WO2021212693A1 - 一种融合Gabor小波的多尺度局部水平集超声图像分割方法 - Google Patents

一种融合Gabor小波的多尺度局部水平集超声图像分割方法 Download PDF

Info

Publication number
WO2021212693A1
WO2021212693A1 PCT/CN2020/107283 CN2020107283W WO2021212693A1 WO 2021212693 A1 WO2021212693 A1 WO 2021212693A1 CN 2020107283 W CN2020107283 W CN 2020107283W WO 2021212693 A1 WO2021212693 A1 WO 2021212693A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
level set
equation
energy
segmentation method
Prior art date
Application number
PCT/CN2020/107283
Other languages
English (en)
French (fr)
Inventor
周智峰
邹慧玲
Original Assignee
北京华科创智健康科技股份有限公司
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 北京华科创智健康科技股份有限公司 filed Critical 北京华科创智健康科技股份有限公司
Priority to CN202080100122.1A priority Critical patent/CN115843373A/zh
Publication of WO2021212693A1 publication Critical patent/WO2021212693A1/zh
Priority to US18/048,527 priority patent/US20230070200A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/168—Segmentation; Edge detection involving transform domain methods
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/08—Detecting organic movements or changes, e.g. tumours, cysts, swellings
    • A61B8/0833—Detecting organic movements or changes, e.g. tumours, cysts, swellings involving detecting or locating foreign bodies or organic structures
    • A61B8/085—Detecting organic movements or changes, e.g. tumours, cysts, swellings involving detecting or locating foreign bodies or organic structures for locating body or organic structures, e.g. tumours, calculi, blood vessels, nodules
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/5207—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of raw data to produce diagnostic data, e.g. for generating an image
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/54—Control of the diagnostic device
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • 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
    • G06T7/174—Segmentation; Edge detection involving the use of two or more images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/12—Diagnosis using ultrasonic, sonic or infrasonic waves in body cavities or body tracts, e.g. by using catheters
    • 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
    • G06T2207/10068—Endoscopic 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
    • G06T2207/10132—Ultrasound 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
    • G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
    • 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
    • G06T2207/20048—Transform domain processing
    • G06T2207/20064—Wavelet transform [DWT]
    • 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
    • G06T2207/20161—Level set
    • 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
    • G06T2207/20172—Image enhancement details
    • G06T2207/20192—Edge enhancement; Edge preservation
    • 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
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/03—Recognition of patterns in medical or anatomical images

Definitions

  • the invention relates to the field of ultrasound image segmentation, and in particular to a multi-scale local level set ultrasound image segmentation method fused with Gabor wavelet.
  • Ultrasound endoscopy uses ultrasound to detect and image the internal organs of the human body. It has no damage to the human body, low detection cost, and high diagnostic accuracy. Therefore, it has become one of the important methods for early gastrointestinal tumor detection and treatment.
  • the location, size and shape of tumors in the digestive tract are important parameters that assist doctors in judging. Therefore, it is of great significance to extract tumor edges from ultrasound images. Due to the detection characteristics of the ultrasound system itself, ultrasound images have the shortcomings of uneven gray distribution, high noise, and poor edge continuity. Therefore, some classic segmentation methods, such as Canny operator, threshold method, etc., are difficult to obtain complete and accurate edge.
  • the proposal of the active contour model is a breakthrough in the field of image segmentation. Its basic idea is to use a continuous curve to fit the edge to be measured, and to perform image segmentation by defining an energy functional equation with the edge as a variable and finding the minimum value.
  • the level set method With the introduction of the level set method, the applicability of the active contour model has been developed.
  • the traditional level set algorithm still has certain shortcomings in actual image processing. For example, the famous Mumford-Shah model is not suitable for practical application due to the complicated calculation iteration process and the large amount of calculation; and the improved Chan-Vase model based on this model is aimed at It is also difficult to obtain accurate segmentation results for images with uneven gray levels.
  • the purpose of the present invention is to propose an ultrasound image segmentation algorithm to improve the accuracy of ultrasound image segmentation.
  • the first aspect of the present invention provides an image segmentation method, including the following steps:
  • Step S1 Obtain an image to be processed
  • Step S2 Use multi-directional Gabor wavelet to filter and decompose the original image to obtain multiple intermediate images
  • Step S3 fusing the multiple intermediate images by taking the maximum value to obtain an enhanced image
  • Step S4 For the intermediate image, construct a corresponding level set energy equation
  • Step S5 optimizing the parameters of the energy equation to make the energy function take the minimum value to obtain the accurate position of the edge
  • Step S6 Repeat the step S5 until the energy function reaches the minimum value, and the final edge is obtained.
  • the multi-directional Gabor wavelet change function in step S2 is
  • g q (x,y) g(x cos ⁇ +y sin ⁇ ,-x sin ⁇ +y cos ⁇ )
  • the intermediate image is shown in the following formula:
  • I (x, y) is the original image data
  • I q (x, y) is the intermediate image data
  • the fusion equation for fusing the plurality of intermediate images in the step S3 is as follows:
  • I'(x,y) is the fusion image data.
  • the energy equation includes: an energy functional term related to the image itself, a length regular term that keeps the edge smooth by limiting the length of the edge, and a distance regular term that avoids reinitialization of the level set equation.
  • the energy functional term is calculated by assuming that the gray-level uneven image is the result of weighting the gray-level bias term with the real image and adding noise, and constructing Gaussian templates with different variances;
  • the Gaussian template is as follows:
  • ⁇ ( ⁇ ,c,b) is the energy functional term related to the image itself
  • ⁇ , c, b are optimized, the energy function is minimized, the three parameters are optimized separately, and the values of the other two parameters are fixed during optimization, and the Euler-Lagrangian formula is used to optimize the level set Equation ⁇ , using partial differential equations to optimize c and b, the optimization equation for ⁇ is as follows:
  • a second aspect of the present invention provides an image processing device, including:
  • the storage unit is used to store the image to be processed
  • the processing unit is configured to obtain edge information in the image in the storage device according to the method described in one of the above technical solutions.
  • a third aspect of the present invention provides an ultrasonic imaging device, including:
  • An ultrasonic probe which is used to send ultrasonic waves to a subject to be inspected, receive ultrasonic waves reflected by the subject to be inspected, and generate echo signals corresponding to the reflected ultrasonic waves;
  • a generating unit configured to generate an ultrasound image related to the object to be inspected according to the echo signal
  • the processing unit is configured to obtain edge information in the ultrasound image according to the method described in one of the above technical solutions.
  • the fourth aspect of the present invention provides an ultrasonic endoscope, which includes an insertion part, a control part, and the ultrasonic imaging device according to the third aspect of the present invention.
  • the present invention provides a multi-scale local clustering level set ultrasound image segmentation method fused with Gabor wavelet, which uses the multi-directionality of Gabor wavelet to process the ultrasound image with uneven gray scale, and uses the maximum value fusion In this way, an intermediate image that enhances the difference between the region to be segmented and the background is obtained.
  • the present invention introduces the idea of multi-scale into the local clustering level set algorithm, sets Gaussian kernel functions with different variances, and performs level set iteration by means of mean fusion to obtain the final edge, thereby overcoming the weaker edge of the ultrasound image and the segmentation Disadvantages of inaccuracy,
  • Fig. 1 is a schematic flowchart of an image segmentation method according to the present invention.
  • Fig. 2 is a schematic diagram of the structure of an image processing device according to the present invention.
  • Fig. 3 is a schematic structural diagram of an ultrasonic imaging device according to the present invention.
  • FIG. 1 The implementation process of the image segmentation method of this embodiment is shown in Fig. 1:
  • Step 1 Read in the acquired ultrasound images.
  • the ultrasound image is a two-dimensional stomach section ultrasound image.
  • Step 2 Use a multi-directional Gabor filter to filter the image.
  • the Gabor function in the two-dimensional space is as follows:
  • ⁇ x and ⁇ y represent the broadening of the Gaussian function in the x direction and the y direction, respectively, and W represents the frequency bandwidth of the Gabor wavelet.
  • the multi-directional Gabor wavelet transform function can be expressed as:
  • g q (x,y) g(x cos ⁇ +y sin ⁇ ,-x sin ⁇ +y cos ⁇ )
  • Step 3 In order to maximize the details of the original image and enhance the difference between the foreground and the background to be segmented, the multiple intermediate images are merged by taking the maximum value to obtain an enhanced image.
  • Step 4 For the intermediate image, construct the corresponding level set energy equation.
  • the energy equation of the whole image consists of three parts, including the energy functional term related to the image itself, the length regular term that keeps the edge smooth by limiting the edge length, and the distance regular term to avoid reinitialization of the level set equation.
  • the energy functional term is calculated by assuming that the gray-level uneven image is the result of weighting the gray-level bias term with the real image and adding noise, and constructing Gaussian templates with different variances.
  • the Gaussian template is as follows:
  • Step 5 Optimize ⁇ , c, b, make the energy function take the minimum value, optimize the three parameters separately, fix the values of the other two parameters when optimizing, and use Euler-Lagrangian formula to optimize the level set Equation ⁇ , using partial differential equations to optimize c and b, the optimization equation for ⁇ is as follows:
  • Step 6 Repeat the calculation of the optimization equation in step 5 until the energy function reaches the minimum value, and the final edge is obtained.
  • this embodiment provides an image processing device, which includes a storage unit and a processing unit.
  • the storage unit is used to store the image to be processed
  • the processing unit is configured to obtain edge information in the image in the storage device according to the method described in any one of the above embodiment 1.
  • this embodiment provides an ultrasonic imaging device, which includes an ultrasonic probe, a generating part, and a processing part.
  • the ultrasonic probe is used to send ultrasonic waves to a subject, receive ultrasonic waves reflected by the subject, and generate echo signals corresponding to the reflected ultrasonic waves.
  • the generating unit is configured to generate an ultrasound image related to the object under inspection based on the echo signal.
  • the processing unit is configured to obtain edge information in the ultrasound image according to the method described in any one of the above embodiment 1.
  • the image segmentation methods related to the above embodiments use ultrasound images as processing objects.
  • the present invention is not limited to this.
  • the image segmentation method according to the various embodiments of the present invention can not only use ultrasound images as processing objects, but also process various grayscale images suitable for the method, such as CT images generated by X-ray computed tomography devices. , X-ray images generated by X-ray diagnostic equipment, MR images generated by magnetic resonance imaging equipment.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Surgery (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Pathology (AREA)
  • Radiology & Medical Imaging (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Multimedia (AREA)
  • Vascular Medicine (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)
  • Image Analysis (AREA)

Abstract

一种融合Gabor小波的多尺度局部水平集超声图像分割方法，该方法将超声图像的灰度不均性看作方向杂乱的纹理，利用Gabor小波的多方向性对图像进行处理，通过取最大值的方式对不同滤波方向的中间图像进行融合，得到减弱纹理效果、增强前景和背景之间差别的中间图像。针对超声图像边缘较弱的特点，用多尺度思想改进传统LIC方法，设定不同方差的高斯卷积核，再通过均值融合的方式得到最终边缘。

Description

一种融合Gabor小波的多尺度局部水平集超声图像分割方法 技术领域
本发明涉及超声图像分割领域，具体涉及一种融合Gabor小波的多尺度局部水平集超声图像分割方法。
背景技术
目前，全世界消化道内的肿瘤发病率较高，且具有较高的致死率，严重威胁人们的生命健康。超声内镜利用超声波对人体内部器官进行检测成像，对人体无损害、检测成本低、诊断准确度高，因此成为早期消化道肿瘤检测与治疗的重要手段之一。消化道内肿瘤的位置、大小和形状是辅助医生判断的重要参数，因此对超声图像进行肿瘤的边缘提取具有重要意义。由于超声系统本身的检测特性，超声图像具有灰度分布不均、噪声较高、边缘连续性较差的缺点，所以一些经典的分割方法，比如Canny算子、阈值法等难以得到完整且准确的边缘。
活动轮廓模型的提出是图像分割领域的一个突破，其基本思想是使用连续曲线来拟合待测边缘，通过定义以边缘为变量的能量泛函方程并求取最小值的方式来进行图像分割。随着水平集方法的引入，活动轮廓模型的适用性得到了发展。但是传统的水平集算法在进行实际图像处理时仍存在一定缺陷，例如著名的Mumford-Shah模型由于计算迭代过程复杂、计算量大，不适宜实际应用；而基于该模型改进的Chan-Vase模型针对灰度不均图像，也难以得到准确的分割结果。
针对灰度不均图像的分割，人们提出了基于局部区域的水平集方法。2007年，Li提出了局部二值化能量(LBF)模型，该模型将高斯核函数引入能量方程提取局部灰度信息，且加入正则项使得水平集迭代时无需进行重新初始化，减小了运算量，但是该算法对初始轮廓敏感，且对噪声图像的鲁棒性较差。2008年，Lankton提出了LRB方法，该方法同样利用局部区域信息对不均匀图像进行分割，但是算法计算效率较低。2011年，Li提出了局部强度聚类水平集方法，该方法针对灰度均匀变化的图像具有较好的分割效果， 但是对超声图像这种边缘弱、灰度变化不均匀的图像，出现了严重的过分割现象。
因此，本领域需要一种能够准确分割超声图像的方法。
发明内容
针对现有技术中存在的上述技术问题，本发明的目的是提出一种超声图像分割算法，以提高超声图像的分割准确性。
本发明的技术方案如下。
本发明第一方面提供一种图像分割方法，包括如下步骤：
步骤S1：获取待处理的图像；
步骤S2：利用多方向的Gabor小波对原图像进行滤波分解，得到多幅中间图像；
步骤S3：通过取最大值的方式对所述多幅中间图像进行融合，得到一幅增强图像；
步骤S4：针对所述中间图像，构造对应的水平集能量方程；
步骤S5：对所述能量方程的参数进行优化，使能量函数取最小值，获得边缘的准确位置；
步骤S6：重复所述步骤S5，直到能量函数取到最小值，获得最终边缘。
优选地，所述步骤S2中的多方向Gabor小波变化函数为
g q(x,y)＝g(x cos θ+y sin θ,-x sin θ+y cos θ)
其中g(x,y)为二维空间下的Gabor函数；θ＝qπ/Q，Q表示小波变换的方向总数，q为方向参数；将不同方向的Gabor小波与原图进行卷积得到多幅中间图像，如下式所示：
I q(x,y)＝I(x,y)*g q(x,y)q＝0,1,……Q-1
其中I(x,y)为原图数据，I q(x,y)为中间图像数据。
优选地，所述步骤S3中对所述多幅中间图像进行融合的融合方程如下：
I′(x,y)＝max{I q(x,y)，q＝0,1,……,Q-1}
其中I′(x,y)为融合图像数据。
优选地，所述能量方程包括：和图像本身相关的能量泛函项、通过限制边缘长度来保持边缘平滑的长度正则项，以及避免水平集方程重新初始化的距离正则项。
优选地，通过假设灰度不均图像是灰度偏置项与真实图像相加权再加入噪声后得到的结果，且构造不同方差大小的高斯模版来计算能量泛函项；高斯模板如下：
Figure PCTCN2020107283-appb-000001
其中σ p＝σ 0×p；所述能量方程形式如下：
Figure PCTCN2020107283-appb-000002
其中，ε(φ,c,b)是和图像本身相关的能量泛函项
Figure PCTCN2020107283-appb-000003
其中P是不同尺度的个数，
Figure PCTCN2020107283-appb-000004
是多尺度圆邻域模板，M 1(φ)＝H(φ)，M 2(φ)＝1-H(φ)；H(x)是Heaviside函数；
Figure PCTCN2020107283-appb-000005
和
Figure PCTCN2020107283-appb-000006
是两项正则项，
Figure PCTCN2020107283-appb-000007
是用来计算水平集方程φ的零水平面轮廓的长度，通过限制弧长来迫使水平集轮廓光滑的长度正则项；它的表达式如下：
Figure PCTCN2020107283-appb-000008
Figure PCTCN2020107283-appb-000009
是距离正则项，它可以使水平集函数在迭代时保持稳定，避免了水平集的重新初始化，该项的表达式如下：
Figure PCTCN2020107283-appb-000010
Figure PCTCN2020107283-appb-000011
优选地，对φ,c,b进行优化，使能量函数取最小值，对三个参数进行分别优化，优化时先固定其他两个参数的值，利用欧拉-拉格朗日公式优化水平集方程φ，利用偏微分方程方程优化c,b，则对φ的优化方程如下所示：
Figure PCTCN2020107283-appb-000012
Figure PCTCN2020107283-appb-000013
固定φ,b，对c进行优化：
Figure PCTCN2020107283-appb-000014
固定φ,c,对b进行优化：
Figure PCTCN2020107283-appb-000015
其中
本发明第二方面提供一种图像处理装置，包括：
存储部，其用于存储待处理的图像；
处理部，其被配置为根据以上技术方案之一所述的方法，获取所述存储装置中的图像中的边缘信息。
本发明第三方面提供一种超声波成像装置，包括：
超声波探头，其用于向被检对象发送超声波，并接收所述被检对象反射的超声波，生成与所述反射的超声波对应的回波信号；
生成部，其用于根据所述回波信号生成与所述被检对象有关的超声波图像；
处理部，其用于根据以上技术方案之一所述的方法，获取所述超声波图像中的边缘信息。
本发明第四方面提供一种超声波内镜，其中包括插入部、控制部，以及根据本发明第三方面所述的超声波成像装置。
通过以上技术方案，本发明提供了一种融合Gabor小波的多尺度局部聚类水平集超声图像分割方法，利用Gabor小波的多方向性对灰度不均的超声图像进行处理，通过最大值融合的方式得到增强待分割区域和背景间差异的中间图像。本发明将多尺度的思想引入局部聚类水平集算法中，设定不同方差大小的高斯核函数，通过均值融合的方式进行水平集迭代，得到最终边缘，从而克服了超声图像边缘较弱，分割不准的缺点，
附图说明
图1是根据本发明的图像分割方法流程示意图。
图2是根据本发明的图像处理装置结构示意图。
图3是根据本发明的超声波成像装置结构示意图。
具体实施方式
下面参照附图对根据本发明的图像分割方法、图像处理装置及超声波成像装置进行说明。
实施例1
本实施例的图像分割方法实施流程如图1所示：
步骤一：读入采集的超声图像。在一优选的实施方式中，所述超声图像为二维胃部剖面超声图像。
步骤二：用多方向的Gabor滤波器对图像进行滤波。二维空间中的Gabor函数如下式所示：
Figure PCTCN2020107283-appb-000016
式中，σ x和σ y分别表示高斯函数在x方向和y方向的展宽，W表示Gabor小波的频率带宽。多方向的Gabor小波变换函数可以表示为：
g q(x,y)＝g(x cos θ+y sin θ,-x sin θ+y cos θ)
其中θ＝qπ/Q，Q表示小波变换的方向总数。将不同方向的Gabor小波与原图进行卷积得到多幅中间图像。如下式所示：
I q(x,y)＝I(x,y)*g q(x,y)q＝0,1,……Q-1
步骤三：为了最大的保留原始图像的细节，增强待分割前景与背景间的差异，利用取最大值的方式对多幅中间图像进行融合，得到增强后的图像。
I′(x,y)＝max{I q(x,y)，q＝0,1,……,Q-1}
步骤四：针对中间图像，构造对应的水平集能量方程。整幅图像的能量方程由三部分组成，包括和图像本身相关的能量泛函项、通过限制边缘长度来保持边缘平滑的长度正则项和避免水平集方程重新初始化的距离正则项。通过假设灰度不均图像是灰度偏置项与真实图像相加权再加入噪声后得到的结果，且构造不同方差大小的高斯模版来计算能量泛函项。高斯模版如下所示：
Figure PCTCN2020107283-appb-000017
Figure PCTCN2020107283-appb-000018
其中
Figure PCTCN2020107283-appb-000019
P是不同尺度的个数，M 1(φ)＝H(φ)，M 2(φ)＝1-H(φ)。H(x)是Heaviside函数。式中
Figure PCTCN2020107283-appb-000020
p(s)的函数形式如下：
Figure PCTCN2020107283-appb-000021
步骤五：对φ,c,b进行优化，使能量函数取最小值，对三个参数进行分别优化，优化时先固定其他两个参数的值，利用欧拉-拉格朗日公式优化水平集方程φ，利用偏微分方程方程优化c,b，则对φ的优化方程如下所示：
Figure PCTCN2020107283-appb-000022
Figure PCTCN2020107283-appb-000023
固定φ,b，对c进行优化：
Figure PCTCN2020107283-appb-000024
固定φ,c,对b进行优化：
Figure PCTCN2020107283-appb-000025
步骤六：重复对步骤5中的优化方程进行计算，直到能量函数取到最小值，获得最终边缘。
实施例2
如图2所示，本实施例提供一种图像处理装置，包括存储部和处理部。
所述存储部用于存储待处理的图像；
所述处理部被配置为根据以上实施例1中任一实施方式所述的方法，获取所述存储装置中的图像中的边缘信息。
实施例3
如图3所示，本实施例提供一种超声波成像装置，包括超声波探头、生成部和处理部。
所述超声波探头用于向被检对象发送超声波，并接收所述被检对象反射的超声波，生成与所述反射的超声波对应的回波信号。
所述生成部用于根据所述回波信号生成与所述被检对象有关的超声波图像。
所述处理部用于根据以上实施例1中任一实施方式所述的方法，获取所述超声波图像中的边缘信息。
本领域技术人员能够理解，虽然以上各实施方式涉及的图像分割方法将超声波图像作为处理对象。然而，本发明并不限定于此。即，本发明各个实施方式涉及的图像分割方法除了可以将超声波图像作为处理对象以外，还可以处理各种具有适于该方法处理的灰度图像，诸如由X射线计算机断层摄影装置生成的CT图像、由X射线诊断装置生成的X射线图像、由磁共振成像装置生成的MR图像。
针对本发明的几个实施方式进行了说明，但这些实施方式是作为例子而示出的，并不意图限定发明的范围。这些新的实施方式可以通过其他各种方式来实施，在不脱离发明的要旨的范围内，可以进行各种省略、置换、变更。这些实施方式和其变形在包含在发明的范围和要旨内，并且包含在权利要求书所记载的发明和它的等同范围内。

Claims (9)

  1. 一种图像分割方法，包括如下步骤：
    步骤S1：获取待处理的图像；
    步骤S2：利用多方向的Gabor小波对原图像进行滤波分解，得到多幅中间图像；
    步骤S3：通过取最大值的方式对所述多幅中间图像进行融合，得到一幅增强图像；
    步骤S4：针对所述中间图像，构造对应的水平集能量方程；
    步骤S5：对所述能量方程的参数进行优化，使能量函数取最小值，获得边缘的准确位置；
    步骤S6：重复所述步骤S5，直到能量函数取到最小值，获得最终边缘。
  2. 根据权利要求1所述的图像分割方法，其特征在于，所述步骤S2中的多方向Gabor小波变化函数为
    g q(x,y)＝g(x cosθ+y sinθ,-x sinθ+y cosθ)
    其中g(x,y)为二维空间下的Gabor函数；θ＝qπ/Q，Q表示小波变换的方向总数，q为方向参数；将不同方向的Gabor小波与原图进行卷积得到多幅中间图像，如下式所示：
    I q(x,y)＝I(x,y)*g q(x,y) q＝0,1,……Q-1
    其中I(x,y)为原图数据，I q(x,y)为中间图像数据。
  3. 根据权利要求1所述的图像分割方法，其特征在于，所述步骤S3中对所述多幅中间图像进行融合的融合方程如下：
    I′(x,y)＝max{I q(x,y)，q＝0,1,……,Q-1}
    其中I′(x,y)为融合图像数据。
  4. 根据权利要求1所述的图像分割方法，其特征在于，所述能量方程包括：和图像本身相关的能量泛函项、通过限制边缘长度来保持边缘平滑的长度正则项，以及避免水平集方程重新初始化的距离正则项。
  5. 根据权利要求4所述的图像分割方法，其特征在于，通过假设灰度不均图像是灰度偏置项与真实图像相加权再加入噪声后得到的结果，且构造不同方差大小的高斯模版来计算能量泛函项；高斯模板如下：
    Figure PCTCN2020107283-appb-100001
    其中σ p＝σ 0×p；所述能量方程形式如下：
    Figure PCTCN2020107283-appb-100002
    其中，ε(φ,c,b)是和图像本身相关的能量泛函项
    Figure PCTCN2020107283-appb-100003
    其中P是不同尺度的个数，
    Figure PCTCN2020107283-appb-100004
    是多尺度圆邻域模板，M 1(φ)＝H(φ)，M 2(φ)＝1-H(φ)；H(x)是Heaviside函数；
    Figure PCTCN2020107283-appb-100005
    和
    Figure PCTCN2020107283-appb-100006
    是两项正则项，
    Figure PCTCN2020107283-appb-100007
    是用来计算水平集方程φ的零水平面轮廓的长度，通过限制弧长来迫使水平集轮廓光滑的长度正则项；它的表达式如下：
    Figure PCTCN2020107283-appb-100008
    Figure PCTCN2020107283-appb-100009
    是距离正则项，它可以使水平集函数在迭代时保持稳定，避免了水平集的重新初始化，该项的表达式如下：
    Figure PCTCN2020107283-appb-100010
    Figure PCTCN2020107283-appb-100011
    其中s表示
  6. 根据权利要求5所述的图像分割方法，其特征在于，对φ,c,b进行优化，使能量函数取最小值，对三个参数进行分别优化，优化时先固定其他两个参数的值，利用欧拉-拉格朗日公式优化水平集方程φ，利用偏微分方程方程优化c,b，则对φ的优化方程如下所示：
    Figure PCTCN2020107283-appb-100012
    Figure PCTCN2020107283-appb-100013
    固定φ,b，对c进行优化：
    Figure PCTCN2020107283-appb-100014
    固定φ,c,对b进行优化：
    Figure PCTCN2020107283-appb-100015
    其中
  7. 一种图像处理装置，其特征在于，包括：
    存储部，其用于存储待处理的图像；
    处理部，其被配置为根据权利要求1-6之一所述的方法，获取所述存储装置中的图像中的边缘信息。
  8. 一种超声波成像装置，其特征在于，包括：
    超声波探头，其用于向被检对象发送超声波，并接收所述被检对象反射的超声波，生成与所述反射的超声波对应的回波信号；
    生成部，其用于根据所述回波信号生成与所述被检对象有关的超声波图像；
    处理部，其用于根据权利要求1-6之一所述的方法，获取所述超声波图像中的边缘信息。
  9. 一种超声波内镜，其特征在于，包括插入部、控制部，以及根据权利要求8所述的超声波成像装置。
PCT/CN2020/107283 2020-04-23 2020-08-06 一种融合Gabor小波的多尺度局部水平集超声图像分割方法 WO2021212693A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202080100122.1A CN115843373A (zh) 2020-04-23 2020-08-06 一种融合Gabor小波的多尺度局部水平集超声图像分割方法
US18/048,527 US20230070200A1 (en) 2020-04-23 2022-10-21 Gabor wavelet-fused multi-scale local level set ultrasonic image segmentation method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CNPCT/CN2020/086248 2020-04-23
CNPCT/CN2020/086248 2020-04-23

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
CNPCT/CN2020/086248 Continuation 2020-04-23 2020-04-23

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/048,527 Continuation-In-Part US20230070200A1 (en) 2020-04-23 2022-10-21 Gabor wavelet-fused multi-scale local level set ultrasonic image segmentation method

Publications (1)

Publication Number Publication Date
WO2021212693A1 true WO2021212693A1 (zh) 2021-10-28

Family

ID=78271085

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/107283 WO2021212693A1 (zh) 2020-04-23 2020-08-06 一种融合Gabor小波的多尺度局部水平集超声图像分割方法

Country Status (3)

Country Link
US (1) US20230070200A1 (zh)
CN (1) CN115843373A (zh)
WO (1) WO2021212693A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114445426A (zh) * 2022-01-28 2022-05-06 深圳大学 内窥镜图像中息肉区域的分割方法、装置及相关组件
CN116843687A (zh) * 2023-09-01 2023-10-03 山东东方智光网络通信有限公司 一种通信光缆表面瑕疵检测方法及装置

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116703954B (zh) * 2023-06-16 2024-04-16 江南大学 一种基于全局预拟合能量驱动的主动轮廓模型方法及系统
CN117058125B (zh) * 2023-09-01 2024-03-15 无锡维凯科技有限公司 基于手机后盖玻璃的检测方法和系统
CN116993625B (zh) * 2023-09-26 2023-12-15 天津市胸科医院 一种气管镜图像增强方法及系统
CN117437279B (zh) * 2023-12-12 2024-03-22 山东艺达环保科技有限公司 一种包装盒表面平整度检测方法及系统
CN117764994B (zh) * 2024-02-22 2024-05-10 浙江首鼎视介科技有限公司 基于人工智能的胆胰成像系统及方法
CN117853494B (zh) * 2024-03-08 2024-06-14 山东中泰药业有限公司 一种药物颗粒生产过程视觉检测方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106296649A (zh) * 2016-07-21 2017-01-04 北京理工大学 一种基于水平集模型的纹理图像分割方法
CN106780508A (zh) * 2016-09-23 2017-05-31 北京联合大学 一种基于Gabor变换的GrabCut纹理图像分割方法
CN107153833A (zh) * 2016-03-03 2017-09-12 富士胶片株式会社 图像处理装置及其工作方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107153833A (zh) * 2016-03-03 2017-09-12 富士胶片株式会社 图像处理装置及其工作方法
CN106296649A (zh) * 2016-07-21 2017-01-04 北京理工大学 一种基于水平集模型的纹理图像分割方法
CN106780508A (zh) * 2016-09-23 2017-05-31 北京联合大学 一种基于Gabor变换的GrabCut纹理图像分割方法

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
BERTA SANDBERG , TONY CHAN , LUMINITA VESE: "A Level-Set and Gabor-based Active Contour Algorithm for Segmenting Textured Images", CITESEERX, 8 January 2016 (2016-01-08), pages 1 - 10, XP055861274 *
LI YUNHONG;WEI NINA;ZHANG XIAODAN: "Image segmentation based on Multi-direction Gabor Filters", FOREIGN ELECTRONIC MEASUREMENT TECHNOLOGY, vol. 36, no. 3, 15 March 2017 (2017-03-15), pages 20 - 23, XP055861266, DOI: 10.19652/j.cnki.femt.2017.03.005 *
ZHANG LI-HE,ZHU LI-LI,MI XIAO-LI: "Localized Multi-Channel Level Set Segmentation Combined with Gabor Texture Feature", ACTA ELECTRONICA SINICA, vol. 39, no. 7, 31 July 2011 (2011-07-31), pages 1569 - 1574, XP055861260, ISSN: 0372-2112 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114445426A (zh) * 2022-01-28 2022-05-06 深圳大学 内窥镜图像中息肉区域的分割方法、装置及相关组件
CN114445426B (zh) * 2022-01-28 2022-08-26 深圳大学 内窥镜图像中息肉区域的分割方法、装置及相关组件
CN116843687A (zh) * 2023-09-01 2023-10-03 山东东方智光网络通信有限公司 一种通信光缆表面瑕疵检测方法及装置

Also Published As

Publication number Publication date
CN115843373A (zh) 2023-03-24
US20230070200A1 (en) 2023-03-09

Similar Documents

Publication Publication Date Title
WO2021212693A1 (zh) 一种融合Gabor小波的多尺度局部水平集超声图像分割方法
Gerard et al. FissureNet: a deep learning approach for pulmonary fissure detection in CT images
Belaid et al. Phase-based level set segmentation of ultrasound images
US9955951B2 (en) Sensor coordinate calibration in an ultrasound system
US8189890B2 (en) Computer-aided classification of anomalies in anatomical structures
US8577129B2 (en) System and method for three-dimensional image rendering and analysis
Ilesanmi et al. A method for segmentation of tumors in breast ultrasound images using the variant enhanced deep learning
US8175348B2 (en) Segmenting colon wall via level set techniques
US20080205717A1 (en) System and method for three-dimensional image rendering and analysis
US9014447B2 (en) System and method for detection of lesions in three-dimensional digital medical image
US7457448B2 (en) Method and system for wavelet based detection of colon polyps
US7480401B2 (en) Method for local surface smoothing with application to chest wall nodule segmentation in lung CT data
JP2006006359A (ja) 画像生成装置、画像生成装置方法、およびそのプログラム
Karargyris et al. Identification of polyps in wireless capsule endoscopy videos using log gabor filters
Hermawati et al. Phase-based thresholding schemes for segmentation of fetal thigh cross-sectional region in ultrasound images
US9545242B2 (en) Sensor coordinate calibration in an ultrasound system
Wen et al. A novel Bayesian-based nonlocal reconstruction method for freehand 3D ultrasound imaging
Silva et al. Algorithm versus physicians variability evaluation in the cardiac chambers extraction
Ratheesh et al. Advanced algorithm for polyp detection using depth segmentation in colon endoscopy
Pengiran Mohamad et al. Transition of traditional method to deep learning based computer-aided system for breast cancer using automated breast ultrasound system (abus) images: a review
Cerrolaza et al. Quantification of kidneys from 3D ultrasound in pediatric hydronephrosis
EP3608872B1 (en) Image segmentation method and system
CN112767403A (zh) 医学影像分割模型训练方法、医学影像分割方法及装置
Lang et al. Multi-scale texture-based level-set segmentation of breast B-mode images
Rahmatullah et al. Anatomical object detection in fetal ultrasound: computer-expert agreements

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: 20932667

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: 20932667

Country of ref document: EP

Kind code of ref document: A1