WO2020038312A1 - 多通道舌体边缘检测装置、方法及存储介质 - Google Patents

多通道舌体边缘检测装置、方法及存储介质 Download PDF

Info

Publication number
WO2020038312A1
WO2020038312A1 PCT/CN2019/101295 CN2019101295W WO2020038312A1 WO 2020038312 A1 WO2020038312 A1 WO 2020038312A1 CN 2019101295 W CN2019101295 W CN 2019101295W WO 2020038312 A1 WO2020038312 A1 WO 2020038312A1
Authority
WO
WIPO (PCT)
Prior art keywords
channel
tongue
image
map
points
Prior art date
Application number
PCT/CN2019/101295
Other languages
English (en)
French (fr)
Inventor
张贯京
葛新科
谭敦
王海荣
高伟明
Original Assignee
深圳市前海安测信息技术有限公司
深圳市易特科信息技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市前海安测信息技术有限公司, 深圳市易特科信息技术有限公司 filed Critical 深圳市前海安测信息技术有限公司
Publication of WO2020038312A1 publication Critical patent/WO2020038312A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20116—Active contour; Active surface; Snakes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing

Definitions

  • the invention relates to the technical field of traditional Chinese medicine tongue image processing, in particular to a multi-channel tongue body edge detection device, method and storage medium.
  • tongue diagnosis is a very important part because it can reflect many of the nature of the disease. Tongue diagnosis has a wide range of applications and has penetrated many aspects of medical and health care. Since the 1980s, researchers in traditional Chinese medicine in various disciplines have devoted themselves to the research of tongue detection methods. With the development of information technology, computer vision theory and image recognition technology are used to carry out computer recognition of traditional tongue image information. Establishing a computerized object recognition method for tongue image has become a hot spot in current research.
  • the tongue body portion needs to be automatically segmented from the original image, and then further detection and identification of the tongue image color and texture information are required.
  • how to effectively and accurately segment the tongue automatically from the original image has a crucial impact on the effective detection and recognition of subsequent tongue image color and texture. Therefore, it is necessary to provide an effective tongue segmentation algorithm to perform tongue segmentation on the tongue image to improve the accuracy of tongue segmentation.
  • the main objective of the present invention is to provide a multi-channel tongue edge detection device, method and machine storage medium, which are aimed at solving the technical problem of low accuracy of tongue segmentation in the prior art.
  • the present invention provides a multi-channel tongue edge detection device including a processor suitable for implementing various computer program instructions and a memory suitable for storing a plurality of computer program instructions.
  • the computer program instructions are processed by The processor loads and executes the following steps: obtaining the tongue RGB image that needs tongue segmentation; using the red area enhancement algorithm to obtain the R-channel grayscale enhancement map of the RGB image, and converting the RGB image into an HSI image and an HSV image, respectively, from the HSI image Extract the H-channel grayscale image from the HSV image, and extract the V-channel grayscale image from the HSV image; use the Otsu binary method to process the R-channel grayscale enhancement image, the H-channel grayscale image, and the V-channel grayscale image separately.
  • the cost function traverses all the potential contour points to obtain the optimal initial contour points, and fills all the optimal initial contour points to obtain the initial contour binary map; in the initial round, In the binary map, the white area of the V-channel binary map and the black areas on the top, left, and right of the white area are removed to obtain the final contour binary map; the final edge point information is obtained based on the final contour binary map, and GVF-snake is used.
  • the segmentation algorithm processes the final edge point information to obtain the final tongue segmentation map.
  • R, G, and B respectively represent the red channel gray value, the green channel gray value, and the blue channel gray value of the image
  • I is the gray value of the R channel gray enhancement map.
  • N is the number of all edge points
  • E is the cost function value of the current traversal point
  • dis () is the Euclidean distance between two coordinate points
  • pt is the initial point. Is the edge point of the current traversal, center is the center point of all edge points
  • abs () is the absolute difference formula, All edge points that are symmetrical to the current traversal point in the abscissa direction with the current traversal point.
  • the step of acquiring a tongue RGB image requiring tongue segmentation includes: acquiring a clear tongue RGB image from a patient's mouth through an image acquisition device of the multi-channel tongue edge detection device; or from the memory Obtain the tongue RGB image that needs tongue segmentation in.
  • the present invention also provides a multi-channel tongue edge detection method, which is applied to a multi-channel tongue edge detection device.
  • the multi-channel tongue edge detection device includes an image acquisition device, a memory, and an output unit.
  • the method includes The following steps: Obtain the tongue RGB image that needs tongue segmentation; use the red area enhancement algorithm to obtain the R channel gray enhancement map of the RGB image, convert the RGB image into HSI image and HSV image, and extract the H channel from the HSI image Grayscale image, and extract the V-channel grayscale image from the HSV image; use the Otsu binarization method to process the R-channel grayscale enhanced image, the H-channel grayscale image, and the V-channel grayscale image to the R-channel enhanced two, respectively.
  • Value map, H-channel binary map, and V-channel binary map extract the edge points of the R-channel enhanced binary map and the H-channel binary map, respectively, to obtain all potential initial contour points; based on the set cost function for all Traverse the potential contour points of L to get the optimal initial contour points, and fill all the optimal initial contour points to get the initial contour binary map; in the initial contour binary map, the V channel is The final contour binary map is obtained by removing the white areas of the value map and the black areas on the top, left, and right of the white area; based on the final contour binary map, the final edge point information is obtained, and the final edge point information is performed using the GVF-snake segmentation algorithm. Process to get the final tongue segmentation map.
  • R, G, and B respectively represent the red channel gray value, the green channel gray value, and the blue channel gray value of the image
  • I is the gray value of the R channel gray enhancement map.
  • N is the number of all edge points
  • E is the cost function value of the current traversal point
  • dis () is the Euclidean distance between two coordinate points
  • pt is the initial point. Is the edge point of the current traversal, center is the center point of all edge points
  • abs () is the absolute difference formula, All edge points that are symmetrical to the current traversal point in the abscissa direction with the current traversal point.
  • the step of acquiring a tongue RGB image requiring tongue segmentation includes: acquiring a clear tongue RGB image from a patient's mouth through an image acquisition device of the multi-channel tongue edge detection device; or from the memory Obtain the tongue RGB image that needs tongue segmentation in.
  • the multi-channel tongue edge detection method further includes the following steps: segmenting the tongue image through a display screen of the output unit, printing the tongue image through a printer of the output unit, or converting the tongue image Send to the doctor's terminal via the communication network.
  • the present invention is a computer-readable storage medium that stores a plurality of computer program instructions, the computer program instructions being loaded by a processor of a computer device and executing the multi-channel tongue edge detection method.
  • the multi-channel tongue edge detection method proposed by the present invention can improve the accuracy of the initial contour, thereby Improve the accuracy of the overall automatic tongue segmentation.
  • a valid initial contour must first be set to achieve the desired segmentation effect.
  • the multi-channel tongue edge detection method of the present invention can effectively extract the initial contour of the tongue image, thereby improving tongue image segmentation.
  • Accuracy In order to obtain better tongue edge information, the present invention obtains gray images of three channels, and then uses the Otsu rate binary algorithm (OTSU) to convert the gray image into a binary image and extracts its edge points respectively. The cost function is set to find the optimal edge point contour information. Based on the edge point contour information, GVF-snake is used to further fit the final tongue segmentation image to improve the accuracy of tongue segmentation.
  • Otsu rate binary algorithm Otsu rate binary algorithm
  • FIG. 1 is a schematic block diagram of a preferred embodiment of a multi-channel tongue edge detection device according to the present invention
  • FIG. 2 is a flowchart of a preferred embodiment of a multi-channel tongue edge detection method according to the present invention
  • FIG. 3 is a schematic diagram of extracting an R channel gray enhancement map from a tongue image
  • FIG. 4 is a schematic diagram of extracting an H-channel grayscale image from a tongue image
  • FIG. 5 is a schematic diagram of extracting a V-channel grayscale image from a tongue image
  • FIG. 6 is a schematic diagram of a process of extracting tongue contour points from a tongue image and obtaining a tongue segmentation image.
  • FIG. 1 is a schematic block diagram of a preferred embodiment of a multi-channel tongue edge detection device of the present invention.
  • the multi-channel tongue edge detection device 1 is equipped with a multi-channel tongue edge detection system 10, and the multi-channel tongue edge detection device 1 may be a multi-channel tongue edge detection system 10 A personal computer, a workstation computer, a four-diagnostic instrument for traditional Chinese medicine, and other computer devices with data processing functions and image processing functions.
  • the multi-channel tongue edge detection device 1 includes, but is not limited to, a multi-channel tongue edge detection system 10, an image acquisition device 11, a memory 12 suitable for storing a plurality of computer program instructions, and executing each A computer program instruction processor 13 and an output unit 14.
  • the image acquisition device 11 is a high-definition camera device including at least a stepping motor and a lens, such as a high-definition camera, for capturing a tongue surface image including the tongue from the tongue of a patient.
  • the memory 12 may be a read-only memory ROM, a random access memory RAM, an electrically erasable memory EEPROM, a flash memory FLASH, a magnetic disk, or an optical disk.
  • the processor 13 is a central processing unit (CPU), a microcontroller (MCU), a data processing chip, or an information processing unit having a data processing function.
  • the output unit 14 may be a display screen for displaying a tongue image, or a printer for printing a tongue image.
  • the multi-channel tongue edge detection system 10 is composed of program modules composed of multiple computer program instructions, including, but not limited to, tongue image acquisition module 101, tongue image processing module 102, and tongue contour extraction.
  • Module 103, tongue contour optimization module 104, and tongue segmentation module 105 are included in the multi-channel tongue edge detection device 1 and can complete fixed functions, which are stored in the memory 12. Explain the specific functions of each module.
  • FIG. 2 is a flowchart of a preferred embodiment of a multi-channel tongue edge detection method according to the present invention.
  • various method steps of the multi-channel tongue edge detection method are implemented by a computer software program, and the computer software program is stored in a computer-readable storage medium (such as the memory 12) in the form of computer program instructions.
  • the computer-readable storage medium may include: a read-only memory, a random access memory, a magnetic disk, or an optical disk.
  • the computer program instructions can be loaded by a processor (for example, the processor 13) and execute the following steps S21 to S27.
  • Step S21 Obtain a tongue RGB image that requires tongue segmentation.
  • the tongue image acquisition module 101 captures a clear tongue RGB image from the patient's mouth through the image acquisition device 11.
  • the tongue RGB image may also be stored in the memory 12 in advance, and the tongue image acquisition module 101 may directly obtain the tongue RGB image from the memory 12 that needs tongue division.
  • Step S22 Use the red area enhancement algorithm to obtain the R channel gray enhancement map of the RGB image, convert the RGB image into HSI image and HSV image, extract the H channel gray image from the HSI image, and extract V from the HSV image.
  • Channel gray map Use the red area enhancement algorithm to obtain the R channel gray enhancement map of the RGB image, convert the RGB image into HSI image and HSV image, extract the H channel gray image from the HSI image, and extract V from the HSV image. Channel gray map.
  • the tongue image processing module 102 can use the red area enhancement algorithm to directly extract the R channel gray enhancement map from the tongue RGB image ( As shown in b in Figure 3), the red area enhancement algorithm is based on the principle that the blue channel reduces the contrast between the tongue and the skin color.
  • the specific implementation formula is as follows:
  • R, G, and B respectively represent the gray value of the red channel, the gray value of the green channel, and the gray value of the blue channel, and I is the gray value of the R channel. Enhance the gray value of the map.
  • the tongue image processing module 102 further converts the tongue RGB image (as shown in FIG. 4A) into an HSI image. Because the tongue region and the skin color region of the tongue RGB image have a certain difference in hue. Therefore, the histogram of the hue channel H in the converted HSI image has a bimodal characteristic, which can separate the tongue and the skin color region. Therefore, the grayscale image of the H channel can be extracted from the HSI image (as shown in b in FIG. 4). As shown).
  • the tongue image processing module 102 then converts the tongue RGB image (shown as a in FIG. 5) into an HSV image. Since the tongue and the lip area will have a part of the shadow when the lips are open, since V The component is the brightness of the color. The brightness determines the intensity of the chromaticity. It is a measure of the amount of colored light. The brightness of the middle part of the lips is obviously different from that of the tongue. In order to effectively distinguish the tongue from the lip area In the beginning, the tongue image processing module 102 converts the tongue RGB image into an HSV image, and extracts a V-channel grayscale image from the HSV image (as shown in FIG. 5B).
  • Step S23 using the Otsu binarization method to process the R-channel grayscale enhancement map, the H-channel grayscale map, and the V-channel grayscale map, respectively, to obtain a binary map of each channel; specifically, the tongue image processing module 102
  • the grayscale enhancement map of R channel, the grayscale map of H channel in HSI channel, and the grayscale map of V channel in HSV channel are obtained by using the Otsu rate binary method (OTSU) to obtain the binarization threshold to obtain different binary maps of each channel, V
  • Otsu rate binary method Otsu rate binary method
  • the binary map obtained by channel inversion is a binary map obtained by inverting the area outside the skin color region.
  • the binary result map is finally obtained, as shown in FIG. 3 c, the R channel enhanced binary map, as shown in FIG.
  • the Otsu rate binary method is a prior art image processing method, that is, selecting a grayscale image of 256 brightness levels through an appropriate threshold so that all pixels smaller than the threshold are set to a value , All pixels larger than the threshold are set to another value, and finally a binary image that is not black or white is obtained.
  • step S24 the edge points of the R-channel enhanced binary map and the H-channel binary map are extracted to obtain all potential initial contour points; the tongue contour extraction module 103 respectively performs R-channel enhanced binary map and H-channel binary map. Extract all edge points of the boundaries of the two binary images in order from left to top, bottom to left and right to bottom from top to bottom, and get all the edge points shown in Figure a in Figure 6 Binary map and use these edge points as potential initial contour points.
  • Step S25 traverse all the potential contour points based on the set cost function to obtain the optimal initial contour points (as shown in FIG. 6b), and fill all the optimal initial contour points to obtain the initial contour.
  • Binary graph shown as c in Figure 6
  • the tongue contour optimization module 104 finds the optimal edge point path through a set cost function, according to the left point from top to bottom, and the bottom point from left to left. All the edge points are traversed in the order from right to right, from bottom to top.
  • the left refers to the point that is smaller than the left of the abscissa of the center of the center (that is, the point whose abscissa is less than the abscissa of center), and the bottom refers to center.
  • the point below the ordinate of the center point that is, the point whose ordinate value is greater than the center ordinate
  • the right side is the point to the right of the center center point abscissa (that is, the point whose abscissa value is greater than the center ordinate)
  • the center center point coordinates are all edges
  • the coordinate average value of the point (such as Equation 2).
  • the uppermost point on the left is selected as the initial point.
  • the cost function is used to position the point from top to bottom according to the position of the left, from the left to the right, and from the bottom to the right.
  • the order of the replacement cost function value is calculated, and the point with the lowest cost function value is used as the next initial point. All the points are traversed in this order to obtain the final initial contour point.
  • the cost function formula is expressed as follows:
  • N is the number of all edge points
  • E is the cost function value of the current traversal point
  • dis () is the Euclidean distance between two coordinate points
  • pt is the initial point.
  • Is the edge point of the current traversal center is the center point of all edge points
  • abs () is the absolute difference formula
  • Step S26 Remove the white area of the V-channel binary image and the black areas on the top, left, and right of the white region in the initial contour binary image to obtain a final contour binary image.
  • the tongue contour optimization module 104 After the traversal, the optimal edge point information is obtained, and the initial contour binary map is obtained by filling. Combined with the brightness V channel in the HSV channel, the lip area and the shadow between the lips and the middle part of the tongue are removed, that is, the V channel is removed. The white area in the binary map and the black areas to the left, top, and right of the white area are used to obtain the final contour binary map (as shown in d in FIG. 6), so as to obtain the initial contour edge information.
  • step S27 the final edge point information is obtained based on the final contour binary map, and the final edge point information is processed by using the GVF-snake segmentation algorithm to obtain the final tongue segmentation map.
  • the tongue segmentation module 105 obtains the final edge point information based on the final contour binary map (shown as e in FIG. 6), and uses a gradient-based vector flow-snake model (GVF-Snake)
  • the segmentation algorithm processes the final edge point information to obtain the final tongue segmentation map (shown as f in Figure 6).
  • the GVF-snake segmentation algorithm is an optimized Snake model image segmentation algorithm in the prior art, which can perform fine segmentation processing on the contour of the tongue effectively to obtain the final tongue segmentation map.
  • the prior art image segmentation algorithm is not described in this embodiment.
  • the tongue segmentation module 105 further divides the tongue image through the display screen of the output unit 14, or prints the tongue image through a printer, or sends the tongue image to the doctor's terminal through the communication network for the doctor to pass
  • the patient's tongue image diagnoses the tongue's size, shape, color, texture, cracks, fetal quality, and presence or absence of tooth marks to help doctors perform a TCM tongue diagnosis to obtain the patient's health status.
  • the present invention is also a computer-readable storage medium.
  • the computer-readable storage medium stores a plurality of computer program instructions, and the computer program instructions are loaded by a processor of a computer device and execute the multi-channel tongue edge detection method of the present invention.
  • the program may be stored in a computer-readable storage medium.
  • the storage medium may include a read-only memory, a random access memory, Disk or CD, etc.
  • the multi-channel tongue edge detection method of the present invention obtains gray images of three channels, and then uses the Otsu rate binary algorithm (OTSU) to convert the gray images to binary values. After extracting the edge points of each image, a cost function is set to obtain the optimal edge point contour information. Based on the edge point contour information, GVF-snake is used to further fit the tongue contour to the final tongue segmentation map.
  • Otsu rate binary algorithm Otsu rate binary algorithm
  • the multi-channel tongue edge detection method proposed by the present invention can improve the accuracy of the initial contour, thereby improving the overall automatic tongue segmentation. Accuracy.
  • a valid initial contour must first be set to achieve the expected segmentation effect.
  • the multi-channel tongue edge detection method of the present invention can effectively extract the initial contour of the tongue image, thereby improving Accuracy of tongue image segmentation.
  • the multi-channel tongue edge detection method proposed by the present invention can improve the accuracy of the initial contour, thereby Improve the accuracy of the overall automatic tongue segmentation.
  • a valid initial contour must first be set to achieve the desired segmentation effect.
  • the multi-channel tongue edge detection method of the present invention can effectively extract the initial contour of the tongue image, thereby improving tongue image segmentation.
  • Accuracy In order to obtain better tongue edge information, the present invention obtains gray images of three channels, and then uses the Otsu rate binary algorithm (OTSU) to convert the gray image into a binary image and extracts its edge points respectively. The cost function is set to find the optimal edge point contour information. Based on the edge point contour information, GVF-snake is used to further fit the final tongue segmentation image to improve the accuracy of tongue segmentation.
  • Otsu rate binary algorithm Otsu rate binary algorithm

Abstract

本发明提供一种多通道舌体边缘检测装置、方法及存储介质，该方法包括步骤：利用红色区域增强算法求取舌体RGB图像的R通道灰度增强图，将RGB图像分别转化成HSI图像和HSV图像，从HSI图像中提取H通道灰度图，并从HSV图像中提取V通道灰度图；基于代价函数对潜在轮廓点进行遍历得到最优的初始轮廓点，并对所有最优的初始轮廓点进行填充得到初始轮廓二值图；在初始轮廓二值图中将V通道二值图的白色区域去除得到最终轮廓二值图；基于最终轮廓二值图得到最终边缘点信息，并利用GVF-snake分割算法对最终边缘点信息进行处理得到最终的舌体分割图。本发明能够有效地提取舌像的初始轮廓，从而提高舌像分割的准确率。

Description

多通道舌体边缘检测装置、方法及存储介质 技术领域
本发明涉及中医舌像处理的技术领域，尤其涉及一种多通道舌体边缘检测装置、方法及存储介质。
 背景技术
在中医诊断过程中，主要通过“望、闻、问、切”进行确诊，而在“望诊”中，舌诊是非常重要的一部分，因为它能反映病症的很多本质。舌诊的应用范围很广，渗透到了医疗和保健领域的很多方面。自本世纪八十年代开始，各学科中医相关研究人员就致力于舌诊检测方法的研究，随着信息技术的发展，运用计算机视觉理论与图像识别技术，对传统的舌像信息进行计算机识别，建立舌像的计算机客观化识别方法成为当前研究的热点。
对舌像信息进行计算机识别，首先需要将舌体部分自动从原图中分割出来，然后再进行进一步的舌像色彩、纹理等信息的检测识别。然而，如何有效精准的将舌体部分自动从原图中分割出来，对后续舌像色彩、纹理等信息的有效检测识别起着至关重要的影响。因此，有必要提供一种有效的舌体分割算法对舌体图像进行舌体分割，提高舌体分割的准确率。
 技术问题
本发明的主要目的在于提供一种基于多通道舌体边缘检测装置、方法及机存储介质，旨在解决现有技术对舌体分割的准确率不高的技术问题。
 技术解决方案
为实现上述目的，本发明提供一种多通道舌体边缘检测装置，包括适于实现各种计算机程序指令的处理器以及适于存储多条计算机程序指令的存储器，该所述计算机程序指令由处理器加载并执行如下步骤：获取需要舌体分割的舌体RGB图像；利用红色区域增强算法求取RGB图像的R通道灰度增强图，将RGB图像分别转化成HSI图像和HSV图像，从HSI图像中提取H通道灰度图，并从HSV图像中提取V通道灰度图；利用大津二值化法分别对R通道灰度增强图、H通道灰度图和V通道灰度图处理得分别到R通道增强二值图、H通道二值图和V通道二值图；分别对R通道增强二值图和H通道二值图的边缘点进行提取得到所有潜在的初始轮廓点；基于设定的代价函数对所有的潜在轮廓点进行遍历得到最优的初始轮廓点，并对所有最优的初始轮廓点进行填充得到初始轮廓二值图；在初始轮廓二值图中将V通道二值图的白色区域及白色区域的上边、左边和右边的黑色区域去除得到最终轮廓二值图；基于最终轮廓二值图得到最终边缘点信息，并利用GVF-snake分割算法对最终边缘点信息进行处理得到最终的舌体分割图。
进一步地，所述红色区域增强算法具体实现公式如下：
Figure dest_path_image001
其中，R、G、B分别代表为图像的红色通道灰度值、绿色通道灰度值、蓝色通道灰度值，I为R通道灰度增强图的灰度值。
进一步地，所述代价函数公式表示如下：
Figure 754266dest_path_image002
；
Figure dest_path_image003
；
其中，N为所有边缘点的个数，E为当前遍历点的代价函数值，dis()为两个坐标点的欧式距离，pt为初始点，
Figure 254517dest_path_image004
为当前遍历的边缘点，center为所有边缘点的中心点，abs()为绝对差公式，
Figure 675134dest_path_image005
为与当前遍历点的横坐标方向上与当前遍历点对称的所有边缘点。
进一步地，所述获取需要舌体分割的舌体RGB图像的步骤包括：通过所述多通道舌体边缘检测装置的图像采集设备从患者嘴部摄取清晰的舌体RGB图像；或者从所述存储器中获取需要舌体分割的舌体RGB图像。
另一方面，本发明还提供一种多通道舌体边缘检测方法，应用于多通道舌体边缘检测装置中，该多通道舌体边缘检测装置包括图像采集设备、存储器以及输出单元，该方法包括如下步骤：获取需要舌体分割的舌体RGB图像；利用红色区域增强算法求取RGB图像的R通道灰度增强图，将RGB图像分别转化成HSI图像和HSV图像，从HSI图像中提取H通道灰度图，并从HSV图像中提取V通道灰度图；利用大津二值化法分别对R通道灰度增强图、H通道灰度图和V通道灰度图处理得分别到R通道增强二值图、H通道二值图和V通道二值图；分别对R通道增强二值图和H通道二值图的边缘点进行提取得到所有潜在的初始轮廓点；基于设定的代价函数对所有的潜在轮廓点进行遍历得到最优的初始轮廓点，并对所有最优的初始轮廓点进行填充得到初始轮廓二值图；在初始轮廓二值图中将V通道二值图的白色区域及白色区域的上边、左边和右边的黑色区域去除得到最终轮廓二值图；基于最终轮廓二值图得到最终边缘点信息，并利用GVF-snake分割算法对最终边缘点信息进行处理得到最终的舌体分割图。
进一步地，所述红色区域增强算法具体实现公式如下：
Figure 227599dest_path_image006
其中，R、G、B分别代表为图像的红色通道灰度值、绿色通道灰度值、蓝色通道灰度值，I为R通道灰度增强图的灰度值。
进一步地，所述代价函数公式表示如下：
Figure dest_path_image007
；
；
其中，N为所有边缘点的个数，E为当前遍历点的代价函数值，dis()为两个坐标点的欧式距离，pt为初始点，
Figure 390913dest_path_image004
为当前遍历的边缘点，center为所有边缘点的中心点，abs()为绝对差公式，
Figure 615221dest_path_image008
为与当前遍历点的横坐标方向上与当前遍历点对称的所有边缘点。
进一步地，所述获取需要舌体分割的舌体RGB图像的步骤包括：通过所述多通道舌体边缘检测装置的图像采集设备从患者嘴部摄取清晰的舌体RGB图像；或者从所述存储器中获取需要舌体分割的舌体RGB图像。
进一步地，所述的多通道舌体边缘检测方法还包括如下步骤：将舌体图像通过输出单元的显示屏分割出的舌体图像，通过输出单元的打印机打印舌体图像，或者将舌体图像通过通信网络发送至医生终端。
再一方面，本发明一种计算机可读存储介质，该计算机存储介质存储多条计算机程序指令，所述计算机程序指令由计算机装置的处理器加载并执行所述多通道舌体边缘检测方法。
 有益效果
相较于现有技术，由于初始轮廓描述的偏差会降低主动轮廓模型（snake）对舌体区域分割的准确率，本发明提出的多通道舌体边缘检测方法能够提高初始轮廓的准确率，从而提高整体舌体自动分割的准确率。对于基于主动轮廓模型的舌像分割算法，首先需要设定有效的初始轮廓才能达到预期的分割效果，本发明多通道舌体边缘检测方法可以有效地提取舌像的初始轮廓，从而提高舌像分割的准确率。为了得到一个较好的舌体边缘信息，本发明通过获取三种通道的灰度图片，然后用大津率二值算法（OTSU）将灰度图片转换为二值图片后分别对其边缘点进行提取，设定代价函数求出最优的边缘点轮廓信息，基于边缘点轮廓信息采用GVF-snake对舌体轮廓进行更进一步的拟合最终的舌体分割图像，从而提高舌体分割的准确率。
 附图说明
图1是本发明多通道舌体边缘检测装置的优选实施例的方框示意图；
图2是本发明多通道舌体边缘检测方法优选实施例的流程图；
图3是从舌体图像提取R通道灰度增强图的示意图；
图4是从舌体图像提取H通道灰度图的示意图；
图5是从舌体图像提取V通道灰度图的示意图；
图6是从舌体图像提取舌体轮廓点并得到舌体分割图像的过程示意图。
本发明目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
 本发明的实施方式
为更进一步阐述本发明为达成预定发明目的所采取的技术手段及功效，以下结合附图及较佳实施例，对本发明的具体实施方式、结构、特征及其功效，详细说明如下。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
参照图1所示，图1是本发明多通道舌体边缘检测装置的优选实施例的方框示意图。在本实施例中，所述多通道舌体边缘检测装置1安装有多通道舌体边缘检测系统10，所述多通道舌体边缘检测装置1可以为安装有多通道舌体边缘检测系统10的个人计算机、工作站计算机、中医四诊仪等具有数据处理功能和图像处理功能的计算机装置。
在本实施例中，所述多通道舌体边缘检测装置1包括，但不仅限于，多通道舌体边缘检测系统10、图像采集设备11、适于存储多条计算机程序指令的存储器12、执行各种计算机程序指令的处理器13以及输出单元14。所述图像采集设备11为一种至少包括步进电机和镜头的高清摄像装置，例如高清摄像机，用于从患者的舌体摄取包含舌体的舌面图像。所述存储器12可以为一种只读存储器ROM，随机存储器RAM、电可擦写存储器EEPROM、快闪存储器FLASH、磁盘或光盘等。所述处理器13为一种中央处理器（CPU）、微控制器（MCU）、数据处理芯片、或者具有数据处理功能的信息处理单元。所述输出单元14可以为一种用于显示舌体图像的显示屏，也可以为一种用于打印舌体图像的打印机。
在本实施例中，所述多通道舌体边缘检测系统10由多条计算机程序指令组成的程序模块组成，包括但不局限于，舌像获取模块101、舌像处理模块102、舌体轮廓提取模块103、舌体轮廓优化模块104和舌体分割模块105。本发明所称的模块是指一种能够被多通道舌体边缘检测装置1的处理器13执行并且能够完成固定功能的一系列计算机程序指令段，其存储在存储器12中，以下结合图2具体说明每一个模块的具体功能。
参考图2所示，是本发明多通道舌体边缘检测方法优选实施例的流程图。在本实施例中，所述多通道舌体边缘检测方法的各种方法步骤通过计算机软件程序来实现，该计算机软件程序以计算机程序指令的形式存储于计算机可读存储介质（例如存储器12）中，计算机可读存储介质可以包括：只读存储器、随机存储器、磁盘或光盘等，所述计算机程序指令能够被处理器（例如处理器13）加载并执行如下步骤S21至步骤S27。
步骤S21，获取需要舌体分割的舌体RGB图像；在本实施例中，舌像获取模块101通过图像采集设备11从患者嘴部摄取清晰的舌体RGB图像。在其它实施例中，所述舌体RGB图像也可以预先存储在存储器12中，舌像获取模块101可以直接从存储器12中获取需要舌体分割的舌体RGB图像。
步骤S22，利用红色区域增强算法求取RGB图像的R通道灰度增强图，将RGB图像分别转化成HSI图像和HSV图像，从HSI图像中提取H通道灰度图，并从HSV图像中提取V通道灰度图。
由于舌体RGB图像（如图3中的a图所示）的边缘主要是红色，因此舌像处理模块102利用红色区域增强算法从舌体RGB图像中可以直接提取出R通道灰度增强图（如图3中的b图所示），该红色区域增强算法基于蓝色通道减少了舌体与肤色对比度的原则，其具体实现公式如下：
Figure dest_path_image009
                          式（1）
在（1）中，R、G、B分别代表为图像的红色（Red）通道灰度值、绿色（Green）通道灰度值、蓝色（Blue）通道灰度值，I为R通道灰度增强图的灰度值。
在本实施例中，舌像处理模块102再将舌体RGB图像（如图4中的a图所示）转换成HSI图像，由于舌体RGB图像的舌体区域与肤色区域的色调存在一定差异，所以转换得到的HSI图像中的色调通道H的直方图存在双峰特性，可以将舌体和肤色区域分开出来，因此可以从HSI图像中提取出H通道灰度图（如图4中的b图所示）。
在本实施例中，舌像处理模块102然后将舌体RGB图像（如图5中的a图所示）转换成HSV图像，由于在嘴唇张开时舌体与嘴唇区域会有一部分阴影部分，由于V分量是颜色的明亮度，明亮度决定了彩色度的光强，是彩色光在量方面的度量，嘴唇与舌体中间部分的明亮度明显存在很大差异，为了有效将舌体与嘴唇区域区分开来，舌像处理模块102将舌体RGB图像转换成HSV图像，并从HSV图像中提取出V通道灰度图（如图5中的b图所示）。
步骤S23，利用大津二值化法分别对R通道灰度增强图、H通道灰度图和V通道灰度图处理得到各通道二值图；具体地，舌像处理模块102将经过增强后的R通道灰度增强图、HSI通道中H通道灰度图、HSV通道中V通道灰度图分别用大津率二值法（OTSU）求取二值化阈值得到各个通道的不同二值图，V通道得到的二值图进行反转得到的二值图，去掉肤色区域之外的区域最终得到的二值结果图，如图3中c图所示的R通道增强二值图、如图4中c图所示的H通道二值图、图5中c图所示的V通道二值图。在本实施例中，所述大津率二值法是一种现有技术的图像处理方法，即将256个亮度等级的灰度图像通过适当的阈值选取，使得所有小于该阈值的像素置为一个值，所有大于该阈值的像素置为另一个值，最终得到一张非黑即白的二值图像。
步骤S24，分别对R通道增强二值图和H通道二值图的边缘点进行提取得到所有潜在的初始轮廓点；舌体轮廓提取模块103分别对R通道增强二值图和H通道二值图按照左边从上到下、下边从左到右、右边从下到上的顺序依次提取这两幅二值图的边界的所有边缘点，得到如图6中的a图所示的所有边缘点的二值图，并将这些边缘点作为潜在的初始轮廓点。
步骤S25，基于设定的代价函数对所有的潜在轮廓点进行遍历得到最优的初始轮廓点（如图6中的b图所示），并对所有最优的初始轮廓点进行填充得到初始轮廓二值图（如图6中的c图所示）。为了得到一个更优的初始轮廓点信息，在本实施例中，舌体轮廓优化模块104通过设定的代价函数来寻找最优的边缘点路径，按照左边点从上到下、下边点从左到右、右边点从下到上的顺序将所有边缘点都遍历一遍，这里左边是指小于center中心点横坐标左边的点（即横坐标点值小于center横坐标的点），下边是指center中心点纵坐标以下的点（即纵坐标值大于center纵坐标的点），右边为center中心点横坐标右边的点（即横坐标值大于center横坐标的点），center中心点坐标为所有边缘点的坐标平均值（如式2），选取左边最上面的点为初始点，利用代价函数根据该点的位置按照左边点从上到下、下边点从左到右、右边点从下到上的顺序求取代价函数值，将代价函数值最小的点作为下一个初始点，以此顺序遍历所有点遍得到最终的初始轮廓点。在本实施例中，所述代价函数公式表示如下：
Figure 739035dest_path_image010
                            式（2）
Figure 23386dest_path_image011
 
式（3）
其中，N为所有边缘点的个数，E为当前遍历点的代价函数值，dis()为两个坐标点的欧式距离，pt为初始点，
Figure 498229dest_path_image004
为当前遍历的边缘点，center为所有边缘点的中心点，abs()为绝对差公式，
Figure 526228dest_path_image008
为与当前遍历点的横坐标方向上与当前遍历点对称的所有边缘点，i为第i个遍历到的候选边缘点，j为与第i个遍历点相对中心点对称的第j个边缘点。
步骤S26，在初始轮廓二值图中将V通道二值图的白色区域及白色区域的上边、左边和右边的黑色区域去除得到最终轮廓二值图；在本实施例中，舌体轮廓优化模块104经过遍历后得到最优的边缘点信息，进行填充得到初始轮廓二值图，结合HSV通道中的明亮度V通道去除掉嘴唇区域及嘴唇与舌体中间部分的阴影部分，即去除掉V通道二值图中的白色区域以及白色区域左边、上边和右边的黑色区域，得到最终轮廓二值图（如图6中的d图所示），从而得到初始轮廓边缘信息。
步骤S27，基于最终轮廓二值图得到最终边缘点信息，并利用GVF-snake分割算法对最终边缘点信息进行处理得到最终的舌体分割图。在本实施例中，所述舌体分割模块105基于最终轮廓二值图得到最终边缘点信息（如图6中的e图所示），并利用基于梯度向量流-蛇模型(GVF-Snake)分割算法对最终边缘点信息进行处理得到最终的舌体分割图（如图6中的f图所示）。在本实施例中，所述GVF-snake分割算法为现有技术中的一种优化的Snake模型图像分割算法，能够针对有效对舌体轮廓进行精细分割处理得到最终的舌体分割图，由于是现有技术的图像分割算法，本实施例不作赘述。
此外，舌体分割模块105还将舌体图像通过输出单元14的显示屏分割出的舌体图像，或者通过打印机打印舌体图像，或者将舌体图像通过通信网络发送至医生终端，供医生通过患者的舌体图像诊断舌体的大小、形状、颜色、纹理、裂纹、胎质、以及有无齿痕等信息，从而辅助医生进行中医舌诊获得患者的健康状况。
本发明还一种计算机可读存储介质，该计算机可读存储介质存储多条计算机程序指令，所述计算机程序指令由计算机装置的处理器加载并执行本发明所述多通道舌体边缘检测方法。本领域技术人员可以理解，上述实施方式中各种方法的全部或部分步骤可以通过相关程序指令完成，该程序可以存储于计算机可读存储介质中，存储介质可以包括：只读存储器、随机存储器、磁盘或光盘等。
为了得到一个较好的舌体边缘信息，本发明所述多通道舌体边缘检测方法获取了三种通道的灰度图片，然后用大津率二值算法（OTSU）将灰度图片转换为二值图片后分别对其边缘点进行提取，设定代价函数求出最优的边缘点轮廓信息，基于边缘点轮廓信息采用GVF-snake对舌体轮廓进行更进一步的拟合最终的舌体分割图。
由于初始轮廓描述的偏差会降低主动轮廓模型（snake）对舌体区域分割的准确率，本发明提出的多通道舌体边缘检测方法能够提高初始轮廓的准确率，从而提高整体舌体自动分割的准确率。对于基于主动轮廓模型（snake）的舌像分割算法，首先需要设定有效的初始轮廓才能达到预期的分割效果，本发明多通道舌体边缘检测方法可以有效地提取舌像的初始轮廓，从而提高舌像分割的准确率。
以上仅为本发明的优选实施例，并非因此限制本发明的专利范围，凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本发明的专利保护范围内。
 工业实用性
相较于现有技术，由于初始轮廓描述的偏差会降低主动轮廓模型（snake）对舌体区域分割的准确率，本发明提出的多通道舌体边缘检测方法能够提高初始轮廓的准确率，从而提高整体舌体自动分割的准确率。对于基于主动轮廓模型的舌像分割算法，首先需要设定有效的初始轮廓才能达到预期的分割效果，本发明多通道舌体边缘检测方法可以有效地提取舌像的初始轮廓，从而提高舌像分割的准确率。为了得到一个较好的舌体边缘信息，本发明通过获取三种通道的灰度图片，然后用大津率二值算法（OTSU）将灰度图片转换为二值图片后分别对其边缘点进行提取，设定代价函数求出最优的边缘点轮廓信息，基于边缘点轮廓信息采用GVF-snake对舌体轮廓进行更进一步的拟合最终的舌体分割图像，从而提高舌体分割的准确率。

Claims (10)

  1. 一种多通道舌体边缘检测装置，包括适于实现各种计算机程序指令的处理器以及适于存储多条计算机程序指令的存储器，其特征在于，该所述计算机程序指令由处理器加载并执行如下步骤：
    获取需要舌体分割的舌体RGB图像；
    利用红色区域增强算法求取RGB图像的R通道灰度增强图，将RGB图像分别转化成HSI图像和HSV图像，从HSI图像中提取H通道灰度图，并从HSV图像中提取V通道灰度图；
    利用大津二值化法分别对R通道灰度增强图、H通道灰度图和V通道灰度图处理得分别到R通道增强二值图、H通道二值图和V通道二值图；
    分别对R通道增强二值图和H通道二值图的边缘点进行提取得到所有潜在的初始轮廓点；
    基于设定的代价函数对所有的潜在轮廓点进行遍历得到最优的初始轮廓点，并对所有最优的初始轮廓点进行填充得到初始轮廓二值图；
    在初始轮廓二值图中将V通道二值图的白色区域及白色区域的上边、左边和右边的黑色区域去除得到最终轮廓二值图；
    基于最终轮廓二值图得到最终边缘点信息，并利用GVF-snake分割算法对最终边缘点信息进行处理得到最终的舌体分割图。
  2. 如权利要求1所述的多通道舌体边缘检测装置，其特征在于，所述红色区域增强算法具体实现公式如下：
    其中，R、G、B分别代表为图像的红色通道灰度值、绿色通道灰度值、蓝色通道灰度值，I为R通道灰度增强图的灰度值。
  3. 如权利要求1所述的多通道舌体边缘检测装置，其特征在于，所述代价函数公式表示如下：
    Figure 913719dest_path_image002
    ；
    ；
    其中，N为所有边缘点的个数，E为当前遍历点的代价函数值，dis()为两个坐标点的欧式距离，pt为初始点，
    Figure 441969dest_path_image004
    为当前遍历的边缘点，center为所有边缘点的中心点，abs()为绝对差公式，
    Figure 805955dest_path_image005
    为与当前遍历点的横坐标方向上与当前遍历点对称的所有边缘点。
  4. 如权利要求1所述的多通道舌体边缘检测装置，其特征在于，所述获取需要舌体分割的舌体RGB图像的步骤包括：
    通过所述多通道舌体边缘检测装置的图像采集设备从患者嘴部摄取清晰的舌体RGB图像；或者
    从所述存储器中获取需要舌体分割的舌体RGB图像。
  5. 一种多通道舌体边缘检测方法，应用于多通道舌体边缘检测装置中，该多通道舌体边缘检测装置包括图像采集设备、存储器以及输出单元，其特征在于，该方法包括如下步骤：
    获取需要舌体分割的舌体RGB图像；
    利用红色区域增强算法求取RGB图像的R通道灰度增强图，将RGB图像分别转化成HSI图像和HSV图像，从HSI图像中提取H通道灰度图，并从HSV图像中提取V通道灰度图；
    利用大津二值化法分别对R通道灰度增强图、H通道灰度图和V通道灰度图处理得分别到R通道增强二值图、H通道二值图和V通道二值图；
    分别对R通道增强二值图和H通道二值图的边缘点进行提取得到所有潜在的初始轮廓点；
    基于设定的代价函数对所有的潜在轮廓点进行遍历得到最优的初始轮廓点，并对所有最优的初始轮廓点进行填充得到初始轮廓二值图；
    在初始轮廓二值图中将V通道二值图的白色区域及白色区域的上边、左边和右边的黑色区域去除得到最终轮廓二值图；
    基于最终轮廓二值图得到最终边缘点信息，并利用GVF-snake分割算法对最终边缘点信息进行处理得到最终的舌体分割图。
  6. 如权利要求5所述的多通道舌体边缘检测方法，其特征在于，所述红色区域增强算法具体实现公式如下：
    Figure 166529dest_path_image006
    其中，R、G、B分别代表为图像的红色通道灰度值、绿色通道灰度值、蓝色通道灰度值，I为R通道灰度增强图的灰度值。
  7. 如权利要求5所述的多通道舌体边缘检测方法，其特征在于，所述代价函数公式表示如下：
    ；
    Figure 985766dest_path_image008
    ；
    其中，N为所有边缘点的个数，E为当前遍历点的代价函数值，dis()为两个坐标点的欧式距离，pt为初始点， 为当前遍历的边缘点，center为所有边缘点的中心点，abs()为绝对差公式，
    Figure 735733dest_path_image010
    为与当前遍历点的横坐标方向上与当前遍历点对称的所有边缘点。
  8. 如权利要求5所述的多通道舌体边缘检测方法，其特征在于，所述获取需要舌体分割的舌体RGB图像的步骤包括：
    通过所述图像采集设备从患者嘴部摄取清晰的舌体RGB图像；或者
    从所述存储器中获取需要舌体分割的舌体RGB图像。
  9. 如权利要求5所述的多通道舌体边缘检测方法，其特征在于，该方法还包括如下步骤：
    将舌体图像通过输出单元的显示屏分割出的舌体图像，或者通过输出单元的打印机打印舌体图像，或者将舌体图像通过通信网络发送至医生终端。
  10. 一种计算机可读存储介质，该计算机可读存储介质存储多条计算机程序指令，其特征在于，所述计算机程序指令由计算机装置的处理器加载并执行如权利要求5至9任一项所述多通道舌体边缘检测方法。
PCT/CN2019/101295 2018-08-20 2019-08-19 多通道舌体边缘检测装置、方法及存储介质 WO2020038312A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810945525.0A CN110866932A (zh) 2018-08-20 2018-08-20 多通道舌体边缘检测装置、方法及存储介质
CN201810945525.0 2018-08-20

Publications (1)

Publication Number Publication Date
WO2020038312A1 true WO2020038312A1 (zh) 2020-02-27

Family

ID=69592242

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/101295 WO2020038312A1 (zh) 2018-08-20 2019-08-19 多通道舌体边缘检测装置、方法及存储介质

Country Status (2)

Country Link
CN (1) CN110866932A (zh)
WO (1) WO2020038312A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117132667A (zh) * 2023-10-26 2023-11-28 湖南半岛医疗科技有限公司 一种基于环境温度反馈的热图像处理方法及相关装置

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112017174A (zh) * 2020-09-03 2020-12-01 湖南省华芯医疗器械有限公司 图像处理方法、装置、电子设备及存储介质
CN112489053B (zh) * 2020-11-26 2021-07-13 深圳市艾合芯科智慧医疗有限公司 一种舌像分割方法、装置及存储介质
CN113052866B (zh) * 2021-03-29 2022-11-18 天津大学 基于局部二值拟合模型的超声图像舌轮廓提取方法
CN113781508B (zh) * 2021-09-09 2022-12-09 中国航发南方工业有限公司 叶型孔图像的轮廓型线提取与分割方法及系统、设备、存储介质
CN116777930B (zh) * 2023-05-24 2024-01-09 深圳汇医必达医疗科技有限公司 应用于舌象提取的图像分割方法、装置、设备及介质
CN116740054B (zh) * 2023-08-08 2023-10-20 天筛(聊城)生物科技有限公司 一种基于图像处理的舌象齿痕检测方法
CN117422720B (zh) * 2023-12-19 2024-03-05 陕西秒康医疗科技有限公司 一种中医治疗舌诊图像智能分割方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060247502A1 (en) * 2005-04-28 2006-11-02 Eastman Kodak Company Method for diagnosing disease from tongue image
CN1920854A (zh) * 2006-09-15 2007-02-28 清华大学 中医舌诊用的舌头彩色数码照片的舌体分割提取方法
CN104658003A (zh) * 2015-03-16 2015-05-27 北京理工大学 舌体图像的分割方法和装置
CN106023151A (zh) * 2016-05-09 2016-10-12 厦门大学 一种开放环境下中医舌象目标检测方法
CN107392904A (zh) * 2017-07-28 2017-11-24 陆杰 一种基于数学形态学的医学图像的分割算法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060247502A1 (en) * 2005-04-28 2006-11-02 Eastman Kodak Company Method for diagnosing disease from tongue image
CN1920854A (zh) * 2006-09-15 2007-02-28 清华大学 中医舌诊用的舌头彩色数码照片的舌体分割提取方法
CN104658003A (zh) * 2015-03-16 2015-05-27 北京理工大学 舌体图像的分割方法和装置
CN106023151A (zh) * 2016-05-09 2016-10-12 厦门大学 一种开放环境下中医舌象目标检测方法
CN107392904A (zh) * 2017-07-28 2017-11-24 陆杰 一种基于数学形态学的医学图像的分割算法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"International Conference on Financial Cryptography and Data Security", vol. 10451 Chap.32, 26 August 2017, article LI, JIANG: "A Tongue Image Segmentation Method Based on Enhanced HSV Convolutional Neural Network", pages: 252 - 260, XP047435453, DOI: 10.1007/978-3-319-66805-5_32 *
CUI ZHENCHAO: "Tongue body segmentation and texture classification in computerized tongue diagnosis (Chapter 3)", CHINESE DOCTORAL DISSERTATIONS FULL-TEXT DATABASE, 1 December 2015 (2015-12-01), pages 1 - 28, XP055688190 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117132667A (zh) * 2023-10-26 2023-11-28 湖南半岛医疗科技有限公司 一种基于环境温度反馈的热图像处理方法及相关装置
CN117132667B (zh) * 2023-10-26 2024-02-06 湖南半岛医疗科技有限公司 一种基于环境温度反馈的热图像处理方法及相关装置

Also Published As

Publication number Publication date
CN110866932A (zh) 2020-03-06

Similar Documents

Publication Publication Date Title
WO2020038312A1 (zh) 多通道舌体边缘检测装置、方法及存储介质
CN103971126B (zh) 一种交通标志识别方法和装置
CN103400150B (zh) 一种基于移动平台进行道路边缘识别的方法及装置
WO2019062092A1 (zh) 基于超像素-多色彩空间的人体轮廓提取方法
CN103914699A (zh) 一种基于色彩空间的自动唇彩的图像增强的方法
CN109241973B (zh) 一种纹理背景下的字符全自动软分割方法
CN108615239B (zh) 基于阈值技术和灰度投影的舌图像分割方法
CN110084135A (zh) 人脸识别方法、装置、计算机设备及存储介质
JPH0877334A (ja) 顔画像の特徴点自動抽出方法
CN110807775A (zh) 基于人工智能的中医舌像分割装置、方法及存储介质
CN107545550B (zh) 细胞图像偏色校正方法
CN111223110B (zh) 一种显微图像增强方法、装置及计算机设备
CN111461036B (zh) 一种利用背景建模增强数据的实时行人检测方法
CN110969631B (zh) 一种精细化照片染发方法及系统
CN104021566A (zh) 基于GrabCut算法的舌诊图像自动分割方法
CN112330561B (zh) 基于交互式前景提取与信息熵分水岭的医学图像分割方法
CN104766316A (zh) 中医望诊中新型嘴唇分割算法
CN111161281A (zh) 一种人脸区域识别方法、装置及存储介质
WO2020108437A1 (zh) 舌下静脉特征提取装置及方法
CN108711160A (zh) 一种基于hsi增强性模型的目标分割方法
CN110648336B (zh) 一种舌质和舌苔的分割方法及装置
CN107239761B (zh) 基于骨架角点检测的果树拉枝效果评价方法
CN104766068A (zh) 一种多规则融合的随机游走舌像提取方法
CN109934152B (zh) 一种针对手语图像的改进小弯臂图像分割方法
CN112598674A (zh) 用于车辆的图像处理方法、装置及车辆

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19852309

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19852309

Country of ref document: EP

Kind code of ref document: A1