A round hole model feature extraction method and device and a storage medium

Abstract

The application provides a round hole model feature extraction method, which comprises the following steps: loading a set of CT images of a round hole model, performing deartifacting and noise reduction processing on the CT images and converting the CT images into a set of binary images; determining the fuzzy center coordinates of a first round hole in the binary images by using the image features of the binary images; intercepting a target pixel block containing the first round hole from the binary images; determining the center of the first round hole and the data of the left and right edges according to the image features of the target pixel block, and determining the coordinates of the center point of the first round hole based on the binary images, the center row coordinates of the first round hole and the left and right edge data. The above technical scheme can effectively improve the speed and accuracy of model feature extraction by collecting a set of CT images of a round hole model, performing deartifacting and noise reduction processing on the set of CT images and converting the CT images into binary images, and determining the round hole center coordinates according to the obtained binary images.

Description

Technical Field

[0001] This invention relates to, but is not limited to, the field of computers, and particularly to a method, apparatus, and storage medium for extracting features from circular hole models. Background Technology

[0002] Current methods for feature extraction of circular holes typically rely on edge detection operators, such as the Canny edge detection operator and the Sobel edge detection operator, to extract the edges of the hole contour. However, due to unavoidable environmental noise from the bed frame and other sources during medical image scanning, artifacts are often generated. Using edge detection operators for feature extraction often results in highly noisy extraction results, failing to effectively remove unwanted elements. Furthermore, conventional manual extraction methods, besides requiring skilled personnel, often cannot meet the demands of rapid and high-precision extraction from a single set of medical images. Summary of the Invention

[0003] The technical problem to be solved in this application is to provide a method, device and storage medium for extracting features from a circular hole model, which can improve the speed and accuracy of model feature extraction.

[0004] This application provides a method for feature extraction of a circular hole model, wherein the circular hole model is provided with a first circular hole, and the method includes:

[0005] Load a set of computed tomography (CT) images of the circular hole model, perform artifact removal and noise reduction processing on the CT images, and convert them into a set of binarized images;

[0006] Using the image features of the binarized image, the coordinates of the blur center of the first circular hole in the binarized image are determined;

[0007] Based on the fuzzy center coordinates and the engineering aperture characteristics of the circular hole model, a target pixel block containing the first circular hole is extracted from the binarized image; based on the image characteristics of the target pixel block, the center row coordinates and left and right edge data of the first circular hole are determined.

[0008] The coordinates of the center point of the first circular hole are determined based on the binarized image, the center row coordinates of the first circular hole, and the data of the left and right edges.

[0009] This application also provides a feature extraction device for a circular hole model, including: a memory and a processor.

[0010] The memory is used to store the program for feature extraction of the circular hole model;

[0011] The processor is configured to read the program for extracting features from a circular hole model and execute any of the aforementioned circular hole model feature extraction methods.

[0012] This application also provides a non-transient computer-readable storage medium storing a computer program, wherein the computer program is configured to execute any of the aforementioned circular hole model feature extraction methods at runtime.

[0013] This application provides a method for feature extraction of a circular hole model, wherein the circular hole model is provided with a first circular hole. The method includes: loading a set of computed tomography (CT) images of the circular hole model; performing artifact removal and noise reduction processing on the CT images and converting them into a set of binarized images; using the image features of the binarized images to determine the blurred center coordinates of the first circular hole in the binarized images; based on the blurred center coordinates and the engineering aperture features of the circular hole model, extracting a target pixel block containing the first circular hole from the binarized images; determining the center row coordinates and left and right edge data of the first circular hole based on the image features of the target pixel block; and determining the coordinates of the center point of the first circular hole based on the binarized images, the center row coordinates and left and right edge data of the first circular hole. The above technical solution acquires a set of CT images of a circular hole model, and performs binarization processing on the set of CT images after removing artifacts and environmental noise. The center coordinates of the circular hole are determined based on the binarized set of CT images. This can replace manual extraction of feature data, thereby effectively improving the speed and accuracy of model feature extraction, reducing the labor cost of manual processing, and reducing the professional skill requirements for operators. Attached Figure Description

[0014] The accompanying drawings are used to provide an understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0015] Figure 1 This is a flowchart of the circular hole model feature extraction method according to an embodiment of this application;

[0016] Figure 2 This is a schematic diagram of a CT image of a circular hole model according to an embodiment of this application;

[0017] Figure 3 This is a flowchart of step S1 in an embodiment of this application;

[0018] Figure 4 This is a flowchart of step S2 in an embodiment of this application;

[0019] Figure 5 This is a flowchart of step S3 in an embodiment of this application;

[0020] Figure 6 This is a flowchart of step S4 in an embodiment of this application;

[0021] Figure 7 This is a schematic diagram of a circular hole model feature extraction device according to an embodiment of this application. Detailed Implementation

[0022] This application describes several embodiments, but these descriptions are exemplary and not restrictive, and it will be apparent to those skilled in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with or in lieu of any other feature or element in any other embodiment.

[0023] This application includes and contemplates combinations of features and elements known to those skilled in the art. The embodiments, features, and elements disclosed in this application may also be combined with any conventional features or elements to form a unique inventive scheme as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive schemes to form another unique inventive scheme as defined by the claims. Therefore, it should be understood that any feature shown and / or discussed in this application may be implemented individually or in any suitable combination. Therefore, the embodiments are not limited except by the limitations imposed by the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

[0024] Furthermore, in describing representative embodiments, the specification may have presented methods and / or processes as a specific sequence of steps. However, the method or process should not be limited to the specific order of steps described herein, to the extent that it does not depend on such a specific order. As will be understood by those skilled in the art, other sequences of steps are also possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation of the claims. Moreover, the claims concerning the method and / or process should not be limited to the steps performed in the written order, and those skilled in the art will readily understand that these orders can be varied and still remain within the spirit and scope of the embodiments of this application.

[0025] First, the parameters used in the model feature extraction process in this application embodiment are explained as follows. The following parameters are all labels that exist in the medical image DICOM file and can be directly obtained from the image.

[0026] SliceThickness: Represents the distance between two CT images, in millimeters;

[0027] PixelSpaceing: Represents the actual physical distance between the same pixel in two CT (Computed Tomography) images, in millimeters; for example, the physical distance between pixel A in CT image 1 and pixel A' in CT image 2.

[0028] WindowWidth: Represents the window width parameter corresponding to the current image state. Adjusting this parameter can display tissues of different densities. It is abbreviated as W below.

[0029] WindowCenter: Represents the window level parameter corresponding to the current image state. Adjusting this parameter can display tissues of different densities. It is abbreviated as L below.

[0030] Slope: Adjusts the slope. Used in conjunction with Intercept, it converts the raw orival values ​​generated by CT into HU values ​​(HU is a unit of measurement for the density of a local tissue or organ in the human body, usually called Henle unit).

[0031] Intercept: The intercept, used in conjunction with Slope, is used to convert the raw orival values ​​generated by CT into HU values;

[0032] The conversion formula is: HU = orival * Slope + Intercept;

[0033] SliceNumber: The current image parsing sequence number.

[0034] like Figure 1 As shown, this embodiment provides a method for feature extraction of a circular hole model, wherein the circular hole model is provided with a first circular hole, and the method includes:

[0035] Step S1: Load a set of computed tomography (CT) images of the circular hole model, perform artifact removal and noise reduction processing on the CT images, and convert them into a set of binarized images;

[0036] Step S2: Using the image features of the binarized image, determine the coordinates of the blurred center of the first circular hole in the binarized image;

[0037] Step S3: Based on the fuzzy center coordinates and the engineering aperture features of the circular hole model, extract the target pixel block containing the first circular hole from the binarized image; based on the image features of the target pixel block, determine the center row coordinates and left and right edge data of the first circular hole;

[0038] Step S4: Determine the coordinates of the center point of the first circular hole based on the binarized image, the center row coordinates of the first circular hole, and the left and right edge data.

[0039] In this embodiment, as Figure 2 The CT image shown depicts a circular hole model. This model can have a central circular hole surrounded by multiple smaller circular holes. Extraction points can be determined based on the specific circular holes to be extracted. For example, the circular hole features at the center of the top layer can be extracted first, followed by the features of the other circular holes.

[0040] The above technical solution acquires a set of target medical images of a circular aperture model. By analyzing specific parameters and extracting features from the target medical images, noise can be accurately removed, and the relative position of the blurred circular aperture can be located. An edge fitting algorithm can effectively improve the extraction accuracy of the geometric center position of the circular aperture, meeting the requirements for high-precision and rapid circular aperture parameter extraction. Simultaneously, by extracting features from the circular aperture model, registration and positioning between CT equipment and the medical navigation system can be achieved.

[0041] It should be noted that the above technical solution can also use other medical images besides CT images for circular aperture model feature extraction. For example, a set of MRI images of the circular aperture model can be acquired, and then the coordinates of the center point of the circular aperture can be determined by the circular aperture model feature extraction method in this embodiment, thereby realizing the registration and positioning of the MRI equipment and the navigation system.

[0042] In one exemplary embodiment, such as Figure 3 As shown, step S1, which involves performing artifact removal and noise reduction on the CT image and converting it into a set of binary images, may include:

[0043] Step S11: Select a CT image containing the first circular hole according to the input command, and adjust the window width and window level of the CT image to remove artifacts in the CT image, and record the adjusted window width W and window level L;

[0044] In this step, a set of CT images of the circular hole model are acquired and loaded. The CT image that can display the first circular hole to be extracted is selected. The window width and window level of the image are manually adjusted to reduce image noise. It is confirmed that there are no obvious metal artifacts in the CT image and that the phantom is complete and the edges are clear. The window width W and window level L at this time are recorded.

[0045] Step S12: Based on the adjustment slope and intercept of each of the CT images in the group, convert the original values ​​of the pixels in the CT images in the group into HU values ​​(CT value, Hounsfield Unit, which reflects the degree of tissue absorption of X-rays), and filter the converted HU values ​​according to W and L.

[0046] In this step, according to the respective adjustment slopes and intercepts of this group of CT images, the original values of the pixel points in this group of CT images are respectively converted into HU values; for each image in this group of CT images, its Slope and Intercept are respectively read, and then each original value orival in the image is converted into an HU value according to the conversion formula (HU = orival * Slope + Intercept).

[0047] Step S13: Convert the filtered HU values of the pixel points in this group of CT images into grayscale values;

[0048] In this step, the HU values can be converted into a set of image grayscale values, marked as the point set data, satisfying the following content: data = (HU + 1024) / 16.

[0049] Step S14: Set the grayscale values of the pixel points within the preset bedplate area in this group of CT images to 0 to filter the noise brought by the bedplate;

[0050] In this step, a binary set Bin[w, h] can be set (w is the total number of pixel rows of the image, h is the total number of pixel columns of the image), the bedplate height threshold is set to 50, and Bin[i, j] = 0 (where, w - 50 < i < w, 0 < j < h), so as to filter the bedplate noise.

[0051] Step S15: Determine the grayscale value threshold according to W, L and the set grayscale value threshold ratio, and binarize some or all of the areas including the first round hole in this group of CT images according to the grayscale values of the pixel points in this group of CT images and the grayscale value threshold, to obtain a group of binarized images.

[0052] In this step, the grayscale value threshold ratio can be set and denoted as Rate (for example, it can take a value of 0.7), and then the grayscale value threshold Threshold_bin = (L + Rate * W - W / 2 + 1024) / 16 is calculated;

[0053] Binarize the ROI area of the grayscale value image, including:

[0054] If data[i, j] > Threshold_bin (0 < i < w - bottom, 0 < j < h), then change the value of Bin[i, j] to 1.

[0055] If data[i, j] < Threshold_bin (0 < i < w - bottom, 0 < j < h), then change the value of Bin[i, j] to 0.

[0056] The above technical solution removes artifacts and couch noise in the CT image, records the window width and window level of the CT image after removing the artifacts, and performs binarization processing on a group of CT images according to the window width and window level, so that the circular hole center coordinates obtained from the group of binarized images are more accurate.

[0057] In an exemplary embodiment, filtering the converted HU values according to W and L in step S12 may include:

[0058] Setting a maximum filtering threshold maxHU and a minimum filtering threshold minHU, where maxHU = L + W / 2 and minHU = L - W / 2;

[0059] For the converted HU value, if HU > maxHU, then modify the HU value to maxHU; if HU < minHU, then modify the HU value to maxHU; if minHU < HU < maxHU, then keep the HU value unchanged.

[0060] In an exemplary embodiment, the first circular hole is located in the middle of the top layer of the model; in step S2, using the image features of the binarized image to determine the fuzzy center coordinates of the first circular hole in the binarized image may include:

[0061] As Figure 4 shown, for each binarized image, determine the row coordinate in the fuzzy center coordinates in the following manner;

[0062] Step S_{21}, determine the threshold value of the number of effective pixels on one row of the top layer according to the engineering length data of the top layer, the physical distance between adjacent pixel points, and the set effective data rate threshold;

[0063] Step S_{22}, count the number of effective pixels in each row of the binarized image in the order from top to bottom, and record the first row with the number of effective pixels greater than the threshold value of the number of effective pixels as the starting row of the top layer;

[0064] Step S_{23}, use the row coordinate corresponding to the starting row as the row coordinate in the fuzzy center coordinates.

[0065] In an exemplary embodiment, step S2 using the image features of the binarized image to determine the fuzzy center coordinates of the first circular hole in the binarized image may further include:

[0066] For each binarized image, determine the column coordinate in the fuzzy center coordinates in the following manner:

[0067] Step S_{24}, determine the number of pixel points corresponding to the model height according to the engineering height data of the top layer and the physical distance;

[0068] Step S25: Determine the search range of the top layer in the height direction of the binarized image based on the position of the starting row and the number of pixels corresponding to the model height;

[0069] Step S26: Count the number of valid pixels in each column within the search range, and find an intermediate column such that the difference between the sum of the number of valid pixels in all columns to the left of the intermediate column and the sum of the number of valid pixels in all columns to the right of the intermediate column is minimized.

[0070] Step S27: Use the column coordinates corresponding to the middle row as the column coordinates in the fuzzy center coordinates.

[0071] In this embodiment, valid pixels refer to the pixels where the model entity is located, and these pixels have a value of 1 in the binarized image. Binarization does not require the entire image to be binarized; an image containing the model can also be called a binarized image after binarization.

[0072] The execution order of steps S21 to S23 and steps S24 to S27 is not specifically limited. Specifically, it can be implemented in the following ways:

[0073] Using engineering data, the length of the first layer (i.e., the top layer) entity is denoted as `first_layer_length` (in mm), and the effective data rate threshold is denoted as `threshold_first_layer_ratio` (for example, a value of 0.55). The effective pixel count threshold `threshold_first_layer` for the first layer is calculated as follows:

[0074] threshold_first_layer=threshold_first_layer_ratio*

[0075] (first_layer_length / PixelSpacing)

[0076] The effective data rate threshold threshold_first_layer_ratio and the corresponding effective pixel count threshold_first_layer are used to sequentially find the row number of the first (top) layer of the model in each CT image.

[0077] Similarly, by combining the actual height of the engineering data, the actual height pixel value of the model can be calculated, denoted as h_layer1_4; where h_layer1_4 = engineering model height / PixelSpacing.

[0078] For the Bin[w, h] output in step S14, sum the elements of each row, which can be compressed into a one-dimensional array, denoted as y[w].

[0079] Set the parameter first_y (first_y represents the row coordinate in the fuzzy center coordinate) equal to 0, and traverse y[j], where 0 < j < w.

[0080] If there exists y[j] > threshold_first_layer, then mark first_y = j, and end the traversal.

[0081] If y[j] is less than threshold_first_layer, then do not mark. If all y[j] are less than threshold_first_layer, it means there is no first round hole in this image.

[0082] ​​​​​​​​​​​​​​​​​​​​​​​

[0090] In one exemplary embodiment, such as Figure 5 As shown, step S3 may include:

[0091] Step S31: Determine the number of rows and columns contained in the target pixel block based on the fuzzy center coordinates and the engineering aperture characteristics of the circular hole model;

[0092] Step S32: Count the number of effective pixels in each row of the target pixel block in the binarized image;

[0093] Step S33: Traverse each row of the target pixel block and record the average row coordinates of the non-zero row with the most effective pixels as the center row coordinate data of the first circular hole.

[0094] Step S34: Count the number of effective pixels in each column of the target pixel block in the binarized image;

[0095] Step S35: Traverse each column of the target pixel block, determine the leftmost and rightmost columns where the number of effective pixels is less than the set minimum pixel count threshold, and record the column coordinates corresponding to the leftmost column and the column coordinates corresponding to the rightmost column as the left edge data and right edge data of the first circular hole, respectively.

[0096] It should be noted that if there is only one column with the number of valid pixels less than the set minimum pixel count threshold, then that column is both the leftmost and rightmost column.

[0097] Steps S31 to S35 can be achieved in the following way:

[0098] First, extract the target pixel block of the first circular hole: For each binarized image, extract the target image region Bin[w1:w2,h1:h2] as the target pixel block containing the first circular hole;

[0099] w1 = first_y - woff; w2 = first_y + w' - woff; h1 = center_x - h' / 2; h2 = center_x + h' / 2; where first_y and center_x are the blurred row coordinates and blurred column coordinates of the center point of the first circular hole in the binarized image, w' and h' are the set number of rows and columns contained in the target pixel block, and woff is the set upward offset.

[0100] For example, if the number of rows w' ​​is 20, the number of columns h' is 30, and the offset w offIf the value is 3, the target binary image region Bin[first_y-3:first_y+20-3,center_x-15:center_x+15] can be extracted and set as patch[20,30]. The sum of each column of patch can be compressed into a one-dimensional array labeled as patch_y

[30] . The sum of each row of patch can be compressed into a one-dimensional array labeled as patch_x

[20] .

[0101] With Xmax = 0 and Max_pos = 0, iterate through patch_x[i], where i is 0...20, and satisfy the following condition:

[0102] If patch_x[i] is greater than Xmax, then Xmax is equal to patch_x[i], Max_pos = i, and the iteration continues;

[0103] If patch_x[i] equals Xmax and patch_x[i] > 0, then Max_pos becomes the set to add element i, and the traversal continues.

[0104] After the traversal is complete, the row pixel coordinates of the center of the first hole can be calculated and set as center_y;

[0105] center_y = average value of the Max_pos set + first_y - 3;

[0106] The `edge` array is assumed to be empty. Iterate through `patch_y[j]`, where `j` is 0...30, satisfying the following condition:

[0107] If patch_y[j] is less than 2 (the smallest pixel), then add element i to the edge array.

[0108] After the traversal is complete, if edge is not empty, then the two edges of the current image can be identified and labeled as edge1 and edge2, where:

[0109] edge1 = edge[0] + center_x - 15 (0 represents the first element)

[0110] edge2 = edge[-1] + center_x - 15 (-1 represents the last element)

[0111] Using the above technical solution, the data of the left and right edges of the first circular hole can be obtained.

[0112] In one exemplary embodiment, such as Figure 6 As shown, step S4, which determines the coordinates of the center point of the first circular hole based on the image, the center row coordinates of the first circular hole, and the left and right edge data, may include:

[0113] Step S41: Select a binarized image of the left and right edge data of the first circular hole that is not empty;

[0114] Step S42: Accumulate and average the left and right edge data of the first circular hole in all selected binarized images to obtain the pixel column coordinates of the center point of the first circular hole;

[0115] Step S43: Accumulate and average the row coordinate data of the center of the first circular hole in all the selected binarized images to obtain the pixel row coordinates of the center point of the first circular hole;

[0116] Step S44: Accumulate the sequence numbers of all selected binarized images and calculate the average value to obtain the image sequence number of the center point of the first circular hole.

[0117] Steps S41 to S44 can be implemented in the following way:

[0118] Assuming the circular hole model corresponds to a set of CT images A1, A2, ..., An, after processing in step S1, a set of corresponding binary images a1, a2, ..., an are obtained.

[0119] Starting with the binarized image a1, we collect the left and right edge data of the first circular hole in a1, label the three-dimensional array set of the first layer of intermediate hole data as hole_0, and then perform the following operations:

[0120] If edge is empty, jump to the next binarized image (e.g., if currently processing a1, jump to a2); if edge is not empty, add two edge elements [edge1, SliceNumber, center_y] and [edge2, SliceNumber, center_y] to the array set hole_0, and jump to the next binarized image (e.g., if currently processing a1, jump to a2).

[0121] After all elements from a1 to an have been traversed, the sum of all elements in the three-dimensional array hole_0 is averaged to obtain the center coordinates of the middle circular hole in the first layer, including the pixel column coordinates, image number, and pixel row coordinates.

[0122] The calculation method for the center coordinates of other circular holes is similar to that for the center coordinates of the middle circular holes in the first layer. Simply refer to step S2 to obtain the fuzzy center coordinates of the corresponding circular holes based on the engineering data and layer thickness data, then refer to step S3 to obtain the data of the left and right edges of the circular holes, and finally, step S4 to calculate the center coordinates of the circular holes.

[0123] Using the above technical solution, the pixel row coordinates and column coordinates of the center point of the first circular hole can be obtained based on the data of the left and right edges of the first circular hole.

[0124] It should be noted that the image sequence number can be a non-integer, which allows for higher precision. Furthermore, for binarized images, the sum of the values ​​of all pixels in each row gives the number of valid pixels in that row.

[0125] The model feature extraction in this application embodiment can achieve the following technical effects:

[0126] 1. The above technical solution records the window width and window level of a CT image after artifact removal, and performs binarization processing on a set of CT images based on this window width and window level, making the coordinates of the circular hole center obtained from the binarized images more accurate. Simultaneously, by preprocessing the target image using CT image pixel spacing parameters and engineering feature parameters, environmental interference noise such as that from the bed board can be effectively removed, thereby significantly improving the accuracy of model feature extraction.

[0127] 2. By extracting key parameters from medical images, the three-dimensional spatial coordinates of the blurred image of the feature hole are located based on engineering data. Then, the left and right edge data of the hole in each CT image are calculated sequentially. Finally, the center coordinates of the hole are accurately fitted, which effectively improves the efficiency and accuracy of feature extraction.

[0128] like Figure 7 As shown, this application also provides a feature extraction device for a circular hole model, including: a memory 10 and a processor 11.

[0129] The memory 10 is used to store the program for feature extraction of the circular hole model;

[0130] The processor 11 is used to read the program for extracting features from a circular hole model and execute any of the aforementioned circular hole model feature extraction methods.

[0131] This application also provides a non-transient computer-readable storage medium storing a computer program, wherein the computer program is configured to execute any of the aforementioned circular hole model feature extraction methods at runtime.

[0132] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A method for feature extraction of a circular hole model, wherein the circular hole model is provided with a first circular hole, the method comprising: Load a set of computed tomography (CT) images of the circular hole model, perform artifact removal and noise reduction processing on the CT images, and convert them into a set of binarized images; Using the image features of each binarized image, determine the fuzzy center coordinates of the first circular hole in the set of binarized images; Based on the fuzzy center coordinates and the engineering aperture characteristics of the circular hole model, a target pixel block containing the first circular hole is extracted from the binarized image; based on the image characteristics of the target pixel block, the center row coordinates and left and right edge data of the first circular hole are determined. The coordinates of the center point of the first circular hole are determined based on each binarized image, the center row coordinates of the first circular hole, and the data of the left and right edges.

2. The method as described in claim 1, characterized in that, The process of performing artifact removal and noise reduction on the CT images and converting them into a set of binary images includes: Select a CT image containing the first circular hole according to the input command, and adjust the window width and window level of the CT image to remove artifacts in the CT image. Record the adjusted window width W and window level L. Based on the adjusted slope and intercept of each CT image in the group, the original values ​​of the pixels in the CT images in the group are converted into HU values, and the converted HU values ​​are filtered according to W and L. Convert the filtered HU values ​​of pixels in this set of CT images to grayscale values; The gray value threshold is determined based on W, L and the set gray value threshold ratio. Based on the gray values ​​of the pixels in the group of CT images and the gray value threshold, part or all of the region including the first circular hole in the group of CT images is binarized to obtain a group of binarized images.

3. The method as described in claim 2, characterized in that: After converting the filtered HU values ​​of the pixels in the group of CT images into grayscale values, the method further includes setting the grayscale values ​​of the pixels located in the preset bed board area in the group of CT images to 0 in order to filter out the noise brought by the bed board.

4. The method as described in claim 1, characterized in that: The first circular hole is located in the middle of the top layer of the circular hole model; determining the fuzzy center coordinates of the first circular hole in the set of binary images using the image features of each binarized image includes: For each binarized image, the row coordinates in the blur center coordinates are determined as follows: Based on the engineering length data of the top layer, the physical distance between adjacent pixels, and the set effective data rate threshold, determine the effective pixel count threshold of the top layer in a row; The effective pixel count of each row in the binarized image is counted from top to bottom, and the first row with an effective pixel count greater than the effective pixel count threshold is recorded as the starting row of the top layer. The row coordinates corresponding to the starting row are used as the row coordinates in the fuzzy center coordinates.

5. The method as described in claim 4, characterized in that: The step of determining the fuzzy center coordinates of the first circular hole in the set of binary images using the image features of each binarized image further includes: For each binarized image, the column coordinates in the blur center coordinates are determined as follows: Based on the engineering height data of the top layer and the physical distance, determine the number of pixels corresponding to the height of the circular hole model; Based on the position of the starting row and the number of pixels corresponding to the height of the circular hole model, the search range of the top layer in the height direction of the binarized image is determined; Count the number of valid pixels in each column within the search range, and find an intermediate column such that the difference between the sum of the number of valid pixels in all columns to the left of the intermediate column and the sum of the number of valid pixels in all columns to the right of the intermediate column is minimized. The column coordinates corresponding to the middle column are used as the column coordinates in the fuzzy center coordinates.

6. The method as described in claim 1, characterized in that: The step of extracting a target pixel block containing the first circular hole from the binarized image based on the blurred center coordinates and the engineering aperture characteristics of the circular hole model includes: The number of rows and columns contained in the target pixel block are determined based on the fuzzy center coordinates and the engineering aperture characteristics of the circular hole model.

7. The method as described in claim 1, characterized in that: The step of determining the center row coordinates and left and right edge data of the first circular hole based on the image features of the target pixel block includes: For each binarized image, the center row coordinates, left edge data, and right edge data of the first circular hole in the binarized image are determined as follows: Count the number of valid pixels in each row of the target pixel block in the binarized image; Traverse each row of the target pixel block and record the average row coordinates of the non-zero row with the most effective pixels as the center row coordinate data of the first circular hole; Count the number of valid pixels in each column of the target pixel block in the binarized image; Traverse each column of the target pixel block, determine the leftmost and rightmost columns where the number of effective pixels is less than the set minimum pixel count threshold, and record the column coordinates corresponding to the leftmost and rightmost columns as the left edge data and right edge data of the first circular hole, respectively.

8. The method as described in claim 7, characterized in that: The step of determining the coordinates of the center point of the first circular hole based on each binarized image, the row coordinates of the center of the first circular hole, and the data of the left and right edges includes: Select a binarized image where the left and right edge data of the first circular hole are not empty; The left and right edge data of the first circular hole in all selected binary images are summed and averaged to obtain the pixel column coordinates of the center point of the first circular hole. The pixel row coordinates of the center point of the first circular hole are obtained by summing the row coordinates of the center point of the first circular hole in all selected binary images and taking the average value. The image number of the center point of the first circular hole is obtained by summing the serial numbers of all selected binarized images and taking the average value.

9. A feature extraction device for a circular hole model, comprising: The memory and processor are characterized by: The memory is used to store the program for feature extraction of the circular hole model; The processor is configured to read the program for extracting features from a circular hole model and execute the circular hole model feature extraction method as described in any one of claims 1 to 8.

10. A non-transient computer-readable storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the circular hole model feature extraction method according to any one of claims 1 to 8 when running.

Images