Oral care monitoring imaging system and method

Abstract

The invention relates to the technical field of oral cavity imaging, in particular to an oral cavity nursing monitoring imaging system and method.The system image structure decomposition module, the local contour recognition module, the image change detection module, the abnormal area filling module and the color gamut dynamic regulation and control module are included. Functional areas are clearly divided, the area recognition precision of images is improved, particularly, in tooth boundary recognition, contrastive analysis of brightness and color differences is utilized, the recognizability of tooth contours and continuous frame tracking of tooth structure changes are remarkably enhanced, the dynamic deformation path of the tooth structure changes is displayed, the monitoring capacity of tiny changes is improved, and meanwhile, the recognition accuracy of the tooth contours is improved. The automatic completion technology is used for repairing the fracture boundary in the image and restoring the boundary continuity, and the color channel is adjusted to ensure the overall visual consistency of the image and avoid visual discordance, so that the image quality and the analysis accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of oral imaging, and particularly to an oral care monitoring imaging system and method. Background Art

[0002] The technical field of oral imaging includes image acquisition, processing, and analysis technologies related to oral health. The core content of this field is to obtain image information inside the oral cavity through imaging devices for diagnosis, treatment planning, and treatment process monitoring. Oral imaging technologies mainly include traditional X-ray imaging, digital imaging technologies, and optical imaging technologies, laser imaging technologies, etc. developed in recent years. With the progress of technology, oral imaging devices are gradually developing towards more accurate, non-invasive, and real-time directions, promoting the improvement of oral medicine in diagnostic accuracy and treatment effects.

[0003] Among them, an oral care monitoring imaging system refers to an imaging monitoring system used in the oral care process. This system mainly improves the imaging technology for health monitoring inside the oral cavity and uses digital imaging technology to monitor oral health in real time. Specifically, this system obtains high-resolution images inside the oral cavity through specific image acquisition devices and uses image processing technology to process the collected data to help users understand the oral health status in real time. This system also includes image display and analysis functions for visual display of oral care data to assist users in making oral care decisions. This system aims to improve the accuracy of oral care and provide users with more comprehensive oral health monitoring through imaging technology.

[0004] The fixed limitation of static images in the time dimension makes the image content only reflect the state at a single moment and unable to present the continuous change trend of teeth and related tissues, resulting in a lag in the perception of the initial manifestations of diseases. The tissue classification process relies on the intuitive differences in brightness distribution and boundaries and lacks in-depth analysis of the internal texture structure of regions, leading to low regional recognition accuracy and common problems such as overlapping boundaries and tissue misjudgment. During the image acquisition process, affected by factors such as posture deviation and light changes, boundary breakage often occurs, but existing methods do not have an automatic completion mechanism, resulting in damaged image integrity and reduced subsequent analysis value. In terms of the visual performance after image reconstruction, local regions appear abrupt due to the lack of color fusion processing, and there are obvious differences between the reconstructed regions and the original regions, affecting the naturalness and observation coherence of the images. Taking multiple shootings as an example, due to angle changes causing edge jumps, it is difficult to accurately compare the differences in the current images, limiting the structure tracking ability, indicating that there are obvious shortcomings in the existing technology in dealing with dynamic changes, boundary interruptions, and visual fusion. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose an oral care monitoring imaging system and method.

[0006] To achieve the above object, the present invention adopts the following technical solution: An oral care monitoring imaging system includes:

[0007] The oral care monitoring image frame sequence obtained by the image structure decomposition module extracts the tooth gray-scale change characteristics, the gum brightness distribution uniformity, and the soft tissue texture characteristics, differentiates the texture types of the image regions and delimits the boundary coordinates, and integrates the regional structure information to obtain an oral imaging partition annotation map;

[0008] The local contour recognition module extracts the brightness difference and color distribution on both sides of the boundary based on the tooth region in the oral imaging partition annotation map, recognizes the feature edges and circumscribes the tooth recognizable region to generate a tooth contour comparison distribution map;

[0009] The image change detection module extracts the continuous frame gray-scale and texture direction fluctuations based on the region indicated by the tooth contour comparison distribution map, tracks the change of the tooth contour boundary on the time axis to obtain a tooth structure dynamic deformation path map;

[0010] The abnormal area filling module extracts image blocks with consistent surrounding texture and continuous brightness based on the broken boundary in the dynamic deformation path map, performs contour reconstruction on the broken area to generate a continuous tooth contour segment map;

[0011] The color gamut dynamic regulation module extracts the color channel offset between the reconstructed area and the surrounding based on the continuous contour segment map, adjusts the color ratio and smooths the abrupt boundary to generate a corrected oral care monitoring imaging image.

[0012] As a further solution of the present invention, the oral imaging partition annotation map includes a tooth region annotation layer, a gum region annotation layer, a soft tissue region annotation layer, a boundary type annotation, a regional texture type label, and a partition number index. The tooth contour comparison distribution map includes a contour boundary line group, a boundary intensity distribution, a brightness anomaly mark, an edge direction vector, and a tooth recognition area mask. The tooth structure dynamic deformation path map includes a deformation trajectory line, a time axis mapping, a deformation trend layer, and a dynamic change area label. The continuous tooth contour segment map includes a contour reconstruction segment, a continuity repair mask, a texture fusion area, and a segment boundary line. The corrected oral care monitoring imaging image includes a color correction layer, a brightness equalization layer, and a reconstructed fusion area.

[0013] As a further solution of the present invention, the image structure decomposition module includes:

[0014] The image frame extraction sub-module obtains the image frame sequence in oral care monitoring, extracts the gray-scale jump value of the tooth surface area, the change range of the average brightness of the gum area, and the texture distribution direction of the oral soft tissue area in the image, samples and collects the regional image features, and establishes a regional image feature set;

[0015] The regional texture classification sub-module determines the texture type to which the image features belong based on the gray-scale jump value, the change range of the average brightness, and the texture distribution direction in the regional image feature set, and generates a result of classifying the image texture type;

[0016] The boundary structure calibration sub-module detects the boundary gray-scale change direction based on the edge position of the region in the image texture type classification result, extracts the gray-scale difference path of adjacent pixels, records the coordinates of continuous edge pixels and numbers them according to the region type, and generates an oral imaging partition annotation map.

[0017] As a further solution of the present invention, the local contour recognition module includes:

[0018] The brightness feature extraction sub-module extracts the distribution of image brightness values on both sides of the boundary line and the pixel density in the color channel based on the tooth boundary area defined in the oral imaging partition annotation map, integrates the pixel rows and columns according to the boundary direction, calculates the brightness difference value of the boundary, and generates boundary brightness difference data;

[0019] The edge anomaly localization sub-module extracts continuous boundary pixel columns based on the boundary section with a drastic change in brightness span in the boundary brightness difference value, compares them with adjacent pixel segments, determines the consistency between the jump direction and the main boundary direction, and screens out the edge paragraphs with stable continuous jump directions as abnormal structure areas, and generates a set of edge brightness jump positions;

[0020] The structure area circumscription sub-module calls the jump areas marked in the set of edge brightness jump positions, circumscribes the boundary closed areas with continuous edge jump directions and stable color distributions according to the boundary extension direction and the change trend of the comparison value, classifies the region types according to the pixel density level, summarizes the contour structure boundaries, and generates a tooth contour comparison distribution map.

[0021] As a further solution of the present invention, the specific calculation formula of the boundary brightness difference value is:

[0022]

[0023] where L d represents the boundary brightness difference value, represents the brightness value of the i-th pixel inside the boundary, represents the brightness value of the i-th pixel outside the boundary, n represents the total number of pixel points considered on both sides of the boundary line, i represents the index of the pixel point, and w irepresents the luminance gradient weight value of the i-th pixel on the boundary line in the boundary normal direction. The letter i associated with the summation symbol ∑ represents the summation over all considered pixel points.

[0024] As a further aspect of the present invention, the image change detection module includes:

[0025] The gray trend extraction sub-module extracts the gray value distribution of each frame region in the continuous image frame sequence based on the region indicated by the tooth contour comparison distribution map, calculates the average gray value between adjacent frames, identifies the continuous gray change trajectory segments, and establishes a regional gray change trend sequence;

[0026] The texture fluctuation monitoring sub-module detects the distribution of texture direction vectors in the corresponding region of the image frame based on the frame sequence region covered by the regional gray change trend sequence, extracts the texture set with prominent direction change rate, and screens the pixel blocks with similar vector change trends to obtain the texture direction fluctuation trajectory data;

[0027] The boundary path arrangement sub-module calls the coordinate sequence in the texture direction fluctuation trajectory data, matches the boundary pixel change positions in the frame sequence, tracks the boundary offset direction and connection order in consecutive frames, integrates the consecutive boundary position coordinates in chronological order and numbers the path segments to generate a dynamic deformation path map of the tooth structure.

[0028] As a further aspect of the present invention, the formula for calculating the average gray value between adjacent frames is specifically:

[0029]

[0030] where ΔG t is the average gray value between adjacent frames, N represents the number of sampling frames for the change in the average gray value, G t is the gray value of the image frame region at time point t, x i,j is the coordinate of the j-th pixel point in the i-th frame, M is the total number of pixels in the image, and G t+i is the gray value of the (t + i)-th frame.

[0031] As a further aspect of the present invention, the abnormal area filling module includes:

[0032] The boundary break extraction sub-module extracts the contour pixel point sequence based on the boundary interruption region in the dynamic deformation path map of the tooth structure, detects the gray interval and extension direction, screens the pixel break points and records the positions and numbers to generate a boundary interruption positioning data set;

[0033] The image block screening and reconstruction sub-module calls the coordinate information of the boundary interruption positioning data set, obtains the texture direction and brightness continuity of adjacent image blocks, calculates the direction consistency rate and brightness difference, screens the image blocks to construct a contour filling structure, and generates an image filling and splicing segment set;

[0034] The contour area embedding sub-module writes the image content into the broken coordinate area based on the structure position and direction identifier in the image filling and splicing segment set, compares the brightness gradient of the surrounding pixels and corrects the gray difference, unifies the edge pixel density, and generates a continuous tooth contour segment map.

[0035] As a further solution of the present invention, the abnormal area filling module includes:

[0036] The color offset extraction sub-module extracts the red, green, and blue channel values based on the color difference between the reconstructed segment and the surrounding area in the continuous tooth contour segment map, calculates the channel mean difference of adjacent areas and screens the pixel points exceeding the offset detection threshold, aggregates and marks the channel difference distribution, and generates a color channel offset range;

[0037] The channel ratio reset sub-module calls the pixel point channel difference in the color channel offset range, calculates the brightness occupancy change coefficient of the channel in the area, adjusts the channel value according to the channel ratio reference value, outputs a channel occupancy structure diagram, and obtains the color channel regulation ratio;

[0038] The boundary brightness leveling sub-module compares the pixel brightness of the reconstructed area boundary and the adjacent area based on the channel value in the color channel regulation ratio, judges the situation where the threshold boundary brightness difference exceeds the range, performs brightness smoothing processing on the offset boundary and synthesizes the image, and generates an oral care monitoring imaging correction image.

[0039] An oral care monitoring imaging method includes the following steps:

[0040] S1: Obtain an oral care monitoring image frame sequence, collect a front view of the tooth surface, an image of the lingual area, and an oral endoscope image, extract the tooth gray scale jump amplitude, judge the gingival brightness balance degree, judge the soft tissue texture direction, screen the areas meeting the characteristics, extract the corresponding boundary coordinates, and generate an oral imaging partition annotation map;

[0041] S2: Based on the tooth boundary in the oral imaging partition annotation map, extract the brightness and color channel density values on both sides of the boundary, screen the mutation segments, judge the brightness extension direction and perform pixel aggregation, and generate a tooth contour comparison distribution map;

[0042] S3: Based on the structure area in the tooth contour comparison distribution map, extract the brightness performance of consecutive frames, judge the inter-frame change trend and texture direction coherence, screen the continuously changing segments and connect the boundary trajectories, and generate a tooth structure dynamic deformation path map;

[0043] S4: Based on the breakpoint positions in the dynamic deformation path diagram of the tooth structure, capture image blocks with consistent texture directions and stable brightness changes, determine whether the edge directions are continuous, and insert them into the contour line to generate a continuous fragment diagram of the tooth contour;

[0044] S5: Based on the inserted area in the continuous fragment diagram of the tooth contour, extract the color channel deviation values, determine whether they exceed the limit, perform proportional adjustment on the abnormal channels, recombine the image and smooth the jumping brightness to generate a corrected image for oral care monitoring imaging.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0046] In the present invention, by highly integrating the structural information in the image, clearly dividing the functional areas, the regional recognition accuracy of the image is improved. Especially in the recognition of the tooth boundary, by using the comparative analysis of the brightness and color differences, the recognizability of the tooth contour is significantly enhanced. The continuous frame tracking of the tooth structure changes shows its dynamic deformation path, increasing the monitoring ability for minor changes. At the same time, the automatic completion technology repairs the broken boundaries in the image, restores the boundary continuity, and ensures the overall visual consistency of the image by adjusting the color channels, avoiding visual disharmony and improving the image quality and analysis accuracy. Description of the Drawings

[0047] Figure 1 is the system flow chart of the present invention;

[0048] Figure 2 is the flow chart of the sub-module of the present invention;

[0049] Figure 3 is the flow chart of the method steps of the present invention. Detailed Embodiments

[0050] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0051] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present invention. In addition, in the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0052] Please refer to Figure 1 , an oral care monitoring imaging system includes:

[0053] The image structure decomposition module obtains the image frame sequence in oral care monitoring, extracts the gray-scale jump features of the tooth surface area, the brightness balance features of the gingival area, and the texture features of the oral soft tissue area, calibrates the areas with boundary structure jumps in the image, differentiates the regional texture types, and delimits the boundary range, integrates the boundary coordinates and the regional types to obtain an oral imaging partition annotation map;

[0054] The local contour recognition module extracts the brightness distribution and color density on both sides of the boundary line based on the tooth boundary area delimited in the oral imaging partition annotation map, identifies the edge segments with abnormal brightness, combines the boundary direction and the comparison extension trend, circles the tooth recognizable area, and summarizes the contour structure to generate a tooth contour comparison distribution map;

[0055] The image change detection module extracts the gray-scale distribution change trend of the area in the continuous image frame sequence based on the area indicated by the tooth contour comparison distribution map, monitors the fluctuation features of the texture direction, correlates and arranges the boundary trajectories of the changed areas on the image time axis to form a continuous spatial change path, and obtains a tooth structure dynamic deformation path map;

[0056] The abnormal area filling module extracts the image blocks with good brightness continuity of the texture direction in the adjacent areas based on the boundary interruption areas shown in the tooth structure dynamic deformation path map, performs contour extension and reconstruction on the fractured areas, and reinserts the processed image blocks into the original contour range to generate a tooth contour continuous segment map;

[0057] The color gamut dynamic regulation module extracts the offset range of the color channels based on the color distribution differences between the reconstructed segments and the surrounding areas in the tooth contour continuous segment map, resets the channel ratio, unifies the regional color tones, performs brightness smoothing on the color mutation boundaries, synthesizes the image, and generates an oral care monitoring imaging correction image.

[0058] The oral imaging partition annotation map includes a tooth area annotation layer, a gingival area annotation layer, a soft tissue area annotation layer, a boundary type annotation, a regional texture type label, and a partition number index. The tooth contour comparison distribution map includes a contour boundary line group, a boundary intensity distribution, a brightness abnormality mark, an edge direction vector, and a tooth recognition area mask. The tooth structure dynamic deformation path map includes a deformation trajectory line, a time axis mapping, a deformation trend layer, and a dynamic change area label. The tooth contour continuous segment map includes a contour reconstruction segment, a continuity repair mask, a texture fusion area, and a segment boundary line. The oral care monitoring imaging correction image includes a color correction layer, a brightness balance layer, and a reconstruction fusion area.

[0059] Please refer to Figure 2 , the image structure decomposition module includes:

[0060] The image frame extraction sub-module obtains the image frame sequence in oral care monitoring, extracts the gray-scale jump value of the tooth surface area, the change range of the average brightness of the gum area, and the texture distribution direction of the oral soft tissue area in the image, samples and aggregates the regional image features, and establishes a regional image feature set;

[0061] First, several consecutive and clear frames of images on the time axis should be extracted according to the image number, and the image areas corresponding to the teeth, gums, and oral soft tissues should be determined in each frame. The tooth area can be preliminarily delimited according to the position of significant changes in the brightness gradient in the image. For example, if there is a continuous edge contour with a gray-scale value jump in the middle section of the lower part of the image, it can be recognized as the tooth surface area. Subsequently, the gray-scale difference between adjacent pixels in each row should be extracted pixel by pixel in this area, and the proportion of the number of jump points should be counted to obtain the density of the jump distribution. If multiple consecutive jump rows are found concentrated in the same area in an image, this area can be confirmed as the active surface layer of the tooth, and the jump density value of this area should be recorded as the basis for subsequent image partition judgment. Then, switch to the gum area. Determine whether the brightness is stable by detecting the mean distribution of the red component in the middle and upper sections of the image. If the change range of the mean brightness of each column in this area is small and relatively continuous, it can be marked as a brightness-balanced area, and its average brightness value and fluctuation range should be recorded. Then, expand to identify the soft tissue area on both sides of the image. By setting several small blocks at equal intervals to extract the area and calculating the change trend of the gray-scale value in the horizontal and vertical directions in turn, determine whether the texture has directionality according to the change amplitude in different directions. If the change trend shows an obvious directional distribution, the main texture direction of this area can be marked as horizontal or vertical, and the regional index value corresponding to the texture direction should be recorded according to the image coordinate position. After establishing the image feature records in the above three types of areas, the recognition results of the jump distribution, average brightness, and texture direction should be corresponding to the spatial positions in the image frame and integrated into a regional image feature set.

[0062] The regional texture classification sub-module determines the texture type to which the image features belong based on the gray-scale jump value, the change range of the average brightness, and the texture distribution direction in the regional image feature set, and generates the image texture type classification result;

[0063] First, extract the sequence of gray - level differences between adjacent pixels in each region, and determine whether there is a continuous gray - level jump. If the gray - level differences of three consecutive groups in the region exceed 5 and have the same direction, it is considered that the region has the gray - level jump feature, and record the number of jump groups as the classification basis. For example, if the pixel gray - levels in the tooth region are 100, 108, 115, 123, 130, the differences are 8, 7, 8, 7, which meet the jump - set feature. Subsequently, extract the luminance channel values of all pixels in the region, calculate the mean value, and obtain the luminance change range. If the difference between the maximum and minimum luminance means is less than 10, it is marked as a region with convergent luminance change, and this feature is common in the gingiva region. Then, divide the image into multiple small region blocks, and judge whether the texture arrangement direction in each block is concentrated. If the main directions are concentrated in a few directions such as horizontal or vertical, it is marked as a region with a clear direction. If the texture is distributed dispersedly in multiple directions, it is considered that there is no dominant direction, and this feature is usually seen in the soft - tissue region. Cross - judge the above three features. If the number of jump groups is greater than 5, the luminance change range is greater than 15, and the texture direction concentration exceeds 40%, it is delimited as the tooth region. If the jump value is low, the luminance change range is small, and the direction is concentrated, it is delimited as the gingiva region. If there are no significant features in each item, it is delimited as the soft - tissue region, complete the classification output of each region, and generate the result of the image texture type division.

[0064] Based on the edge positions of the regions in the result of the image texture type division, the boundary structure calibration sub - module detects the boundary gray - level change direction, extracts the path of the gray - level differences between adjacent pixels, records the coordinates of the edge continuous pixels and numbers them according to the region type, and generates an oral imaging partition annotation map.

[0065] Based on the edge positions of the regions in the image texture type division result, it is necessary to first extract the boundary starting points of each texture type region in the image. The method is to determine whether there is a difference in the type numbers between the current pixel and its upper, lower, left, and right pixels. If there is, record the current pixel as a boundary point and extract its coordinate information. For example, in the region where the tooth and the gum meet, a dense band of boundary pixels is often formed due to the sudden change in gray level, and an initial boundary index can be formed based on this. Next, it is necessary to detect the gray level change direction of these boundary points, extract the gray level values of consecutive boundary pixels in the horizontal and vertical directions, and compare their gray level change trends. If there is a continuous upward or downward trend exceeding three pixels, then this boundary segment is marked as a section with consistent direction. For example, if the gray level of the gum edge decreases from 110 to 85 and remains continuous for three groups, then it is determined that this direction is a stable downward path. After that, based on the boundary points with stable directions, further extract the gray level difference path of adjacent pixels, extend one pixel layer inward and outward from the edge, calculate the gray level difference between the current pixel and its adjacent points. If the fluctuation range of the difference sequence is within the specified interval and the trend is consistent, then this path is recognized as an effective edge segment, and the coordinates of all pixels in this path are extracted. Finally, integrate the coordinate sets according to the starting point, ending point, and path number of each boundary segment, and assign independent numbers to the tooth, gum, and soft tissue regions respectively, and finally form multiple groups of complete contour paths to generate an oral imaging partition annotation map.

[0066] Please refer to Figure 2 , the local contour recognition module includes:

[0067] The brightness feature extraction sub-module extracts the image brightness value distribution on both sides of the boundary line and the pixel density in the color channel based on the tooth boundary region defined in the oral imaging partition annotation map, integrates the pixel rows and columns according to the boundary direction, calculates the brightness difference value of the boundary, and generates boundary brightness difference data;

[0068] The specific calculation formula for the brightness difference value of the boundary is:

[0069]

[0070] Among them, L d represents the brightness difference value of the boundary, represents the brightness value of the i-th pixel inside the boundary, represents the brightness value of the i-th pixel outside the boundary, n represents the total number of pixel points considered on both sides of the boundary line, i represents the index of the pixel point, w i represents the brightness gradient weight value of the i-th pixel on the boundary line in the boundary normal direction, and the letter i associated with the summation symbol ∑ represents the summation over all considered pixel points;

[0071] Suppose the measured data of the first 3 boundary pixel points are as follows:

[0072] I in=[210,208,205],I out =[190,186,184];

[0073] μ in =(210+208+205) / 3=207.67;

[0074] Weight value calculation:

[0075] w1=(52-45) / (52-45)=1.0;

[0076] w2=(48-45) / (52-45)=0.429;

[0077] w3=(46-45) / (52-45)=0.143; Calculate item by item as follows:

[0078] The first pixel item:

[0079]

[0080] w1=0.25;

[0081]

[0082] The second pixel item:

[0083]

[0084] w2=0.50;

[0085]

[0086] The third pixel item:

[0087]

[0088] w3=0.75;

[0089]

[0090] The average calculation is:

[0091]

[0092] The results show that this value represents the intensity of brightness change in the tooth boundary area in the image, and is a feature used to judge the degree of boundary prominence in subsequent image segmentation and boundary extraction. The higher the value, the stronger the grayscale contrast inside and outside the boundary, and the clearer the image contour, which is suitable for boundary judgment and structure recognition. When the value is low, it may reflect that the grayscale transition in the edge area is not obvious, and further analysis is required in combination with other features.

[0093] Based on the boundary section with a drastic change in brightness span among the boundary brightness difference values, the edge anomaly localization sub-module extracts continuous boundary pixel columns, compares them with adjacent pixel segments, judges the consistency between the jump direction and the main boundary direction, filters out the edge paragraphs with stable continuous jump directions as the abnormal structure area, and generates the edge brightness jump position set;

[0094] First, determine the starting coordinates and boundary orientation of the boundary area. Sample the brightness value differences between each pixel point in the boundary section and its adjacent pixels row by row or column by column, and extract the pixel points with a brightness span greater than the boundary detection reference value. The reference value is set to 10. For example, if the brightness of a certain pixel is 135 and the adjacent pixels are 120, 122, and 128, the differences are 15, 13, and 7. The first two are identified as drastic change points and included in the initial edge detection band. Then, extract the pixel columns with a continuous jump trend among them. If the length of the continuous jump columns is not less than 3 columns, they are marked as continuous boundary columns to form a preliminary edge band structure. On the edge band, continue to extract the brightness difference trends of the adjacent pixels on both sides of the edge column to judge whether the jump direction is consistent with the main boundary direction. The judgment condition is that at least two of the three consecutive columns have the same direction. The direction consistency judgment threshold is 66%. For example, if the brightness of the boundary column is 142, 130, 118, and the adjacent column is 110, 124, 135, when judging that the jump is in the same direction, keep this column. Subsequently, traverse the continuous jump columns, filter out the structural segments with continuous and non-reverse jump directions as the effective jump segments. Define the ratio of the sum of the brightness jump amplitudes in the structural segment to the segment length as the jump density value. If this value is greater than 5, then this segment is considered as the area with concentrated brightness changes. All effective jump segments are marked, numbered, and archived in the abnormal edge segment set. Finally, generate a coordinate mapping table, output the edge jump distribution positions, and generate the edge brightness jump position set.

[0095] The structure area circumscribing sub-module calls the jump areas marked in the edge brightness jump position set, circumscribes the boundary closed areas with continuous edge jump directions and stable color distributions according to the boundary extension direction and the change trend of the comparison value, classifies the area types according to the pixel density level, summarizes the contour structure boundaries, and generates the tooth contour comparison distribution map;

[0096] First, the starting and ending coordinates of each jump area are collected, and the jump points with continuous boundary features are connected in series into a line segment structure, and the boundary extension direction is marked. In actual operation, if a jump segment is composed of coordinate points A(20,45), B(21,46), and C(22,47), the extension direction is tilted to the upper right, that is, the main direction is oblique extension. Then, the adjacent pixel columns of the line segment in the extension direction are extracted, and the pixel brightness change trend in the adjacent columns is counted and the consistency with the jump direction is judged. If the change direction is the same and the change amplitude maintains increasing or decreasing continuity, the pixel column is included in the jump area boundary extension range. When performing trend judgment, a direction consistency judgment threshold must be set. The threshold is set to the same change trend in two of the three consecutive columns, and the extension direction is judged to be established. After completing the construction of the preliminary closed area, the color distribution structure of the pixels in the area is continued to be extracted. The channel mean of each area is calculated according to the three channels of red, green, and blue, and any two adjacent sub-areas in the area are judged. Whether the channel mean difference between blocks is within the acceptable range, the range is set as a single channel difference less than 15. If this condition is met, it is considered to be a stable color distribution area. For example, the channel means in a certain area are R=112, G=115, B=117, and the adjacent areas are R=108, G=113, B=120, then the mean differences are 4, 2, and 3, respectively, all of which do not exceed the offset threshold and are marked as stable segments. Finally, a closed structure area with stable color and consistent jump direction is formed. After the closed area is constructed, the coordinates of the pixel points in all areas are extracted and the number of pixels per unit area is calculated. Statistics are performed in units of 10×10 pixel blocks. If the total number of pixels in a single block exceeds 80, it is marked as a high-density area, otherwise it is marked as a low-density area. After the density classification is completed, it is numbered according to high, medium, and low levels. After numbering, all regional density labels, coordinate data, and boundary structures are integrated into a unified boundary distribution structure table, and the structure table is output to generate a tooth contour comparison distribution map.

[0097] See also Figure 2 , the image change detection module includes:

[0098] The grayscale trend extraction submodule extracts the grayscale value distribution of each frame area in the continuous image frame sequence based on the area indicated by the tooth contour contrast distribution map, calculates the grayscale average value between adjacent frames, identifies the continuous grayscale change trajectory fragments, and establishes the regional grayscale change trend sequence;

[0099] The calculation formula of the grayscale average value between adjacent frames is:

[0100]

[0101] Among them, ΔG t is the grayscale average between adjacent frames, N represents the number of sampling frames for which the grayscale average changes, G t is the grayscale value of the image frame area at time point t, xi,j is the coordinate of the j-th pixel in the i-th frame, M is the total number of pixels in the image, and G t+i is the grayscale value of the (t + i)-th frame;

[0102] Suppose there is a set of consecutive image frames currently. The grayscale value of each frame can be obtained through an image processing tool. To calculate the grayscale change at time point t, it is first necessary to calculate the grayscale value change between consecutive frames;

[0103] Select N = 5 consecutive frames of images for calculation. Suppose their grayscale values are respectively: G t = 120, G t+1 = 125, G t+2 = 128, G t+3 = 130, G t+4 = 135;

[0104] Calculate the absolute difference of the grayscale value change:

[0105] |G t+1 - G t | = |125 - 120| = 5, |G t+2 - G t | = |128 - 120| = 8, |G t+3 - G t | = |130 - 120| = 10, |G t+4 - G t | = |135 - 120| = 15;

[0106] Calculate the pixel position change: The measure of the pixel position change in each frame is usually obtained through coordinate calculation. Suppose x i,j represents the coordinate of the j-th pixel in the i-th frame, and the pixel coordinate change within each frame is relatively small. The coordinate difference is obtained by calculating the coordinate offset of each pixel frame by frame, and further the change of the pixel position is calculated through the Euclidean distance. Suppose that among all pixels, the sum of the squares of the pixel offsets of all frames is 2000.

[0107] Calculate the sum of the squares of the coordinate offsets:

[0108]

[0109] Now these values can be substituted into the formula for calculation:

[0110]

[0111] The result indicates that the gray - scale change amount at time point t is 339.23, which means the magnitude of the gray - scale change amount in the image area within the continuous frame sequence being calculated. This value is relatively large, indicating that the gray - scale change between image frames is very obvious. According to the calculation result of the formula, this value directly reflects the gray - scale trend change of the image during this period of time.

[0112] The texture fluctuation monitoring sub - module detects the distribution of texture direction vectors in the corresponding area of the image frame based on the frame sequence area covered by the regional gray - scale change trend sequence, extracts the texture set with prominent direction change rate, and screens the pixel blocks with similar vector change trends to obtain the texture direction fluctuation trajectory data.

[0113] First, extract the texture change vectors in each frame of the image. For each pixel point, calculate the gray - scale change direction of its surrounding area. The gray - scale change trend is calculated by comparing the gray - scale value differences of adjacent pixel points. If the gray - scale difference of three consecutive columns of pixels is greater than a certain set threshold (for example, 10), it is considered that there is significant texture change in this area. Then, mark these areas with significant changes and use their gray - scale change trends as vectors. Determine the dominant direction of the texture by calculating the direction and length of the vectors. If the texture direction in a certain image area changes successively to 0 degrees, 45 degrees, and 90 degrees, it indicates that there is an obvious direction fluctuation in the texture of this area. Furthermore, extract the texture set with the most significant changes. The texture direction change frequency in these sets is relatively high. In the example, if the texture direction in an area gradually changes from the horizontal line to the vertical line, it indicates that there is a significant fluctuation in the texture direction of this area. Next, further screen the texture direction vectors in the area. For these pixel blocks with a relatively high texture direction change frequency, calculate the standard deviation of the change of the texture direction vectors within each pixel block, and judge the stability of the vector change according to this standard deviation. If the standard deviation is greater than the set threshold (for example, set to 5 degrees), it is considered that the texture change trend within this pixel block is unstable, and it is excluded from the stable texture area. The texture areas screened in this way are usually texture areas with relatively consistent direction changes and stable changes. For example, the tooth surface usually shows relatively straight and continuous local texture directions, in contrast to the scattered and irregular arrangement in the soft tissue area. Finally, screen out the texture areas with relatively consistent direction changes and small fluctuations, merge the pixel blocks in these areas into continuous texture direction fluctuation trajectories, and assign numbers to each trajectory to ensure that each texture fluctuation trajectory has a clear record of spatial position and direction. Finally, obtain the texture direction fluctuation trajectory data.

[0114] The boundary path arrangement sub - module calls the coordinate sequence in the texture direction fluctuation trajectory data, matches the boundary pixel change positions in the frame sequence, tracks the boundary offset direction and connection order in consecutive frames, integrates the consecutive boundary position coordinates in chronological order and numbers the path segments to generate the dynamic deformation path map of the tooth structure.

[0115] First, extract the boundary position points and texture change regions in each frame of the image, mark the position changes of the boundary coordinates in consecutive image frames, compare the displacements between the current frame and the previous frame boundary positions frame by frame based on the texture direction vectors in each frame, calculate the offset of each pixel point. If the offset value is greater than the preset offset detection threshold (for example, set to 5 pixels), it is marked as a valid boundary offset region. Suppose in a certain image frame, the coordinates of a pixel point change from (30, 40) to (35, 45), then the calculated offset is 5 pixels and the direction is the same, meeting the offset condition. Subsequently, determine whether the jump direction of each boundary point is consistent with its main direction, and compare the similarity between the vector direction of consecutive jump points and the main direction. If the main direction offset is less than the set threshold (such as the deviation is less than 15 degrees), it is considered that the jump direction in this region is consistent. If the main direction deviation is greater than this value, it is considered that the jump direction is inconsistent with the main boundary trend, and this region is not regarded as a valid boundary. Then, for all boundary points with valid offsets, integrate their position data in chronological order to construct a boundary trajectory. If the displacement values in a certain region are consistent and the direction is the same in consecutive frames, this displacement is regarded as a stable trajectory. Number each boundary position on the trajectory, number the path segments in sequence. Each path segment starts from the first offset point and records until the end of this region. The path segment number is the path identifier for subsequent analysis. Finally, combine all the path segment data into a complete path map, output this map and generate a dynamic deformation path map of the tooth structure.

[0116] Please refer to Figure 2 , the abnormal area filling module includes:

[0117] Based on the boundary interruption regions in the dynamic deformation path map of the tooth structure, the boundary fracture extraction sub-module extracts the sequence of contour pixel points, detects the gray interval and the extension direction, filters the pixel fracture points and records their positions and numbers, generating a boundary interruption location data set;

[0118] First, extract the sequence of contour pixel points of these regions, scan the marked boundary positions in the image frame by frame, determine the spatial coordinates of each boundary pixel point and the gray value difference of its adjacent positions. If the gray difference between adjacent pixel points exceeds the set threshold (for example, 15), it is considered that there is an obvious gray jump in this region, indicating that the boundary is broken, and record the pixel coordinates of this position; Next, detect the gray interval and the extension direction. In each boundary break region, analyze the gray change trend on both sides of the break point. If the gray value change in the left and right neighborhoods of a certain break point shows a large jump and the jump direction is stable, then judge this position as a break point. For example, if the gray values of the neighborhood pixels at a certain position are 100, 130, and 160 in sequence, and the gray difference is 30, it is regarded as a large interval, and this position is judged as the boundary break; Through the comparison of consecutive frames, identify and track the extension direction of the break point, compare all break positions with adjacent pixel points, judge the boundary extension trend, and sort them according to their relative positions to ensure the correct tracking of the break positions. If the boundary break position shows a trend from the upper left to the lower right, record it as the extension direction of this break and number it, and mark the serial number of this break point; Finally, screen out all pixel points that meet the break characteristics and their extension paths, summarize and number the positions and extension directions of each boundary break region, and generate a boundary interruption positioning data set. This data set records the specific positions and corresponding numbers of each break point for subsequent processing and analysis.

[0119] The image block screening and reconstruction sub-module calls the coordinate information of the boundary interruption positioning data set, obtains the texture direction and brightness continuity of adjacent image blocks, calculates the direction consistency rate and brightness difference, screens image blocks to construct a contour filling structure, and generates an image filling and splicing segment set;

[0120] First, extract the adjacent image patches at the fracture position. The pixel sequence within each image patch contains its texture direction and brightness continuity. By analyzing the texture direction vectors and gray-scale change trends of adjacent pixel patches, calculate the direction consistency rate and brightness difference of each pixel patch. In actual operation, for two adjacent image patches, if their texture direction changes slightly and the brightness difference is less than a set threshold (such as 10 gray-scale units), it is considered that they are relatively consistent in texture and brightness and meet the conditions for constructing a continuous contour. If the texture direction changes from the upper left corner to the lower right corner, calculate the amount of this direction change and count it; Next, select the image patches with high texture direction consistency and small brightness differences as the filling areas. These areas have strong continuity with the areas where the boundary is fractured and are suitable as candidate areas for boundary filling. Calculate the texture consistency rate of each image patch. By comparing the direction differences and brightness differences of pixel points in adjacent image patches, if the texture direction difference between adjacent patches is less than 15 degrees and the brightness difference is less than 20 gray-scale units, then this image patch is recognized as a suitable filling area. If the number of image patches that meet the conditions is sufficient, select these patches for splicing to form a filling structure for the contour; Finally, reconstruct the continuity of the fractured boundary through the selected image patches and integrate these spliced image patches into a filling fragment set. Each image patch in the spliced fragment set is numbered in sequence, and each number represents the position and order of the image patch during the filling process, and finally generate an image filling and splicing fragment set.

[0121] Based on the structural position and direction identifier in the image filling and splicing fragment set, the contour area embedding sub-module writes the image content into the fractured coordinate area, compares the brightness gradients of surrounding pixels and corrects the gray-scale difference, unifies the edge pixel density, and generates a continuous fragment map of the tooth contour;

[0122] First, extract the position coordinates and corresponding directions of each filling segment. The image segment is compared with the boundary of the surrounding area according to its filling direction to determine the docking direction and angle of the filling segment. Then, compare the gray-scale difference between the filling segment and the boundary of the original image, extract the gray-scale values near the boundary, and calculate the gray-scale gradient of the boundary pixel points. If the gray-scale of the boundary of the filling segment differs significantly from the surrounding pixels, gray-scale smoothing processing is required. The gray-scale difference threshold is set to 20. If the boundary gray-scale difference exceeds this threshold, the boundary is corrected to bring the gray-scale values of the surrounding area closer to ensure continuity. At the same time, calculate the brightness continuity between the filling segment and the surrounding area, analyze the difference in the brightness distribution between the filling segment and the surrounding area. If the brightness difference is large (for example, the brightness difference exceeds 25 gray levels), the brightness of the filling segment is adjusted to balance its brightness value, reducing the brightness difference between the filling segment and the surrounding area and avoiding the formation of an obvious seam. Then, according to the texture structure of the surrounding area, adjust the texture density of the pixels in the filling segment to ensure a match with the texture of the original image. In areas with a higher pixel density, the pixels in the filling segment are refined to reduce the local density, making the image transition smoother. Finally, unify the density of the edge pixels to ensure that the pixel density of the entire image edge is within a reasonable range, making the image transition natural and seamless, and finally generating a continuous segment map of the tooth contour.

[0123] Please refer to Figure 2 , the abnormal area filling module includes:

[0124] Based on the color difference between the reconstructed segment and the surrounding area in the continuous segment map of the tooth contour, the color offset extraction sub-module extracts the red, green, and blue channel values, calculates the channel mean difference of adjacent areas and filters out the pixel points that exceed the offset detection threshold, aggregates and marks the channel difference distribution, and generates the color channel offset range.

[0125] First, extract the pixel values of the red, green, and blue (RGB) channels of each reconstructed segment and its adjacent area, calculate the RGB values of each pixel point, and compare the channel means of the adjacent areas. If the difference between the RGB channel means of a certain segment and the means of the adjacent area exceeds a preset offset detection threshold (e.g., set to 10), it is considered that the color of this segment has an offset. Then, by traversing each pixel point in the image, extract all pixel points with large color differences and perform screening. If the mean value of the red channel in a certain area is 120, while the mean value of the adjacent area is 105, and their difference is greater than the offset detection threshold, then mark this point as a color offset point and record the channel difference of this point. Then, integrate the color differences of all offset points into a region to generate a channel difference distribution map, record the specific positions of each offset point and their corresponding channel difference values, and finally generate the color channel offset range. For example, in a certain image area, the mean value of the red channel is 150, the mean value of the green channel is 130, the mean value of the blue channel is 120, and the channel means of its adjacent area are 145, 125, and 115 respectively. After calculation, the differences are 5, 5, and 5 respectively. These difference values are lower than the offset detection threshold, so these areas are not regarded as offset areas. On the contrary, if the difference is greater than the threshold, it is marked as a color offset area.

[0126] The channel ratio reset sub-module calls the pixel point channel differences in the color channel offset range, calculates the change coefficient of the brightness occupancy ratio of the channel in the area, adjusts the channel value according to the channel ratio reference value, outputs the channel occupancy structure diagram, and obtains the color channel regulation ratio;

[0127] First, extract the RGB channel values of each offset region, calculate the mean difference between the RGB channels of each pixel in each region and those of the adjacent region. If the difference exceeds the set offset detection threshold (for example, set to 15), then mark this pixel as a color offset point and conduct aggregation. Count the number and location of the offset points. Then, according to the distribution of the offset points within the region, calculate the coefficient of change in brightness proportion within the region. Specifically, by comparing the brightness distributions of the red, green, and blue channels in this region, calculate the proportion changes of each channel within the region. If the proportion of a certain channel has a significant change compared to the reference value (for example, the proportion change of the green channel is greater than 20%), then it is considered that the color distribution in this region has an obvious offset and needs to be adjusted. Then, based on the channel proportion reference value of each region, adjust each channel of the offset region. The adjustment method is to scale the pixel values of each channel to within the reference value range to ensure that the adjusted image has the same color as the surrounding regions. If the proportion of the red channel in a certain region is too high, then reduce the brightness of this channel until it returns to the reference value. Finally, generate a channel proportion structure diagram, which shows the proportion of each channel in different regions, helping to monitor the specific distribution of color offset. Ultimately, obtain the color channel regulation ratio, that is, the adjustment ratio of different channels in each region, for use in subsequent image optimization processes.

[0128] Based on the channel values in the color channel regulation ratio, the boundary brightness leveling sub-module compares the pixel brightness of the reconstructed region boundary with that of the adjacent region, determines the situation where the threshold boundary brightness difference exceeds the range, performs brightness smoothing on the offset boundary and synthesizes the image to generate an oral care monitoring imaging corrected image.

[0129] First, extract the pixel brightness values of the reconstructed region boundary and the adjacent region. For each pixel point, calculate the difference between its brightness and the pixel brightness of the adjacent region. Assume that the brightness of the pixel on the region boundary is 120, while the brightness value of the adjacent region is 100. By calculating the brightness difference, if this difference exceeds the set brightness threshold (for example, the threshold is 10), then mark it as an offset point with too large a brightness difference. Then, analyze the difference between the brightness of the boundary region and that of the adjacent region. If it exceeds the set range, then perform brightness smoothing. The brightness smoothing method is to adjust the brightness of this boundary pixel according to the brightness difference between this region and the adjacent region through weighted average or interpolation method to make it close to the brightness of the adjacent region until the brightness balance requirement is met. During the continuous processing, if the color or brightness difference between this boundary and the adjacent region is greater than the set range, then gradually adjust the color and brightness values of the boundary region to ensure the overall coherence of the image. Finally, the smoothed image will be coordinated with the color and brightness of the surrounding regions to generate a coherent and natural oral care monitoring imaging image. Finally, the generated image is the oral care monitoring imaging corrected image, which provides a clearer and more coherent oral health image for the user to assist in further diagnosis and treatment.

[0130] Please refer to Figure 3 , an oral care monitoring imaging method, comprising the following steps:

[0131] S1: Obtain an oral care monitoring image frame sequence, collect front views of tooth surfaces, images of the lingual region, and oral endoscope images, extract the amplitude of tooth gray-scale jumps, judge the evenness of gingival brightness, judge the direction of soft tissue texture, screen the regions that meet the features, extract the corresponding boundary coordinates, and generate an oral imaging partition annotation map;

[0132] S2: Based on the tooth boundaries in the oral imaging partition annotation map, extract the density values of brightness and color channels on both sides of the boundaries, screen the mutation segments, judge the brightness extension direction and perform pixel aggregation to generate a tooth contour comparison distribution map;

[0133] S3: Based on the structural regions in the tooth contour comparison distribution map, extract the brightness performance of consecutive frames, judge the inter-frame change trend and the coherence of the texture direction, screen the continuously changing segments and connect the boundary trajectories to generate a tooth structure dynamic deformation path map;

[0134] S4: Based on the breakpoint positions in the tooth structure dynamic deformation path map, capture the image blocks with consistent texture directions and stable brightness changes, judge whether the edge directions are continuous, and insert them into the contour line to generate a tooth contour continuous segment map;

[0135] S5: Based on the inserted regions in the tooth contour continuous segment map, extract the color channel deviation values, judge whether they exceed the limit, perform proportional adjustment on the abnormal channels, recombine the image and smooth the jumping brightness to generate an oral care monitoring imaging correction image.

[0136] The above is only a preferred embodiment of the present invention, and it does not limit the present invention in other forms. Any person skilled in the relevant art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An oral care monitoring imaging system, characterized in that, The system comprises: The image structure decomposition module obtains the oral care monitoring image frame sequence, extracts the tooth grayscale change characteristics, gum brightness distribution balance, and soft tissue texture characteristics, distinguishes the image area texture type and delineates the boundary coordinates, integrates the regional structure information, and obtains the oral imaging partition annotation map; The local contour recognition module extracts the brightness difference and color distribution on both sides of the boundary based on the tooth area in the oral imaging partition annotation map, identifies the feature edge and delineates the identifiable tooth area, and generates a tooth contour comparison distribution map; The image change detection module extracts the grayscale and texture direction fluctuations of continuous frames based on the area indicated by the tooth contour contrast distribution map, tracks the changes of the tooth contour boundary on the time axis, and obtains a dynamic deformation path map of the tooth structure; The abnormal area filling module extracts image blocks with consistent surrounding texture and continuous brightness based on the fracture boundaries in the dynamic deformation path map, performs contour reconstruction on the fractured area, and generates a continuous segment map of the tooth contour; The color gamut dynamic control module extracts the color channel offset between the reconstructed area and the surrounding area based on the contour continuous segment map, adjusts the color ratio and smoothes the mutation boundary, and generates an oral care monitoring imaging correction image.

2. The oral care monitoring imaging system according to claim 1, wherein: The oral imaging partition annotation map includes a tooth area annotation layer, a gum area annotation layer, a soft tissue area annotation layer, a boundary type annotation, a regional texture type label, and a partition number index; the tooth contour contrast distribution map includes a contour boundary line group, a boundary intensity distribution, a brightness abnormality mark, an edge direction vector, and a tooth identification area mask; the tooth structure dynamic deformation path map includes a deformation trajectory line, a time axis mapping, a deformation trend layer, and a dynamic change area label; the tooth contour continuous segment map includes a contour reconstruction segment, a continuity repair mask, a texture fusion area, and a segment boundary line; the oral care monitoring imaging correction image includes a color correction layer, a brightness balance layer, and a reconstruction fusion area.

3. The oral care monitoring imaging system according to claim 1, wherein The image structure decomposition module comprises: The image frame extraction submodule obtains the image frame sequence in oral care monitoring, extracts the grayscale jump value of the tooth surface area, the brightness mean change range of the gum area, and the texture distribution direction of the oral soft tissue area in the image, samples and collects the regional image features, and establishes a regional image feature set; The regional texture classification submodule determines the texture type to which the image feature belongs based on the grayscale jump value, the brightness mean value variation range and the texture distribution direction in the regional image feature set, and generates an image texture type classification result; The boundary structure calibration submodule detects the direction of boundary grayscale change based on the edge position of the region in the image texture type division result, extracts the adjacent pixel grayscale difference path, records the edge continuous pixel coordinates and numbers them according to the region type, and generates an oral imaging partition annotation map.

4. The oral care monitoring imaging system according to claim 1, wherein The local contour recognition module comprises: The brightness feature extraction submodule extracts the image brightness value distribution and the pixel density in the color channel on both sides of the boundary line based on the tooth boundary area delineated in the oral imaging partition annotation map, integrates the pixel rows and columns according to the boundary direction, calculates the brightness difference value of the boundary, and generates boundary brightness difference data; The edge anomaly positioning submodule extracts continuous boundary pixel columns based on the boundary segments where the brightness span changes dramatically in the boundary brightness difference values, and compares them with adjacent pixel segments to determine the consistency between the jump direction and the main direction of the boundary, and selects edge segments with stable continuous jump directions as abnormal structure areas to generate edge brightness jump position sets; The structural area delineation submodule calls the jump area where the edge brightness jump position is concentratedly marked, and delineates the closed boundary area with continuous edge jump direction and stable color distribution according to the boundary extension direction and contrast value change trend, and divides the area type according to the pixel density level, summarizes the contour structure boundary, and generates a tooth contour contrast distribution map.

5. The oral care monitoring imaging system according to claim 1, wherein The calculation formula of the boundary brightness difference value is specifically: Among them, L d represents the boundary brightness difference value, represents the brightness value of the i-th pixel inside the boundary, represents the brightness value of the i-th pixel outside the boundary, n represents the total number of pixels considered on both sides of the boundary line, i represents the index of the pixel, w i represents the brightness gradient weight value of the i-th pixel on the boundary line in the boundary normal direction, and the letter i associated with the summation symbol ∑ represents the summation over all considered pixels.

6. The oral care monitoring imaging system according to claim 1, wherein The image change detection module comprises: The grayscale trend extraction submodule extracts the grayscale value distribution of each frame region in the continuous image frame sequence based on the region indicated by the tooth contour contrast distribution map, calculates the grayscale average value between adjacent frames, identifies the continuous grayscale change trajectory segments, and establishes the regional grayscale change trend sequence; The texture fluctuation monitoring submodule detects the texture direction vector distribution of the corresponding area in the image frame based on the frame sequence area covered by the regional grayscale change trend sequence, extracts the texture set with prominent direction change rate, and screens the pixel blocks with similar vector change trends to obtain texture direction fluctuation trajectory data; The boundary path arrangement submodule calls the coordinate sequence in the texture direction fluctuation trajectory data, matches the boundary pixel change position in the frame sequence, tracks the boundary offset direction and connection order in continuous frames, integrates the continuous boundary position coordinates in chronological order and numbers the path segments, and generates a dynamic deformation path map of the tooth structure.

7. The oral care monitoring imaging system according to claim 1, wherein The grayscale average value calculation formula between adjacent frames is specifically: Among them, ΔG t is the average grayscale value between adjacent frames, N represents the number of sampled frames for the change in the average grayscale value, G t is the grayscale value of the image frame area at time point t, x i,j is the coordinate of the j-th pixel in the i-th frame, M is the total number of pixels in the image, G t+i is the grayscale value of the (t + i)-th frame.

8. The oral care monitoring imaging system according to claim 1, wherein The abnormal area filling module includes: The boundary break extraction submodule extracts the contour pixel point sequence based on the boundary break area in the dynamic deformation path map of the tooth structure, detects the grayscale interval and extension direction, screens the pixel break points and records the positions and numbers, and generates a boundary break positioning data set; The image block screening and reconstruction submodule calls the coordinate information of the boundary interruption positioning data set, obtains the texture direction and brightness continuity of adjacent image blocks, calculates the direction consistency rate and brightness difference, screens image blocks to construct contour filling structures, and generates an image filling splicing fragment set; The contour region embedding submodule fills in the structural position and direction identifiers in the spliced fragment set based on the image, writes the image content into the fracture coordinate area, compares the brightness gradient of the surrounding pixels and corrects the grayscale difference, unifies the edge pixel density, and generates a continuous fragment image of the tooth contour.

9. The oral care monitoring imaging system according to claim 1, wherein The color gamut dynamic control module includes: The color shift extraction submodule extracts the red, green and blue channel values based on the color difference between the reconstructed segment and the surrounding area in the tooth contour continuous segment image, calculates the mean difference of the channels in adjacent areas and filters out the pixel points that exceed the shift detection threshold, collects and marks the channel difference distribution, and generates the color channel shift range; The channel ratio reset sub-module calls the pixel channel difference in the color channel offset range, calculates the change coefficient of the brightness proportion of the channel in the region, adjusts the channel value according to the channel ratio reference value, outputs the channel proportion structure diagram, and obtains the color channel regulation ratio; The boundary brightness leveling sub-module, based on the channel value in the color channel regulation ratio, compares the pixel brightness of the reconstructed region boundary and the adjacent region, judges the situation where the threshold boundary brightness difference exceeds the range, performs brightness smoothing processing on the offset boundary and synthesizes the image, and generates an oral care monitoring imaging correction image.

10. An oral care monitoring imaging method, which is used to implement the urban utility tunnel intelligent management system according to any one of claims 1-9, and is characterized in that, It includes the following steps: S1: Obtain the oral care monitoring image frame sequence, collect the front view of the tooth surface, the image of the lingual region and the oral endoscope image, extract the amplitude of the tooth gray level jump, judge the gingival brightness balance degree, judge the soft tissue texture direction, screen the regions that meet the characteristics, extract the corresponding boundary coordinates, and generate an oral imaging partition annotation map; S2: Based on the tooth boundary in the oral imaging partition annotation map, extract the brightness and color channel density values on both sides of the boundary, screen the mutation segments, judge the brightness extension direction and perform pixel aggregation, and generate a tooth contour comparison distribution map; S3: Based on the structural region in the tooth contour comparison distribution map, extract the brightness performance of consecutive frames, judge the inter-frame change trend and the coherence of the texture direction, screen the continuously changing segments and connect the boundary trajectories, and generate a tooth structure dynamic deformation path map; S4: Based on the breakpoint position in the tooth structure dynamic deformation path map, capture the image blocks with consistent texture direction and stable brightness change, judge whether the edge trend is continuous, and insert them into the contour line to generate a tooth contour continuous segment map; S5: Based on the inserted region in the tooth contour continuous segment map, extract the color channel deviation value, judge whether it exceeds the limit, perform ratio adjustment on the abnormal channel, reorganize the image and smooth the jumping brightness, and generate an oral care monitoring imaging correction image.

Images