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0072] Figure 1 The basic flowchart of a titanium alloy forming defect detection method provided by the embodiments of the present application is shown in FIG. 1, which specifically includes the following steps: Figure 1

[0073] Step S100: acquiring optical image data and thermal imaging image data of the surface of titanium alloy in the hot rolling process of titanium alloy;

[0074] Step S200: identifying the defect area in the optical image data, and combining the thermal imaging image data to determine the suspected defect area in the optical image data;

[0075] ​Step S300: determining a real defect probability of the suspected defect area based on temperature changes of a corresponding area of the suspected defect area in the thermal imaging image data;

[0076] Step S400: determining a defect type of the suspected defect area based on the real defect probability.

[0077] The titanium alloy forming defect detection method provided by the embodiment of the present application realizes online accurate detection of titanium alloy forming defects by acquiring multi-modal image data of the surface of the titanium alloy in the hot rolling process of the titanium alloy and realizing the correlation analysis of the dynamic change characteristics of the temperature field and the optical color anomaly, and effectively improves the defect detection precision.

[0078] The specific steps of the titanium alloy forming defect detection method provided by the embodiment of the present application will be described in detail below.

[0079] Step S100: acquiring optical image data and thermal imaging image data of the surface of the titanium alloy in the hot rolling process of the titanium alloy.

[0080] Specifically, the surface temperature distribution of the titanium alloy after hot rolling has a certain relationship with internal defects, but single thermal imaging or optical detection cannot take into account high-temperature interference and surface details, therefore, in the embodiment of the present application, an equidistant thermal imaging camera array is deployed to continuously capture the dynamic evolution of the temperature field of the titanium alloy, and a high-resolution surface optical camera is deployed to collect high-resolution surface optical images, thereby making up for the defect of insufficient spatial resolution of thermal imaging, and the cooperation of the two types of image data can ensure the spatio-temporal alignment of the data and cover the full area without blind area detection.

[0081] In a specific example, a plurality of equidistantly distributed thermal imaging cameras are deployed in the conveying belt area of the titanium alloy hot rolling production line. At the end of the whole row of thermal imaging cameras (i.e. the last thermal imaging camera), a conventional optical camera is synchronously installed to ensure that the collection range of the optical camera is consistent with the collection range of the last thermal imaging camera at the end. The installation interval between any two adjacent thermal imaging cameras is set as L, and the sampling time interval of the thermal imaging camera is determined as t0 (in the case that the conveying belt speed is v, t0=L / v), so as to ensure that the shooting areas of the adjacent thermal imaging cameras are continuous and have no overlap. The models of the thermal imaging camera and the optical camera can be reasonably selected according to the needs, for example, the thermal imaging camera is selected as a medium-wave infrared thermal imager (FLIR A8300sc), because the medium-wave band (3-5 μm) is more sensitive to the radiation of high-temperature titanium alloy, and the temperature measurement accuracy is ±1℃; the optical camera is selected as a high-dynamic-range industrial camera (Basler ace 2), which has a wide dynamic range (>120 dB) and can overcome strong light reflection and retain bright and dark details.

[0082] In the hot rolling process of the titanium alloy, the thermal imaging camera and the optical camera installed above are used to collect thermal imaging images and optical images (i.e., RGB images) of a surface area of the titanium alloy, so as to obtain optical image data and thermal imaging image data of the surface of the titanium alloy in the hot rolling process of the titanium alloy.

[0083] Step S200: identifying a defect area in the optical image data, and determining a suspected defect area in the optical image data in combination with the thermal imaging image data.

[0084] Specifically, the titanium alloy defect area causes abnormal heat conduction, resulting in temperature distribution difference, and surface deformation causes change of optical reflection characteristics, so that the abnormal color area is located by the end optical image, and the temperature evolution rule thereof in the thermal imaging image data is traced back, so that the real defect and the oxide scale and other interference can be distinguished, and the accuracy of defect identification is improved.

[0085] Further, in a possible implementation manner, as shown in Figure 2 the suspected defect area is determined, including:

[0086] Step S201: identifying a defect area in a target frame optical image in the optical image data, and determining a color difference area.

[0087] Specifically, the target frame optical image (i.e., any one frame of optical image) in the optical image data is subjected to image preprocessing, such as non-local mean filtering, to reduce random noise in a high-temperature environment, so as to obtain the target frame optical image after image preprocessing. Further, the defect area is identified in the target frame optical image after image preprocessing, and the color difference area possibly being a defect area is determined.

[0088] Further, in a possible implementation manner, the color difference area is determined, including: converting the target frame optical image to a CIELAB color space to obtain a LAB image; determining pixel mean values of each channel of the LAB image; determining a color difference value of each pixel point in the LAB image based on the difference between the pixel value of each pixel point in the LAB image in each channel and the corresponding pixel mean value; and performing threshold segmentation on the LAB image based on the color difference value to obtain the color difference area.

[0089] In a specific example, the target frame optical image (RGB image) is converted to a CIELAB color space to obtain a LAB image, and pixel mean values of the LAB image in the L channel, the A channel and the B channel in the CIELAB color space are calculated and are denoted as the L channel pixel value L i , the A channel pixel value A i and the B channel pixel value B iComparing with the pixel mean value of the corresponding channel, the color difference value of any ith pixel point in the LAB image is calculated

[0090] Further, based on the color difference value ΔE of each pixel point in the LAB image, further color difference region extraction, i.e. image binary segmentation using adaptive threshold k, when ΔE>k exists, the corresponding pixel point is determined as a pixel point in the color difference region, when ΔE≤k exists, the corresponding pixel point is determined as a pixel point in the normal region. The continuous region composed of the pixel points in the color difference region is determined, and the hole filling processing in the continuous region is performed by using the morphological closing operation method, so as to finally obtain the color difference region in the target frame optical image.

[0091] Step S202: determining a plurality of target frame thermal imaging images of the target frame optical image in the thermal imaging image data, the target frame thermal imaging image corresponding to the same titanium alloy surface region as the target frame optical image.

[0092] Specifically, when determining the titanium alloy forming surface defect according to the color difference region in the optical image, it may be affected by interference factors such as surface oxide skin, resulting in misjudgment. In order to more accurately verify the possibility of defects, the series of thermal imaging images previously shot can be traced back, and the temperature difference at the corresponding position is observed. If abnormal temperature change is also found in the thermal imaging image, it can be further confirmed that the region indeed has defects.

[0093] In a specific example, in the thermal imaging image data collected by 9 thermal imaging cameras, 9 thermal imaging images corresponding to the same titanium alloy surface region as the target frame optical image are determined, and the 9 thermal imaging images are taken as the target frame thermal imaging images of the target frame optical image. For example, when the collection time corresponding to the target frame optical image is t, the thermal imaging image collected by the terminal thermal imaging camera at time t is taken as one frame of target frame thermal imaging image, the thermal imaging image collected by the thermal imaging camera before the terminal thermal imaging camera at time t-t0 is taken as one frame of target frame thermal imaging image, and the thermal imaging image collected by the thermal imaging camera before the terminal thermal imaging camera at time t-2×t0 is taken as one frame of target frame thermal imaging image, and so on, so that 9 target frame thermal imaging images of the target frame optical image can be obtained.

[0094] Step S203: determining the local temperature difference of the color difference region corresponding region and the normal region in the target frame thermal imaging image.

[0095] Specifically, the chromatic aberration region in the target optical image is mapped to the target frame thermal imaging image to obtain a corresponding region of the chromatic aberration region. In a specific example, the chromatic aberration region in the target optical image is extracted, and a geometric center point A of the chromatic aberration region is determined. Then, the coordinate position of the point A is accurately determined in the 9 target frame thermal imaging images. Then, the chromatic aberration region in the 9 target frame thermal imaging images is located based on the coordinate position of the point A, so as to ensure that the chromatic aberration region in the target optical image is accurately matched with the corresponding point coordinates in each target frame thermal imaging image. Thus, the corresponding region of the chromatic aberration region can be determined in all target frame thermal imaging images.

[0096] In the target frame thermal imaging image, the corresponding region of the chromatic aberration region and the normal region are compared in temperature. At this time, since the titanium alloy has poor thermal conductivity, there is a relatively obvious difference in temperature in the entire image, thereby causing a large local temperature difference between the corresponding region of the chromatic aberration region and the normal region.

[0097] Further, in a possible implementation, the local temperature difference between the corresponding region of the chromatic aberration region and the normal region in the target frame thermal imaging image includes: determining a reference region of a rolling region where the corresponding region of the chromatic aberration region is located in the target frame thermal imaging image; taking a region other than the corresponding region of the chromatic aberration region in the reference region as a reference normal region; determining a mean value of pixel values of all pixel points in the corresponding region of the chromatic aberration region in the target frame thermal imaging image to obtain a chromatic aberration region temperature pixel; determining a mean value of pixel values of all pixel points in the reference normal region to obtain a normal region temperature pixel; and determining the local temperature difference between the corresponding region of the chromatic aberration region and the normal region based on the difference between the chromatic aberration region temperature pixel and the normal region temperature pixel.

[0098] In a specific example, in any target frame thermal imaging image, a normal region of a region rolled at the same time as the chromatic aberration region can be selected for temperature comparison. In the target frame thermal imaging image, a two-dimensional coordinate system is constructed with a rolling direction (i.e., a conveying belt transmission direction) as a horizontal coordinate and a vertical direction of the rolling direction as a vertical coordinate. At this time, the region rolled at the same time as the chromatic aberration region appears as a same horizontal coordinate region in the image. Therefore, first, the maximum value X max and the minimum value X min of the horizontal coordinate of the current corresponding region of the chromatic aberration region are determined in the target frame thermal imaging image. Then, two straight lines parallel to the vertical axis, x=X max and x=X min are determined. Then, a region between the two straight lines is taken as a reference region of a rolling region where the corresponding region of the chromatic aberration region is located. Finally, a region other than the corresponding region of the chromatic aberration region in the reference region is taken as a reference normal region, which maintains the same rolling sequence as the chromatic aberration region.

[0099] In the thermal imaging image of any target frame, average value calculation is performed on the temperature of all pixels in the region corresponding to the μth chromatic aberration region to obtain the chromatic aberration region temperature pixel T μ,1 Meanwhile, average value calculation is performed on the temperature of all pixels in the reference normal region corresponding to the μth chromatic aberration region to obtain the normal region temperature pixel T μ,2 The absolute value of the difference between the chromatic aberration region temperature pixel T μ,1 and the normal region temperature pixel T μ,2 is calculated to obtain the local temperature difference ΔT μ between the chromatic aberration region and the normal region, where the local temperature difference ΔT μ = |T μ,1 -T μ,2 |.

[0100] In the above manner, the local temperature difference between the region corresponding to the chromatic aberration region and the normal region in the thermal imaging image of any target frame can be determined.

[0101] Step S204: When any local temperature difference is greater than a set local temperature difference threshold, the corresponding chromatic aberration region is determined as a suspected defect region.

[0102] Specifically, a local temperature difference threshold γ is set in advance, which needs to be determined according to actual working conditions. The local temperature difference ΔT μ between the region corresponding to the chromatic aberration region and the normal region is compared with the local temperature difference threshold γ. When there is a local temperature difference ΔT μ > γ in the thermal imaging image of any target frame, the corresponding chromatic aberration region is determined as a suspected defect region, otherwise it is indicated that the corresponding chromatic aberration region is likely to be interference of the oxide scale, and no further analysis is performed.

[0103] The above step S200 determines the chromatic aberration region in the optical image data, maps the chromatic aberration region to the thermal imaging image data, determines the local temperature difference between the region corresponding to the chromatic aberration region and the normal region, and then accurately identifies the suspected defect region based on the local temperature difference, thereby avoiding the interference of the oxide scale.

[0104] Step S300: Based on the temperature change of the suspected defect region in the thermal imaging image data, the real defect probability of the suspected defect region is determined.

[0105] Specifically, in the process of hot rolling and cooling of titanium alloy, the temperature change trend is a key factor for evaluating the possibility of defects. Under standard cooling conditions, the temperature field should follow a preset uniform downward path to decrease smoothly. However, if there are defects such as micro-cracks or uneven microstructure inside the material, the temperature change trend of these areas will show significant differences from normal areas. For example, the cooling rate of the crack area may be lower than that of the surrounding normal area due to the obstruction of heat conduction, resulting in an abnormal lag or fluctuation in the temperature drop trend. Based on this, by analyzing the temperature change of the suspected defect area in the thermal imaging image data, the probability of the suspected defect area being a real defect can be inferred.

[0106] Further, in a possible implementation, as shown in Figure 3 the probability of a real defect of the suspected defect area is determined, including:

[0107] Step S301: determining a first defect probability of the suspected defect area based on the local temperature difference size of the suspected defect area in the thermal imaging images of the plurality of target frames and the difference of the local temperature difference in the adjacent thermal imaging images of the target frames.

[0108] Specifically, in the process of hot rolling and cooling of titanium alloy, the local temperature difference of the actual defect area should persistently exist. When the suspected defect area continuously maintains a high local temperature difference, and the trend of the local temperature difference is worse, i.e., the local temperature difference changes greatly, the probability of being a real defect is higher. Therefore, by analyzing the local temperature difference size of each suspected defect area in the thermal imaging images of the corresponding plurality of target frames and the difference of the local temperature difference in the adjacent thermal imaging images of the target frames, the first defect probability of the suspected defect area can be determined.

[0109] Further, in a possible implementation, the first defect probability of the suspected defect area is determined, including: determining an accumulated value of the local temperature difference of the suspected defect area in the thermal imaging images of the plurality of target frames to obtain a local temperature difference cumulative value; determining a mean value of the difference of the local temperature difference of the suspected defect area in each two adjacent thermal imaging images of the target frames to obtain a local temperature difference difference mean value; and determining the first defect probability of the suspected defect area based on the local temperature difference cumulative value and the local temperature difference difference mean value.

[0110] In a specific example, for any μth suspected defect region in a target frame optical image (i.e. any one frame of optical image) in the optical image data, the local temperature difference of the μth suspected defect region in all target frame thermal imaging images of the target frame optical image can be determined, and the local temperature differences are arranged in the order of the acquisition time of the target frame thermal imaging images, thereby a local temperature difference sequence can be determined. The greater the sum of all local temperature differences in the local temperature difference sequence, the higher the probability of the μth suspected defect region being a real defect. Therefore, the cumulative sum of all local temperature differences in the local temperature difference sequence is calculated, and the cumulative sum is taken as the local temperature difference cumulative value. At the same time, the absolute value of the difference between each local temperature difference and the next local temperature difference in the local temperature difference sequence is determined, the local temperature difference difference is obtained, and the average of all local temperature difference differences is obtained, thereby obtaining the local temperature difference difference average. The local temperature difference cumulative value and the local temperature difference difference average are fused in a positive direction, that is, the product of the local temperature difference cumulative value and the local temperature difference difference average is calculated, and the product is normalized to [0, 1] by using the maximum and minimum value normalization method, thereby obtaining the first defect probability P1 of the μth suspected defect region.

[0111] The step S301 determines the local temperature difference cumulative value and the local temperature difference difference average by analyzing the size of the local temperature difference of each suspected defect region in several target frame thermal imaging images and the difference of the local temperature difference in adjacent target frame thermal imaging images, and then based on the local temperature difference cumulative value and the local temperature difference difference average, the first defect probability of the suspected defect region can be accurately determined.

[0112] Step S302: determining the second defect probability of the suspected defect region based on the difference of the temperature pixels of the color difference region of the suspected defect region in adjacent target frame thermal imaging images and the change trend of the local temperature difference in several target frame thermal imaging images.

[0113] Specifically, the temperature change of the suspected defect region in the cooling process is analyzed. Among all the thermal imaging images, the more stable the temperature change of the suspected defect region, the higher the probability of being a real defect. At the same time, the local temperature difference retention of the suspected defect region in the cooling process is analyzed. In the cooling process of titanium alloy, if the temperature change rate of the suspected defect region at a certain time deviates seriously, and the change of the local temperature difference of the thermal imaging taken at the corresponding time is also relatively large, then the probability of the suspected defect region being a real defect is higher. Therefore, by analyzing the difference of the temperature pixels of the color difference region of each suspected defect region in adjacent target frame thermal imaging images and the change trend of the local temperature difference in all target frame thermal imaging images and the difference of the local temperature difference in different target frame thermal imaging images, the second defect probability of the suspected defect region can be determined.

[0114] Further, in a possible implementation, the second defect probability of the suspected defect region is determined by: determining a color difference region temperature variation stability based on a difference value of color difference region temperature pixels of the suspected defect region in each two adjacent target frame thermal imaging images; determining a target frame thermal imaging image with most abnormal local temperature difference trend based on a change trend of local temperature difference of the suspected defect region in the target frame thermal imaging images; determining a local temperature difference retention degree based on a difference between the local temperature difference of the suspected defect region in the target frame thermal imaging image and the local temperature difference in other target frame thermal imaging images; and determining the second defect probability of the suspected defect region based on the color difference region temperature variation stability and the local temperature difference retention degree.

[0115] Further, in a possible implementation, the target frame thermal imaging image with most abnormal local temperature difference trend is determined by: performing polynomial fitting on the local temperature difference of the suspected defect region in the target frame thermal imaging images to obtain a fitted local temperature difference; determining a maximum value in residuals of all local temperature differences and corresponding fitted local temperature differences to obtain a maximum residual; and determining the target frame thermal imaging image corresponding to the maximum residual as the target frame thermal imaging image with most abnormal local temperature difference trend.

[0116] In a specific example, for any μth suspected defect region in a target frame optical image (i.e., any frame of optical image) in the optical image data, a color difference region temperature pixel Tμ of the μth suspected defect region in all target frame thermal imaging images of the target frame optical image can be determined. μ,1 The color difference region temperature pixel Tμ of the μth suspected defect region in the target frame thermal imaging image is determined by: determining a color difference region temperature pixel Tμ of the μth suspected defect region in the target frame thermal imaging image based on a color difference region temperature pixel Tμ of the μth suspected defect region in the target frame optical image. μ,1 The color difference region temperature pixels are arranged in the order of the acquisition time of the target frame thermal imaging images, so that a color difference region temperature pixel sequence can be determined. The sum of squares of the difference between each color difference region temperature pixel and the next color difference region temperature pixel in the color difference region temperature pixel sequence is calculated, and the reciprocal of the sum of all squares is obtained, so that the color difference region temperature variation stability U is obtained. When the value of the color difference region temperature variation stability U is larger, the stability of the temperature variation is higher, and the probability that the μth suspected defect region is a real defect is larger.

[0117] Meanwhile, based on the local temperature difference sequence corresponding to the μth suspected defect region and the acquisition time of the target frame thermal imaging image corresponding to each local temperature difference in the local temperature difference sequence, polynomial fitting is performed on the local temperature difference sequence changing with the acquisition time to obtain any local temperature difference ΔTμ of the local temperature difference sequence. μ The corresponding fitted local temperature difference ΔT′μ of the μth suspected defect region is determined by: determining a fitted local temperature difference ΔT′μ of the μth suspected defect region based on the polynomial fitting. μ For the μth suspected defect region, if the local temperature difference ΔTμ in a certain ith target frame thermal imaging image deviates from the theoretical fitted local temperature difference ΔT′μ after the polynomial fitting, μ,i the probability that the μth suspected defect region is a real defect is determined by: determining a deviation degree of the local temperature difference ΔTμ in the ith target frame thermal imaging image from the theoretical fitted local temperature difference ΔT′μ after the polynomial fitting; and determining the probability that the μth suspected defect region is a real defect based on the deviation degree. μThe larger the value (the larger the regression difference), the greater the local temperature difference ΔT in the thermal imaging image of the target frame acquired at the corresponding time. μ,i Local temperature difference ΔT between the thermal imaging image of the j-th target frame and any other target frame μ,j The greater the difference, the higher the local temperature difference retention of the μth suspected defect region, and the higher the probability that the μth suspected defect region is a real defect. Therefore, the local temperature difference ΔT of any i-th target frame thermal imaging image is calculated. μ,i The corresponding fitted local temperature difference ΔT′ μ,i The absolute value of the difference between them is used to obtain the residual, and then the maximum residual R among the residuals corresponding to all target frame thermal imaging images is found. max And the maximum residual R max The corresponding target frame thermal imaging image is taken as the thermal imaging image of the frame to be analyzed, where the local temperature difference trend is most abnormal. The local temperature difference in this thermal imaging image is denoted as maxΔT. μ And calculate the maxΔT μ Local temperature difference ΔT in thermal images of other target frames μ The absolute value of the difference is used to obtain the local temperature difference. The sum of all local temperature difference differences is calculated to obtain the local temperature difference retention degree F. The larger the value of the local temperature difference retention degree F, the higher the degree of local temperature difference retention, and the greater the probability that the μ-th suspected defect area is a real defect.

[0118] Furthermore, the temperature change stability U and local temperature difference retention F of the color difference region corresponding to the μth suspected defect region are positively integrated. That is, the product of the temperature change stability U and the local temperature difference retention F of the color difference region is calculated, and the product is normalized to the range [0,1] using the maximum and minimum value normalization method, thereby obtaining the second defect probability P2 of the μth suspected defect region.

[0119] Step S302 above analyzes the temperature change and local temperature difference retention of the suspected defect area during the cooling process to determine the temperature change stability and local temperature difference retention of the color difference area corresponding to the suspected defect area, thereby accurately determining the second defect probability of the suspected defect area.

[0120] Step S303: Based on the first defect probability and the second defect probability, determine the actual defect probability of the suspected defect area.

[0121] Specifically, for any μ-th suspected defect region, the true defect probability Q of that μ-th suspected defect region is determined by combining its corresponding first defect probability P1 and second defect probability P2. In a specific example, the true defect probability Q of that μ-th suspected defect region is calculated using the following formula:

[0122] Q = ω1P1 + ω2P2;

[0123] In the formula: ω1, ω2 are weight coefficients, which are determined by the relative size of the residual cumulative sum obtained by polynomial fitting of the local temperature difference sequence and the temperature pixel sequence of the color difference area, if the residual cumulative sum obtained by polynomial fitting of the local temperature difference sequence is R1, and the residual cumulative sum obtained by polynomial fitting of the temperature pixel sequence of the color difference area is R2, then: ω2 = 1 - ω1.

[0124] The above step S300 determines the first defect probability and the second defect probability of the suspected defect area by analyzing the temperature change of the corresponding area of the suspected defect area in the thermal imaging image data, and then calculates the probability Q that the suspected defect area is a real defect.

[0125] Step S400: determining the defect type of the suspected defect area based on the real defect probability.

[0126] Specifically, according to the real defect probability of each suspected defect area, the defect type corresponding to each suspected defect area is determined. The greater the value of the real defect probability, the more likely it is that the corresponding suspected defect area is a defect area with high defect risk.

[0127] Further, in a possible implementation, determining the defect type of the suspected defect area includes: if the real defect probability is greater than a high-risk probability threshold, it is determined that the corresponding suspected defect area exists a high-risk defect; if the real defect probability is less than or equal to the high-risk probability threshold and greater than a low-risk probability threshold, it is determined that the corresponding suspected defect area exists a medium-risk defect; otherwise, it is determined that the corresponding suspected defect area does not exist a risk defect. The specific values of the high-risk probability threshold and the low-risk probability threshold need to be determined according to actual conditions.

[0128] In a specific example, determining the defect type of the suspected defect area includes:

[0129] First, defect threshold calibration is performed:

[0130] Based on the real defect probability Q value distribution of the historical qualified samples and the defect samples, the best segmentation threshold Q is determined by the ROC curve crit . For example, when Q>0.7, the defect probability exceeds 95%, then 0.7 is set as the high-risk probability threshold; when 0.4<Q≤0.7, it is a medium risk, and when Q≤0.4, it is a low risk, then 0.4 is set as the low-risk probability threshold.

[0131] Second, defect type mapping is performed:

[0132] High-risk area (Q>0.7): There are high-risk defects in the corresponding area, that is, there may be cracks, inclusions and other serious defects, which need to be handled urgently.

[0133] Medium-risk area (0.4<Q≤0.7): There are medium-risk defects in the corresponding area, that is, there may be scale indentation or micro-cracks, which need to be adjusted locally.

[0134] Low-risk area (Q≤0.4): There are no risk defects in the corresponding area, which belongs to the normal production area and needs to be monitored.

[0135] Finally, real-time defect type determination and process control:

[0136] Based on the real defect probability Q of the suspected defect area in the optical image data obtained in real time during the hot rolling process of the titanium alloy, the real defect probability Q is compared with the high-risk probability threshold and the low-risk probability threshold determined by the above defect threshold calibration, the specific defect type of the suspected defect area is determined, and the risk area with high-risk defects and medium-risk defects is marked, and the titanium alloy hot rolling control system is triggered to adjust the rolling parameters. Subsequently, based on the real defect probability Q of the suspected defect area in the optical image data obtained in real time after the rolling parameter adjustment, it is verified whether it returns to the safety threshold. If the real defect probability Q value is continuously high, a secondary alarm is triggered and manual intervention is notified.

[0137] Based on the same inventive concept, the embodiments of the present application also provide a titanium alloy forming defect detection device, as shown in Figure 4 The device comprises:

[0138] An image data acquisition module for acquiring optical image data and thermal imaging image data of the surface of the titanium alloy during the hot rolling process of the titanium alloy;

[0139] A suspected defect area acquisition module for identifying the defect area in the optical image data and determining the suspected defect area in the optical image data in combination with the thermal imaging image data;

[0140] A defect probability acquisition module for determining the real defect probability of the suspected defect area based on the temperature change of the corresponding area of the suspected defect area in the thermal imaging image data;

[0141] A defect type determination module for determining the defect type of the suspected defect area based on the real defect probability.

[0142] It should be noted that: the device provided in the above embodiment is only used for example to divide the above functional modules, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the titanium alloy forming defect detection method provided in the above embodiment belongs to the same concept, and its specific implementation process is described in the method embodiment, which will not be repeated here.

[0143] Based on the same inventive concept, the embodiments of the present application also provide a titanium alloy forming defect detection system, as shown in the system includes: a memory 501, a processor 502, and computer program code 503 stored in the memory 501 and running on the processor 502, wherein the processor 502 executes the computer program code 503, so that the system can execute any one of the titanium alloy forming defect detection methods described above. Figure 5

[0144] The embodiments of the present application can divide the system into functional modules according to the above method examples, for example, each functional module can be corresponding, or two or more functions can be integrated in one processing module, and the integrated module can be realized in the form of hardware. It should be noted that the division of modules in this embodiment is illustrative, and is only a logical function division, and another division method can be used in actual implementation.

[0145] Based on the same inventive concept, the embodiments of the present application also provide a computer program product, which includes: computer program code, when the computer program code runs on a computer, so that the computer executes any one of the titanium alloy forming defect detection methods described above.

[0146] Based on the same inventive concept, the embodiments of the present application also provide a computer readable storage medium, which stores computer program code, when the computer program code runs on a computer, so that the computer executes any one of the titanium alloy forming defect detection methods described above.

[0147] It should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.​

Claims

1. A method for detecting forming defects in titanium alloys, characterized in that, Includes the following steps: Acquire optical and thermal imaging data of the titanium alloy surface during the hot rolling process; Defect areas in the optical image data are identified, and combined with the thermal imaging image data, suspected defect areas in the optical image data are determined. Based on the temperature change of the corresponding area of ​​the suspected defect area in the thermal imaging image data, the actual defect probability of the suspected defect area is determined. Based on the actual defect probability, the defect type of the suspected defect area is determined; Identifying suspected defect areas in the optical image data includes: Defect regions in the target frame optical image of the optical image data are identified to determine the color difference regions; Several target frame thermal imaging images are determined from the thermal imaging image data, wherein the target frame thermal imaging images correspond to the same titanium alloy surface area as the target frame optical image; In the thermal imaging image of the target frame, determine the local temperature difference between the region corresponding to the color difference region and the normal region; When any local temperature difference exceeds the set local temperature difference threshold, the corresponding color difference area is identified as a suspected defect area. Determining the local temperature difference between the region corresponding to the color difference region and the normal region in the target frame thermal imaging image includes: In the target frame thermal imaging image, determine the reference area of ​​the rolling region where the color difference region corresponds to the region; The area in the reference region other than the area corresponding to the color difference region is taken as the reference normal region; The average pixel value of all pixels in the region corresponding to the color difference region in the thermal imaging image of the target frame is determined to obtain the temperature pixel of the color difference region. The average pixel value of all pixels in the reference normal region is determined to obtain the temperature pixel of the normal region. Based on the difference between the temperature pixels in the color difference region and the temperature pixels in the normal region, the local temperature difference between the color difference region and the normal region is determined; Determining the actual defect probability of the suspected defect area includes: Based on the magnitude of the local temperature difference of the suspected defect area in the thermal imaging images of several target frames, and the difference in the local temperature difference in the thermal imaging images of adjacent target frames, the first defect probability of the suspected defect area is determined. Based on the difference in temperature pixels of the color difference region in the suspected defect region in the thermal imaging images of adjacent target frames, and the changing trend of local temperature difference in the thermal imaging images of several target frames, the second defect probability of the suspected defect region is determined. Based on the first defect probability and the second defect probability, the actual defect probability of the suspected defect area is determined.

2. The method for detecting forming defects in titanium alloys according to claim 1, characterized in that, Determine the color difference area, including: The optical image of the target frame is converted to the CIELAB color space to obtain a LAB image; Determine the average pixel value of the LAB image in each channel; Based on the difference between the pixel value of each pixel in each channel and the corresponding pixel mean in the LAB image, the color difference value of each pixel in the LAB image is determined; Based on the color difference value, the LAB image is thresholded to obtain the color difference region.

3. The method for detecting forming defects in titanium alloys according to claim 1, characterized in that, Determining the first defect probability of the suspected defect region includes: The cumulative value of the local temperature difference in the suspected defect area in the thermal imaging images of the target frames is determined to obtain the cumulative value of the local temperature difference; The mean difference in local temperature difference of the suspected defect area in each pair of adjacent target frame thermal imaging images is determined to obtain the mean difference in local temperature difference. Based on the cumulative value of the local temperature difference and the average value of the local temperature difference, the first defect probability of the suspected defect area is determined.

4. The method for detecting forming defects in titanium alloys according to claim 1, characterized in that, Determining the second defect probability of the suspected defect region includes: Based on the difference in temperature pixels of the color difference region in each two adjacent target frame thermal imaging images of the suspected defect area, the stability of the temperature change in the color difference region is determined. Based on the local temperature difference variation trend of the suspected defect area in the thermal imaging images of several target frames, the thermal imaging image of the frame to be analyzed with the most abnormal local temperature difference trend is determined. Based on the difference between the local temperature difference of the suspected defect area in the thermal imaging image of the frame to be analyzed and the local temperature difference in the thermal imaging image of other target frames, the local temperature difference retention is determined. Based on the temperature change stability of the color difference region and the local temperature difference retention, the second defect probability of the suspected defect region is determined.

5. The method for detecting forming defects in titanium alloys according to claim 4, characterized in that, Identify the thermal imaging frames to be analyzed that exhibit the most anomalous local temperature difference trends, including: The local temperature difference of the suspected defect area in the thermal imaging images of the target frames is fitted by a polynomial to obtain the fitted local temperature difference. The maximum value among the residuals of all the local temperature differences and the corresponding fitted local temperature differences is obtained to obtain the maximum residual; The thermal imaging image of the target frame corresponding to the maximum residual is determined as the thermal imaging image of the frame to be analyzed with the most abnormal local temperature difference trend change.

6. The method for detecting forming defects in titanium alloys according to claim 1, characterized in that, Determining the defect type of the suspected defect area includes: If the actual defect probability is greater than the high-risk probability threshold, then it is determined that there is a high-risk defect in the corresponding suspected defect area; If the actual defect probability is less than or equal to the high-risk probability threshold and greater than the low-risk probability threshold, then it is determined that there is a medium-risk defect in the corresponding suspected defect area. Otherwise, it is determined that there is no risk defect in the corresponding suspected defect area.

7. A titanium alloy forming defect detection system, characterized in that, The device includes a memory, a processor, and executable computer program code stored in the memory and executable on the processor. When the processor executes the computer program code, it performs a method for detecting forming defects in titanium alloys as described in any one of claims 1 to 6.

Images