A threshold-based centrifuge test image processing method

Abstract

The present disclosure relates to a threshold-based centrifuge test image processing method. The method comprises: collecting a set of images for a marker point in a centrifuge test; determining a first parameter of each image in the set of images; determining a first parameter threshold range based on the first parameter of a specific type of image in the set of images; obtaining a to-be-segmented target image corresponding to the first parameter based on the first parameter and the first parameter threshold range; determining a second parameter and a third parameter of each image in the to-be-segmented target image to determine an image segmentation threshold based on the second parameter and the third parameter, wherein the second parameter is used to represent the marker point and the third parameter is used to represent a non-marker point; and segmenting the to-be-segmented target image using the image segmentation threshold to obtain a segmented target image including the marker point. In this way, image quality evaluation and threshold segmentation can be performed in a double-threshold combination manner, and the deformation of the geotechnical body surface during the entire test process can be quickly and accurately obtained.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of centrifuge test image processing and image quality analysis, in particular to a threshold-based centrifuge test image processing method. BACKGROUND

[0002] In recent years, centrifuge model tests have been widely used in geotechnical engineering and geological disaster prevention. At present, the non-contact monitoring methods for model surface deformation during the test mainly include laser displacement sensor monitoring method and particle image velocimetry method. The laser displacement sensor monitoring method uses sensors arranged on the model to monitor the displacement of a certain place on the model surface during the test. The particle image velocimetry method obtains the particle velocity and flow direction of the entire phase field by cross-correlation operation on multiple images.

[0003] Although the laser displacement sensor monitoring method can truly reflect the position change of the measured place, one laser displacement sensor can only obtain the displacement change of one measuring point. To understand the displacement change of the entire model surface during the test, a large number of sensors need to be arranged, which is not only uneconomical but also difficult to carry out in the centrifuge model test due to space limitations. The particle image velocimetry method has high requirements for image quality, and it is necessary to ensure that the lighting conditions of all images are unchanged. Small changes in light can cause unstable results. With the rotation of the centrifuge and the lifting of the test platform during the centrifuge model test, the images collected during the test inevitably have the phenomenon of shadow and uneven brightness. When the cross-correlation calculation of the particle image velocimetry method is applied to such images, good results cannot be obtained. SUMMARY

[0004] The purpose of the present disclosure is to provide a threshold-based centrifuge test image processing method to at least partially solve the above problems existing in the prior art.

[0005] According to a first aspect of the present disclosure, a threshold-based centrifuge test image processing method is provided. The method comprises: collecting a set of images for at least one marker point in a centrifuge test, the at least one marker point being arranged on a test model in a predetermined manner and being associated with deformation of the test model in the centrifuge test; determining at least one first parameter of each image in the set of images; determining a first parameter threshold range based on the first parameter of a specific type of image in the set of images; obtaining a to-be-segmented target image corresponding to the first parameter based on the first parameter and the first parameter threshold range; determining a second parameter and a third parameter of each image in the to-be-segmented target image, so as to determine an image segmentation threshold based on the second parameter and the third parameter, wherein the second parameter is used to represent the marker point in the to-be-segmented target image and the third parameter is used to represent a non-marker point in the to-be-segmented target image; and segmenting the to-be-segmented target image by using the image segmentation threshold, to obtain a segmented target image including the at least one marker point.

[0006] In embodiments according to the present disclosure, image quality evaluation and threshold segmentation can be performed in a manner of combination of double thresholds, the deformation of the geotechnical body surface in the whole test process can be quickly and accurately obtained, and the data processing amount can be greatly reduced and the test efficiency can be improved. In embodiments of the present disclosure, image quality evaluation is performed by setting a first parameter threshold before marker point extraction, and images with insufficient exposure, excessive exposure, too small or too large light and shade difference are removed, thereby reducing the influence of light changes during centrifuge operation on image recognition effect, and the obtained results can correspond to the real data collected by the displacement sensor, thereby making up for the poor effect of the particle image velocimetry method in unstable light field. Secondly, the high-quality images screened by embodiments of the present disclosure are segmented again by using the threshold method, so as to accurately separate the marker point images and ensure the accuracy of the test model result measurement of the centrifuge test.

[0007] In some embodiments, segmenting the to-be-segmented target image by using the image segmentation threshold comprises: fine-tuning the image segmentation threshold during the segmentation process, wherein the image segmentation threshold is dynamically set by the difference between the second parameter and the third parameter. In such embodiments, when the image segmentation threshold method is used to segment the images of the to-be-segmented target images that have passed the quality screening, the dynamic threshold makes the segmentation of the marker points more accurate and achieves the best segmentation effect.

[0008] In some embodiments, the method further comprises: filtering and eroding the segmented target image; numbering the landmark points or the regions where the landmark points are located in the segmented target image and determining the correspondence between the landmark points in the segmented target image and the landmark points in the target image to be segmented; and in response to determining that the landmark points in the segmented target image and the target image to be segmented have correspondence, obtaining the coordinate information, the number and the time information of the landmark points in the segmented target image. In such embodiments, the influence of fine noise can be reduced, and by comparing the acquisition of the landmark points in the segmented target image and the originally acquired image, the extraction of the landmark points can be checked to ensure that there is no missing extraction or over-extraction.

[0009] In some embodiments, the method further comprises: based on the coordinate information, the number and the time information of the landmark points in each image of the segmented target image, verifying the accuracy of the coordinate information; and converting the coordinate information that passes the verification into the real displacement of the landmark points in the test process to obtain at least one of the displacement curve, the speed curve and the acceleration curve of the landmark points. In such embodiments, abnormal data can be effectively removed, or the target data can be re-extracted.

[0010] In some embodiments, verifying the accuracy of the coordinate information based on the coordinate information, the number and the time information of the landmark points in each image of the segmented target image comprises: determining whether the coordinate value change amount of each landmark point at each time meets the coordinate value change interval estimated based on the motion speed of the landmark points; in response to determining that the change amount meets the coordinate value change interval, retaining the coordinate information of the landmark point at the corresponding time; in response to determining that the change amount does not meet the coordinate value change interval, fine-tuning the image segmentation threshold, re-obtaining the coordinate information and re-determining the change amount; and in response to determining that the change amount does not meet the coordinate value change interval for multiple times, omitting the coordinate information of the landmark point at the corresponding time. In such embodiments, the availability of the target landmark point data can be determined based on the coordinate value change interval estimated based on the approximate speed of the landmark point motion.

[0011] In some embodiments, converting the coordinate information that passes the verification into the real displacement of the landmark points in the test process to obtain at least one of the displacement curve, the speed curve and the acceleration curve of the landmark points comprises: based on the actual distance corresponding to each pixel in the segmented target image being inversely calculated based on the point spacing of the landmark points actually set in the test model, converting the pixel coordinate value change amount of the landmark points into the actual physical distance change amount. In such embodiments, the target measurement curve of all the landmark points in the entire test process can be obtained to provide a data basis for subsequent analysis.

[0012] In some embodiments, converting the coordinate information passed by the inspection into real displacement of the mark point during the test to obtain at least one of the displacement curve, the velocity curve and the acceleration curve of the mark point includes: correcting the slope of the obtained deformation data according to the angle between the camera imaging plane and the surface of the test model, and converting the apparent deformation into the real deformation of the surface of the test model. In such embodiments, the real displacement, velocity and acceleration parameters of the slope of the mark point can be obtained.

[0013] In some embodiments, collecting a set of images for at least one mark point in the centrifuge test includes: selecting a mark point of a predetermined color and size according to the test model; estimating the deformation direction and deformation speed of the test model, selecting the arrangement direction and spacing of the mark point; and after the mark point is laid out, adjusting the camera shooting parameters and the angle of view. In such embodiments, the image can achieve a better contrast effect, further making the size of the mark point adapt to the size of the model, and ensuring the clarity of the mark point in the camera shot image.

[0014] In some embodiments, the spacing of the mark point is greater than the maximum value of the deformation of the surface of the test model within the time of continuously shooting two images by the camera. In such embodiments, the numbering of the mark point can be prevented from being confused during image post-processing.

[0015] In some embodiments, the method further includes: the first parameter includes at least one of the brightness average value and the brightness standard deviation; and / or the first parameter threshold range includes at least one of the brightness average value threshold range and the brightness standard deviation threshold range; and / or at least one of the second parameter or the third parameter includes the pixel color parameter; and / or the specific type of image includes one or more of the normal image, the overexposed image, the underexposed image and the image with too large light and dark difference. In such embodiments, the threshold parameters involved in the image segmentation of the specific feasible double threshold method are provided.

[0016] It should be understood that the content described in the summary section is not intended to limit the key or important features of the embodiments of the present disclosure, nor to limit the scope of the present disclosure. Other features of the present disclosure will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0017] The above and other objects, features and advantages of the embodiments of the present disclosure will be more apparent from the following detailed description read in conjunction with the accompanying drawings, in which:

[0018] Figure 1 The overall framework diagram of the threshold-based centrifuge test image processing method according to some embodiments of the present disclosure.

[0019] Figure 2 A schematic diagram of the layout of the marker points for a centrifuge test according to some embodiments of the present disclosure.

[0020] Figure 3 A schematic diagram of the change in the average value and standard deviation of the image brightness during a centrifuge test according to some embodiments of the present disclosure.

[0021] Figure 4 A schematic diagram of the marker points after threshold segmentation according to some embodiments of the present disclosure.

[0022] Figure 5 A schematic diagram of the displacement curve of the marker points after data processing according to some embodiments of the present disclosure.

[0023] In the various drawings, identical or corresponding reference numbers represent identical or corresponding parts. DETAILED DESCRIPTION

[0024] Embodiments of the present disclosure will be described in more detail with reference to the drawings. Although certain embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be embodied in various forms and should not be interpreted as being limited to the embodiments set forth herein, but rather, these embodiments are provided so as to more thoroughly and completely understand the present disclosure. It is understood that the drawings and embodiments of the present disclosure are for exemplary purposes only and are not intended to limit the scope of protection of the present disclosure.

[0025] In the description of embodiments of the present disclosure, the term "comprising" and its conjugations should be understood to encompass the meanings of "including" and "consisting of" as well as "consisting essentially of". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "an embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. can refer to different or identical objects. Other explicit and implicit definitions can also be included below.

[0026] As mentioned above, the laser displacement sensor monitoring method requires the arrangement of a large number of sensors, which is neither economical nor easy to carry out in a centrifugal model test due to space limitations, and the application of such images to the particle image velocimetry cross-correlation calculation cannot obtain good results. Thus, the present disclosure aims to address the problems of the current centrifugal model test, i.e., the laser displacement sensor monitoring is not comprehensive, and the particle image velocimetry method is greatly disturbed and cannot obtain high-quality deformation data. The present disclosure proposes a model surface deformation extraction method based on an image threshold segmentation algorithm and a matching model surface marker point layout method, which can obtain comprehensive and reliable model surface deformation data and save test costs. At the same time, since error or false image data is excluded, the data processing amount can also be greatly reduced, and the test efficiency can be improved.

[0027] Embodiments of the present disclosure will be described below in detail with reference to the accompanying drawings. It should be noted that in the embodiments of the present disclosure, "picture" and "image" belong to the same or similar concepts, and thus can be used interchangeably in the context herein.

[0028] Figure 1 A general framework diagram 100 of a threshold-based centrifuge test image processing method according to some embodiments of the present disclosure. Generally, the framework diagram 100 includes a landmark and camera arrangement 110, image quality evaluation 130, image segmentation 150, and data processing 170. It should be understood that the overall structure of the framework diagram 100 is described for exemplary purposes only, without implying any limitation on the scope of the present disclosure. Implementations of the present disclosure can also be applied to environments with different structures and / or functions, and at least one of the landmark and camera arrangement 110, image quality evaluation 130, image segmentation 150, and data processing 170 of the present disclosure can be omitted or replaced in a specific use scenario.

[0029] In the landmark and camera arrangement 110, reference is made to Figure 1 A set of images for at least one landmark is collected in a centrifuge test, the at least one landmark being arranged on a test model in a predetermined manner and being associated with deformation of the test model in the centrifuge test. The steps of the landmark and camera arrangement 110 will be introduced below in conjunction with Figure 2 .

[0030] Figure 2 A schematic diagram 200 of a centrifuge test landmark arrangement according to some embodiments of the present disclosure. In the embodiment as shown in Figure 2 , a centrifuge test can be performed in a centrifuge 210, in which a test model 230 can be placed. The test model 230 may, for example, be a geotechnical model or any other model that needs to be tested. The test model 230 can also be a model of any shape, such as a planar model, an inclined plane model, or a curved surface model, etc., without limitation of the present disclosure.

[0031] In some embodiments, in order to analyze the deformation of the test model 230 under the centrifugal action of the centrifuge 210, a set of landmarks 250 can be arranged on the test model 230, the set of landmarks 250 including at least one landmark 201. In this embodiment, reference is made to Figure 2 , the landmark 201 is arranged in a predetermined manner, for example, arranged as Figure 2The illustrated manner is straight in the vertical direction and staggered between different columns. In this way, each marker point 201 can generate deformation and displacement when subjected to the centrifugation of the centrifuge 210. By comparing the parameter changes of the images of the marker points 201 before and after the centrifugal test, and inverting the data, the parameters associated with the deformation of the test model 230 can be obtained, providing data for subsequent analysis.

[0032] It should be noted that the embodiment is only exemplary, and at least one marker point 201 can also be arranged in other forms according to actual needs, and the present disclosure does not limit this.

[0033] In some embodiments, in order to obtain marker points 201 that are clearly visible and easy to distinguish in images, the marker points can be selected in appropriate colors and sizes according to the test model 230 before the centrifuge test. The color of the marker points can be the complementary color of the surface color of the test model 230 to achieve a better contrast effect, and the size of the marker points can be adapted to the size of the model to ensure that they are clearly visible in the images taken by the camera. The marker points 201 can be pasted on the surface of the test model 230. For example, the color of the marker points 201 of the test model 230 can be red, green or any other appropriate color, and the present disclosure does not limit this. The color of the marker points 201 will be described in detail below taking red as an example.

[0034] In some embodiments, next, the deformation direction and deformation speed of the test model 230 can be estimated, and the arrangement direction and spacing of the marker points 201 can be selected. In one embodiment, the spacing of the marker points can be greater than the maximum value of the deformation of the model surface within the time of two consecutive images taken by the camera, to prevent the problem of marker point 201 numbering confusion during image post-processing, which will be explained in more detail below. After the marker points 201 are arranged, the camera shooting parameters and viewing angle can be adjusted to make the model exposure correct and the position appropriate, so as to take more high-quality images. In one embodiment, the camera can use a high frame rate recording mode to obtain more images and get better post-processing effect. After obtaining the images of the entire centrifuge test process, image quality evaluation 130 can be performed.

[0035] In the image quality evaluation 130, at least one first parameter of each image in the group of images obtained in the marker point and camera arrangement 110 operation can be determined, and based on the first parameter of a specific type of image in the group of images, a first parameter threshold range is determined, and based on the first parameter and the first parameter threshold range, a target image corresponding to the first parameter to be segmented is obtained.

[0036] In some embodiments, a first parameter can be calculated for each image captured throughout the experiment. The first parameter can be a single parameter or a set of multiple parameters. For example, the first parameter can be the average brightness and standard deviation of brightness of each image, or one of them. Of course, it can also be any other suitable parameter or set of parameters, and this disclosure does not limit it.

[0037] The following description uses the average brightness and standard deviation of brightness as the first parameters as an exemplary embodiment. That is, the exemplary method described below uses two factors, the average brightness and the standard deviation of brightness, of the image to control the image quality.

[0038] In some embodiments, the average image brightness can be calculated using the following equations (1) and (2). and the standard deviation of lightness S Exp :

[0039]

[0040]

[0041] Figure 3 This reflects the changes in image brightness during the experiment. Figure 3 300 is a schematic diagram of the changes in average and standard deviation of image brightness during a centrifuge test according to some embodiments of the present disclosure. See also, in some embodiments, […]. Figure 3 The diagram 300 illustrating the changes in mean and standard deviation may include a diagram 301 illustrating the changes in mean and a diagram 303 illustrating the changes in standard deviation. The mean brightness reflects the overall brightness of the image; controlling the brightness within a certain range ensures the image is neither overexposed nor underexposed. The standard deviation of brightness reflects the differences in brightness within the image; excessive differences in brightness are detrimental to image segmentation. After obtaining the mean and standard deviation of brightness for each image, the range of the mean brightness and the threshold range of the standard deviation are determined using pre-selected typical images.

[0042] In some embodiments, reference Figure 1 The range of average brightness and the standard deviation threshold range can be determined by selecting typical images of a specific type. This specific type could be one or more of the following: normal images, overexposed images, underexposed images, images with excessive brightness differences, or other types of images. In one embodiment, several typical normally exposed images can be manually selected. The range of average brightness is then determined based on the average brightness of these normally exposed images, and the standard deviation threshold range is determined based on the standard deviation of brightness of these normally exposed images. In other words, images falling within the range of average brightness and the standard deviation threshold range can be classified as normal images, and these images matching the threshold range are the images to be segmented in the next step.

[0043] It should be noted that other types of images, such as overexposed images, underexposed images, and images with too large light-dark differences, can also be used to determine the luminance average value range and the standard deviation threshold value range, so that images not falling within the range can be used as the target image to be segmented. In addition, other suitable methods can be used to determine the first parameter threshold value range, and the present disclosure is not limited in this regard.

[0044] It should be noted that the first parameter threshold value range can be a specific range, and in the case of requiring high-precision screening of image quality, the first parameter threshold value range can also be set to a specific threshold value, and the present disclosure is not limited in this regard.

[0045] Returning to Figure 1 After the target image to be segmented is obtained in the image quality evaluation 130, image segmentation 150 can be performed. In the image segmentation 150, the second parameter and the third parameter of each image in the target image to be segmented are determined to determine the image segmentation threshold based on the second parameter and the third parameter, wherein the second parameter is used to represent the landmark points in the target image to be segmented and the third parameter is used to represent the non-landmark points in the target image to be segmented.

[0046] In some embodiments, the second parameter or the third parameter can be a color parameter of a pixel, that is, the second parameter can be a color parameter of a pixel of a landmark point, and the third parameter can be a color parameter of a pixel at a non-landmark point position. In one embodiment, the color parameters of the pixels at the landmark points and the non-landmark point positions in the image can be measured, and the image segmentation threshold can be set according to the difference between the two.

[0047] In one embodiment, with reference to Figure 4 The threshold method can be used to segment the target image to be segmented that meets the above quality evaluation requirements, and the selected factors can be three indexes in the image HSV (H: hue, S: saturation, V: brightness) space, and the thresholds of the three indexes can be set according to the color of the landmark points and the specific circumstances of the test. For example, the threshold range of red landmark points is approximately H>0.7, S<1, and V>0.1. Subsequently, as shown in Figure 4 After the landmark point pixels in each image are extracted according to the above thresholds, the extracted pixels are newly created on a blank image of the same size as the original image. In this way, at least one threshold segmented landmark point schematic diagram 400 composed of extracted landmark points 401 is generated, and in the case of ideal extraction effect, each extracted landmark point 401 can correspond to Figure 2 the landmark point 201 at the corresponding position in the image in FIG. 2.

[0048] It should be noted that in addition to the HSV index, the threshold method can also use other index types, such as the RGB index and the grayscale index, and the present disclosure is not limited in this regard.

[0049] In one embodiment, the image segmentation threshold can be constantly fine-tuned in actual operation to achieve the best segmentation effect. In this way, a segmented target image including at least one extracted landmark point 401 can be obtained.

[0050] In one embodiment, with continued reference to Figure 1 Next, the extracted image can be filtered and eroded to remove small interference pixels that can exist in the extraction process and reduce the influence of fine noise. Subsequently, the segmented target image can be binarized to label each extraction region. Each region can be numbered in a certain order, such as from the top of the image to the bottom of the image and from the left of the image to the right of the image. The specific order can be set according to actual needs, and the key is that all images should adopt the same numbering method to ensure that the coordinates of the extracted landmark points can be accurately corresponded. Because there can be individual cases of poor image segmentation quality, which can result in a number of extracted landmark points that does not match the actual number of landmark points, the number of labels in the image range can be checked after segmentation to determine whether it is the same as the actual number of landmark points pasted on the model. If they are the same, the image is considered to be a valid image, and the centroid of each label region is then calculated, and the centroid coordinates of each label, the image shooting time information, and the like are stored in the information library for subsequent data processing. If they are different, the image is discarded.

[0051] It should be noted that the checking can also be performed by directly labeling the landmark points, as long as the corresponding effect can be achieved, and the present disclosure does not limit this.

[0052] At this point, the segmentation of an image and the extraction of the landmark point position are completed, and the next image at a time is imported after the extraction of one image is completed, and the above operation is repeated until all images are processed.

[0053] Returning to Figure 1 In data processing 170, the accuracy of the coordinates of the landmark points can be judged to again check the accuracy of the coordinates of the extracted landmark points. In one embodiment, the coordinate value change of each point at each time can be checked to determine whether it is within the coordinate value change interval of the estimated speed of the landmark point movement. If it is, the coordinate of the point at the time is retained, and if the coordinate value changes greatly, it can be caused by image extraction error. In this case, the threshold value after fine-tuning can be used to re-extract the sub-image, and the coordinate value change is calculated again. If the coordinate value change is normal, the new point coordinate information is saved to the storage list, and if the abnormality still exists after multiple re-extractions, the point position information at the time is omitted.

[0054] In some embodiments, after the coordinates of the points are verified, the actual distance corresponding to each pixel in the image is inversed according to the distance between the actual pasted mark points, and then the pixel coordinate value change of all the points is converted into the actual physical distance change. In one embodiment, when the observed object is a plane, the distance conversion is not needed. If the observed object is an inclined plane, the distance change value needs to be projected onto the plane where the mark points are located according to the inclination of the inclined plane to obtain the real displacement change of the mark points. If the observed object is a curved surface, the distance change value needs to be projected onto the curved surface where the mark points are located according to the curvature of the curved surface to obtain the real displacement change of the mark points.

[0055] Figure 5 A schematic diagram of the displacement curve of the mark points after data processing according to some embodiments of the present disclosure is shown. In some embodiments, as shown in FIG. 6, the data obtained after all the above steps are completed can be subjected to curve smoothing and other steps to obtain the displacement change curve of the model surface. In some embodiments, the speed change curve, the acceleration change curve or other change curves required by the test can also be obtained to provide corresponding data for more in-depth analysis. Figure 5

[0056] The various embodiments of the present disclosure use threshold values to evaluate the quality of the images collected during the centrifuge test process, and again use threshold values to segment the images and the mark points. The two operation steps are organically combined to extract the positions of the mark points in each image to obtain the deformation of the model surface during the entire test process, which can better obtain the deformation of the model surface during the test process, while greatly reducing the data processing amount and improving the test efficiency. The first parameter threshold value is set before the extraction of the mark points to evaluate the quality of the images, and the underexposed, overexposed, images with too small or too large light and dark differences are removed, which reduces the influence of the light changes during the operation of the centrifuge on the image recognition effect, and the obtained results can correspond to the real data collected by the displacement sensor, which makes up for the poor effect of the particle image velocimetry method in unstable light fields. Secondly, the embodiments of the present disclosure use the threshold value method to segment the high-quality images selected again to accurately separate the mark point images, which ensures the accuracy of the measurement of the test model results of the centrifuge test.

[0057] Although the above discussion contains a number of implementation details, these should not be construed as limiting the scope of the present disclosure. Certain features described in the context of separate embodiments can also be implemented in combination. Conversely, various features described in the context of a single embodiment can also be implemented separately or in any suitable subcombination.

[0058] ​Moreover, while operations are depicted in a particular order, this should not be understood as requiring such an order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing can be advantageous. Likewise, while several specific implementation details are contained in the above discussion, these should not be construed as limitations on the scope of the disclosure, but rather as descriptions of particular implementations. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single implementation. Conversely, various features that are described in the context of a single implementation can also be implemented in subcombination or as separate implementations in accordance with the embodiments set forth herein.

[0059] Embodiments of the present disclosure have been described above, with examples illustrating, but not exhaustively, and without limitation, the various embodiments disclosed. Numerous modifications and adaptations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The selection of the terms to be used in the description is intended to best convey the principles of the embodiments, practical application, or improvement to the art, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A threshold-based image processing method for centrifuge tests, characterized in that, include: In the centrifuge test, a set of images is acquired for at least one marker point, the at least one marker point being set on the test model in a predetermined manner and associated with the deformation of the test model in the centrifuge test; Determine at least one first parameter for each image in the set of images; Based on the first parameter of a specific type of image in the set of images, determine the threshold range of the first parameter; Based on the first parameter and the threshold range of the first parameter, the target image to be segmented corresponding to the first parameter is obtained; A second parameter and a third parameter are determined for each image in the target image to be segmented, and an image segmentation threshold is determined based on the second parameter and the third parameter, wherein the second parameter is used to characterize the marker points in the target image to be segmented and the third parameter is used to characterize the non-marker points in the target image to be segmented; as well as The target image to be segmented is segmented using the image segmentation threshold to obtain a segmented target image including at least one of the marker points; The method includes: The segmented target image is then subjected to filtering and erosion processing. The marker points or regions located in the segmented target image are numbered, and the correspondence between the marker points in the segmented target image and the marker points in the target image to be segmented is determined; and In response to determining that a marker point in the segmented target image and a marker point in the target image to be segmented have the corresponding relationship, the coordinate information, number, and time information of the marker point in the segmented target image are obtained; and the method further includes: Based on the coordinate information, the number, and the time information of the marker points in each image of the segmented target image, the accuracy of the coordinate information is verified; and The verified coordinate information is converted into the actual displacement of the marker point during the test process to obtain at least one of the displacement change curve, velocity change curve and acceleration change curve of the marker point. The accuracy of the coordinate information is verified based on the coordinate information, the number, and the time information of the marker points in each image of the segmented target image, including: Determine whether the change in coordinate value of each of the aforementioned marker points at each moment conforms to the range of coordinate value changes predicted by the movement speed of the marker points; If it is determined that the change amount conforms to the coordinate value change range, the coordinate information of the marker point at the corresponding time is retained; If, in response to the determination that the change amount does not conform to the coordinate value change range, the image segmentation threshold is fine-tuned, the coordinate information is reacquired, and the change amount is re-determined; and If it is determined that the change amount does not conform to the coordinate value change range multiple times, the coordinate information of the marker point at the corresponding time is omitted.

2. The method according to claim 1, characterized in that, Segmenting the target image to be segmented using the image segmentation threshold includes: The image segmentation threshold is fine-tuned during the segmentation process, wherein the image segmentation threshold is dynamically set by the difference between the second parameter and the third parameter.

3. The method according to claim 2, characterized in that, The verified coordinate information is converted into the actual displacement of the marker point during the test, so as to obtain at least one of the displacement change curve, velocity change curve, and acceleration change curve of the marker point, including: Based on the actual spacing of the marker points set in the experimental model, the actual distance corresponding to each pixel in the segmented target image is inverted; and The change in the pixel coordinates of the marker is converted into the actual change in physical distance.

4. The method according to claim 2, characterized in that, The verified coordinate information is converted into the actual displacement of the marker point during the test, so as to obtain at least one of the displacement change curve, velocity change curve, and acceleration change curve of the marker point, including: The deformation data obtained is corrected by tilting the camera's imaging plane to the surface of the test model, and the apparent deformation is converted into the actual deformation of the test model's surface.

5. The method according to claim 1, characterized in that, The set of images acquired during the centrifuge test for at least one marker point includes: The marker points of predetermined color and size are selected according to the experimental model; Estimate the deformation direction and deformation rate of the test model, and select the orientation and spacing of the marker points; and After the markers are set up, adjust the camera shooting parameters and angle.

6. The method according to claim 5, characterized in that, The spacing between the marker points is greater than the maximum value of the surface deformation of the test model within the time interval between two consecutive image captures by the camera.

7. The method according to claim 6, characterized in that, The method further includes: The first parameter includes at least one of: the average lightness and the standard deviation of lightness; and / or The first parameter threshold range includes at least one of: the average lightness threshold range and the standard deviation lightness threshold range; and / or At least one of the second parameter or the third parameter includes: a pixel color parameter; and / or The specific image types include one or more of the following: normal images, overexposed images, underexposed images, and images with excessive brightness differences.

Images