CN110060275B - Method and system for detecting human body microcirculation blood flow velocity - Google Patents
Method and system for detecting human body microcirculation blood flow velocity Download PDFInfo
- Publication number
- CN110060275B CN110060275B CN201910288803.4A CN201910288803A CN110060275B CN 110060275 B CN110060275 B CN 110060275B CN 201910288803 A CN201910288803 A CN 201910288803A CN 110060275 B CN110060275 B CN 110060275B
- Authority
- CN
- China
- Prior art keywords
- image
- module
- projection
- frame image
- moving target
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
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
- G06T7/12—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
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/215—Motion-based segmentation
-
- 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
- G06T2207/10056—Microscopic 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
- G06T2207/20036—Morphological image processing
-
- 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
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
- G06T2207/30104—Vascular flow; Blood flow; Perfusion
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for detecting the blood flow velocity of human microcirculation, wherein the method comprises the following steps: carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V; performing background modeling on the video image sequence V by adopting a Gaussian mixture model; segmenting the blood vessels in each frame image to obtain the outlines and the central lines of the blood vessels; detecting and tracking a moving target in a blood vessel by adopting a background difference method, calculating to obtain the position of the moving target and generating a motion trail diagram; and performing projection calculation on the motion trail graph to obtain the speed of the moving target. The invention applies the background difference and projection methods to the detection of the human microcirculation blood flow velocity, can quickly and accurately realize the detection of the microcirculation blood flow velocity, has important reference value and auxiliary diagnosis function for the clinical diagnosis of some diseases, and has great value for the detection of the health condition of the human body, the judgment of the curative effect of the medicine and the like.
Description
Technical Field
The invention relates to a method and a system for detecting the blood flow velocity of human microcirculation, belonging to the technical field of artificial intelligence.
Background
In recent years, with the intensive research on cardiovascular diseases by traditional and western medicine, a great deal of attention is paid to the microcirculation research by medical systems, and particularly, the traditional Chinese medicines in many traditional Chinese medicines have remarkable effects on treating cardiovascular diseases in clinic, but the mechanisms of the traditional Chinese medicines are not clear, and the traditional Chinese medicines cannot be superior to each other in medicinal effects. For the study of human microcirculation, it is mainly to study the measurement of morphological parameters of microcirculation, such as capillary density, blood flow velocity in capillary, cell aggregation degree, etc. Wherein the blood flow velocity in the microcirculation is an intuitive and important parameter indicator.
There are two main methods for measuring the blood flow velocity of the microcirculation, one is to measure by a doppler analyzer, and the other is to analyze and calculate the microcirculation video. The method of the Doppler analyzer is troublesome to use and high in cost, and the manufacturing cost of one Doppler analyzer is hundreds of thousands. Thus, much research has been conducted by the video analysis method, which has been developed. At present, the main research methods of video analysis include a spatial correlation method, an optical flow method, a particle image velocimetry method and a method based on a space-time diagram.
The spatial correlation method obtains the moving distance of the position of the regional window in the video sequence by searching the maximum cross correlation coefficient, and the instantaneous speed is the result of removing the interval time of the video sequence by spatial bits. The optical flow method is used for estimating the flow velocity in the X and Y directions, but the assumed condition is difficult to satisfy, namely, the brightness of the continuous frame image pixels is kept unchanged in a small range, so the optical flow method is more suitable for high-frame-rate and high-resolution video images. The particle image velocimetry mainly calculates the flow velocity by measuring the displacement of tracer particles added in a fluid during two times of passing a certain plane, but the particle image velocimetry has high requirements on equipment and needs higher space-time resolution. In recent years, a method based on ST maps is increasingly applied to quantitative measurement of blood flow velocity. ST-maps are an abbreviation of space-time-maps, and the idea of this method is to map the trajectory of the movement of the cell into a two-dimensional space-time map, converting the video-based flow velocity measurement into a directional measurement of the trajectory in the ST-map. However, this method is slightly insufficient in real-time because of a large amount of calculation when calculating the ST map.
Disclosure of Invention
Aiming at the defects of the method, the invention provides the method for detecting the human body microcirculation blood flow velocity, which can quickly and accurately detect the microcirculation blood flow velocity, has important reference value and auxiliary diagnosis function for clinical diagnosis of some diseases, and has great value for detecting the health condition of a human body, judging the curative effect of medication and the like.
The technical scheme adopted for solving the technical problems is as follows:
on one hand, the method for detecting the blood flow velocity of human microcirculation provided by the embodiment of the invention comprises the following steps:
step 1: carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V;
step 2: carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model;
and step 3: segmenting the blood vessel in each frame of image to obtain the contour and the central line of the blood vessel;
and 4, step 4: detecting and tracking a moving target in a blood vessel by adopting a background difference method, calculating to obtain the position of the moving target, and generating a motion trail diagram according to the position;
and 5: and projecting the motion trail diagram of the moving target, and calculating to obtain the speed of the moving target.
As a possible implementation manner of this embodiment, in step 1, the process of debouncing the nail fold microcirculation video image to obtain the stable video image sequence V includes the following steps:
step 11: acquiring a video image file of microcirculation of the anonymous nail plica by a microscopic camera;
step 12, graying each frame image G (x, y) of the nail plica microcirculation video by using a formula (1.1) to obtain an image G (x, y);
Gray＝0.114×B+0.587×G+0.299×R (1.1)
b, G, R is the three color components of each pixel point in the image;
step 13, selecting the first frame image of the micro-circulation video image sequence of the nail folds as a reference frame image, performing offset correction on each frame later by referring to the first frame image, projecting the images in the row direction and the column direction respectively by using a formula (1.2) to obtain two independent one-dimensional data,
wherein the content of the first and second substances,
respectively represent the k frame image g k The gray projection values of the ith row and the jth column of (i, j), M and N are the height and the width of the image respectively, M and N represent the length of the projection interval respectively, and M and N satisfy the following conditions: m is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to N;
step 14, calculating the correlation ρ (ω) between the line gray level projection curve of the k frame image and the line projection curve of the reference frame image according to the formula (1.3):
wherein, P k (i) Is the gray projection value of the ith line of the k frame image, P 1 (i) The gray projection value of the ith line of the first frame image is taken as q, the offset is a search range of one side relative to the reference frame, M-2q represents the height of the image after the search ranges on two sides are removed, namely the height of a matching area, and omega is the search offset;
step 15, assume ω min The value of omega when the rho (omega) is minimum, then the motion vector delta of the k frame image relative to the reference frame image in the vertical direction is calculated according to the formula (1.4) k ，
δ k ＝m+1-ω min (1.4)
Wherein, delta k Indicating for positive a movement of | δ of the frame to be matched upwards k L pixels, δ k Negative indicates a downward movement of | δ k L pixels;
step 16, calculating the motion vector delta of the current frame image in the horizontal direction by the same method as step 15 k For the horizontal direction shift-delta of the current frame image k Translating by-delta in the vertical direction k And obtaining the stable video image sequence V.
As a possible implementation manner of this embodiment, in step 2, the process of performing background modeling on the video image sequence V by using the gaussian mixture model includes the following steps:
step 21, selecting a first frame image from the video image sequence V to initialize a background model, a Gaussian functionThe desired initialization of numbers is the gray value of the image pixel, the standard deviation is initialized to a constant c, and the variance is c 2 ；
Step 22, background modeling is carried out on the video image sequence V by using a formula (2.1),
where g (x, y, t) is the gray value at the image coordinate (x, y) at time t, P (g (x, y, t)) is the probability of this gray value occurring, μ t And σ t Respectively the expectation and variance of the Gaussian distribution of the pixel at the time t;
step 23, detecting background pixels according to the formula (2.2),
|g(x,y,t)-μ t-1 (x,y)|<λσ t-1 (2.2)
wherein λ is a gaussian background parameter;
step 24, respectively updating the expectation, the standard deviation and the variance of the Gaussian model by using a formula (2.3), a formula (2.4) and a formula (2.5);
μ t (x,y)=(1-α)*μ t-1 (x,y)+α*g(x,y,t) (2.3)
and 25, repeating the step 23 and the step 24 until the background pixels are detected, and obtaining a background model GM.
As a possible implementation manner of this embodiment, in step 3, the process of segmenting the blood vessel in each frame of image to obtain the contour and the centerline of the blood vessel includes the following steps:
step 31, selecting a proper threshold value Th, binarizing the gray value GM (x, y) of any point in the scene model GM image according to a formula (3.1) to obtain a binary image B (x, y), and obtaining a blood vessel region set H by taking a white connected region, wherein the blood vessel region set H comprises a plurality of segmented blood vessels H;
and step 32, selecting a target blood vessel H from the blood vessel set H, refining by using a mathematical morphology skeleton extraction method, and extracting a skeleton to obtain a centerline E of the blood vessel H.
As a possible implementation manner of this embodiment, in step 4, the process of detecting and tracking the moving target in the blood vessel by using a background difference method, and calculating to obtain the position and the motion trajectory diagram of the moving target includes the following steps:
step 41, differentiating the current frame image f (x, y) and the background image GM (x, y) by using a formula (4.1) to obtain a moving target image D (x, y);
D(x,y)＝|f(x,y)-GM(x,y)| (4.1)
step 42, setting a threshold value T, binarizing the moving target image, if D (x, y) > T, then D (x, y) =255, otherwise, D (x, y) =0, and obtaining a white pixel area in the new binary image D (x, y) as the moving target;
step 43, performing and operation on the central line obtained in step 32 and the moving target D (x, y) to obtain an intersection point set HJ, and calculating the distance from each point in the HJ to one end of the central line to obtain a distance set Hd;
step 44, generating a motion trail graph W according to the distance set Hd of each frame, setting the initial pixel value of W to be 0, the abscissa represents the number of video frames, the ordinate represents the distance from each point on the central line to the central line end point, and for the ith frame image, the distance set is Hd i Then to Hd i Let W (i, a) =255, and the motion trajectory graph is a motion trajectory graph obtained by converting the measurement of the blood flow velocity in the microcirculation into the measurement of the motion target trajectory direction in the two-dimensional image, that is, calculating the slope of the trajectory in the motion trajectory graph to obtain the blood flow velocity.
As a possible implementation manner of this embodiment, the process of projecting the motion trajectory diagram of the moving object and calculating the speed of the moving object includes the following steps:
step 51, projecting the motion trail graph W in the S direction by using a formula (5.1), wherein the value of a projection angle theta is rotated to pi from 0 to obtain a projection set Hw;
wherein, M and N are the height and width of the image respectively, and the operator is ┚ to indicate the corresponding relation between the lower rounding and the projection coordinate (t, s) and the original coordinate (x, y);
step 52, calculating the variance of each histogram in the set Hw and putting the variance into the set H sigma;
step 53, selecting the largest element from the variance set H σ, and finding the corresponding projection histogram from the set Hw according to the element, thereby finding the projection angle θ of the projection histogram;
step 54, calculating the cosecant of the angle θ, and obtaining the blood flow velocity v, i.e. v = cot θ.
In another aspect, a system for detecting a blood flow velocity of human microcirculation according to an embodiment of the present invention includes:
the video image sequence acquisition module is used for carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V;
the background modeling module is used for carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model;
the image blood vessel segmentation module is used for segmenting blood vessels in each frame of image to obtain the outlines and the central lines of the blood vessels;
the motion trail module is used for detecting and tracking a motion target in a blood vessel by adopting a background difference method, calculating the position of the motion target and generating a motion trail graph according to the position;
and the projection module is used for projecting the motion trail diagram of the moving target and calculating to obtain the speed of the moving target.
As a possible implementation manner of this embodiment, the video image sequence obtaining module includes:
the video image file module is used for acquiring a video image file of the microcirculation of the anonymous nail plica through the microscopic camera;
the image graying module is used for graying each frame image G (x, y) of the nail fold microcirculation video to obtain an image G (x, y);
the image projection module is used for selecting a first frame image of the nail fold microcirculation video image sequence as a reference frame image, performing offset correction on each frame later by referring to the first frame image, and projecting the images in the row direction and the column direction respectively to obtain two independent one-dimensional data;
a correlation calculation module for calculating the correlation ρ (ω) between the line gray level projection curve of the kth frame image and the line projection curve of the reference frame image;
a motion vector calculation module for calculating a motion vector delta of the k frame image in a vertical direction with respect to the reference frame image k ；
An image translation module for calculating the motion vector delta of the current frame image in the horizontal direction k And horizontally shifting the current frame image by-delta k Is translated by-delta in the vertical direction k And obtaining the stable video image sequence V.
As a possible implementation manner of this embodiment, the background modeling module includes:
a background model initialization module for selecting a first frame image from the video image sequence V to initialize a background model, where the expectation of the Gaussian function is initialized to the gray value of the image pixel, the standard deviation is initialized to a constant c, and the variance is c 2 ；
The modeling module is used for carrying out background modeling on the video image sequence V;
the background pixel detection module is used for detecting background pixels;
and the updating module is used for updating the expectation, the standard deviation and the variance of the Gaussian model.
As a possible implementation manner of this embodiment, the image vessel segmentation module includes:
the background model binarization module is used for selecting a proper threshold Th, binarizing gray values GM (x, y) of any point in the scene model GM image to obtain a binary image B (x, y), and taking a white connected region to obtain a blood vessel region set H which comprises a plurality of segmented blood vessels H;
and the skeleton extraction module is used for selecting a target blood vessel H from the blood vessel set H, refining the target blood vessel H by using a mathematical morphology skeleton extraction method, and extracting a skeleton to obtain a central line E of the blood vessel H.
As a possible implementation manner of this embodiment, the motion trajectory module includes:
the difference module is used for carrying out difference on the current frame image f (x, y) and the background image GM (x, y) to obtain a moving target image D (x, y);
a moving target binarization module, configured to set a threshold T, binarize a moving target image, if D (x, y) > T, D (x, y) =255, otherwise, D (x, y) =0, and obtain a white pixel area in a new binary image D (x, y) as a moving target;
the distance set module is used for performing AND operation on the central line and the moving target D (x, y) to obtain an intersection set HJ, and calculating the distance from each point in the HJ to one end of the central line to obtain a distance set Hd;
a motion trail graph generating module for generating a motion trail graph W according to the distance set Hd of each frame, setting the initial pixel values of the motion trail graph W to be 0, the abscissa represents the number of video frames, the ordinate represents the distance from each point on the central line to the central line end point, and for the ith frame image, the distance set is Hd i Then to Hd i Let W (i, a) =255, and the motion trajectory graph is a motion trajectory graph obtained by converting the measurement of the blood flow velocity in the microcirculation into the measurement of the motion target trajectory direction in the two-dimensional image, that is, calculating the slope of the trajectory in the motion trajectory graph to obtain the blood flow velocity.
As a possible implementation manner of this embodiment, the projection module includes:
the S direction projection module is used for projecting the motion trail diagram W in the S direction, and the value of the projection angle theta is rotated to pi from 0 to obtain a projection set Hw;
the variance calculation module is used for calculating the variance of each histogram in the set Hw and putting the variance into the set H sigma;
the projection angle module is used for selecting the largest element from the variance set H sigma and finding the corresponding projection histogram from the set Hw according to the element so as to find the projection angle theta of the projection histogram;
and the angle cotangent calculation module is used for calculating the cotangent of the angle theta to obtain the blood flow velocity v, namely v = cot theta.
The technical scheme of the embodiment of the invention has the following beneficial effects:
the method for detecting the human microcirculation blood flow velocity comprises the steps of firstly carrying out debouncing operation on a nail fold microcirculation video image to obtain a stable video image sequence V, then carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model, then segmenting blood vessels in each frame of image to obtain the outline and the central line of the blood vessels, then carrying out detection and tracking on a moving target in the blood vessels by adopting a background difference method, calculating to obtain the position of the moving target and generating a moving track graph according to the position, and finally projecting the moving track graph of the moving target to calculate the velocity of the moving target. The method is mainly used for automatically detecting the speed of the microcirculation blood flow of the nail folds, applies the methods of background difference and projection to the detection of the speed of the microcirculation blood flow of the human body, can quickly and accurately realize the detection of the speed of the microcirculation blood flow, has important reference value and auxiliary diagnosis function for the clinical diagnosis of certain diseases, and has great value for the detection of the health condition of the human body, the judgment of the curative effect of medication and the like.
The system for detecting the human body microcirculation blood flow speed in the technical scheme of the embodiment of the invention comprises: the video image sequence acquisition module is used for carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V; the background modeling module is used for carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model; the image blood vessel segmentation module is used for segmenting blood vessels in each frame of image to obtain the outlines and the central lines of the blood vessels; the motion trail module is used for detecting and tracking a motion target in a blood vessel by adopting a background difference method, calculating to obtain the position of the motion target and generating a motion trail map according to the position; and the projection module is used for projecting the motion trail diagram of the moving target and calculating to obtain the speed of the moving target. The method is mainly used for automatically detecting the speed of the microcirculation blood flow of the nail folds, applies the methods of background difference and projection to the detection of the speed of the microcirculation blood flow of the human body, can quickly and accurately realize the detection of the speed of the microcirculation blood flow, has important reference value and auxiliary diagnosis function for the clinical diagnosis of certain diseases, and has great value for the detection of the health condition of the human body, the judgment of the curative effect of medication and the like.
Description of the drawings:
FIG. 1 is a flow chart illustrating a method of human microcirculation blood velocity detection according to an exemplary embodiment;
FIG. 2 is a frame image of a nail fold microcirculation video;
FIG. 3 is a graph of plasma movement traces of the nail fold microcirculation blood vessel;
fig. 4 is a schematic diagram of the correspondence between the projection coordinates (t, s) of the motion trajectory and the original coordinates (x, y);
FIG. 5 is a schematic diagram illustrating a system for human microcirculation blood flow velocity detection according to an exemplary embodiment.
Detailed Description
The invention is further illustrated by the following examples in conjunction with the accompanying drawings:
in order to clearly explain the technical features of the present invention, the following detailed description of the present invention is provided with reference to the accompanying drawings. The following disclosure provides many different embodiments, or examples, for implementing different features of the invention. To simplify the disclosure of the present invention, the components and arrangements of specific examples are described below. Moreover, the present disclosure may repeat reference numerals and/or letters in the various examples. This repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and/or configurations discussed. It should be noted that the components illustrated in the figures are not necessarily drawn to scale. Descriptions of well-known components and processing techniques and procedures are omitted so as to not unnecessarily limit the invention.
FIG. 1 is a flow chart illustrating a method of human microcirculation blood velocity detection according to an exemplary embodiment. As shown in fig. 1, a method for detecting blood flow velocity in human microcirculation according to an embodiment of the present invention includes the following steps:
step 1: and carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V.
The process of debouncing a plica microcirculation video image to obtain a stabilized video image sequence V comprises the following steps:
step 11: acquiring a video image file of microcirculation of the anonymous nail plica by a microscopic camera;
step 12, performing graying on each frame image G (x, y) of the fold microcirculation video shown in fig. 2 by using a formula (1.1) to obtain an image G (x, y);
Gray＝0.114×B+0.587×G+0.299×R (1.1)
b, G, R is the three color components of each pixel point in the image;
step 13, selecting the first frame image of the nail fold microcirculation video image sequence as the reference frame image, making offset correction for each frame later with reference to the first frame image, projecting the images in the row direction and the column direction respectively by using a formula (1.2) to obtain two independent one-dimensional data,
wherein the content of the first and second substances,
respectively represent the k-th frameImage g k (i, j) the ith row and the jth column of gray projection values, M and N are the height and width of the image, respectively, M and N represent the projection interval length, respectively, and M and N satisfy the following conditions: m is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to N;
step 14, calculating the correlation ρ (ω) between the line gray level projection curve of the k frame image and the line projection curve of the reference frame image according to the formula (1.3):
wherein, P k (i) Is the gray projection value of the ith line of the k frame image, P 1 (i) The gray projection value of the ith line of the first frame image is obtained, q is a search range of the offset on one side relative to a reference frame, M-2q represents the height of the image after the search ranges on two sides are removed, namely the height of a matching area, and omega is the search offset;
step 15, assume ω min The value of omega when the rho (omega) is minimum, then the motion vector delta of the k frame image relative to the reference frame image in the vertical direction is calculated according to the formula (1.4) k ，
δ k ＝m+1-ω min (1.4)
Wherein, delta k Indicating for positive a movement of | δ of the frame to be matched upwards k L pixels, δ k Negative indicates a downward movement of | δ k L pixels;
step 16, calculating the motion vector Δ of the current frame image in the horizontal direction by the same method as step 15 k For the horizontal direction shift-delta of the current frame image k Is translated by-delta in the vertical direction k And obtaining the stable video image sequence V.
And 2, step: and carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model.
The process of background modeling a video image sequence V using a gaussian mixture model comprises the steps of:
step 21, selecting a first frame image from the video image sequence V to initialize a background model, a period of a gaussian functionIf the gray value of the image pixel is initialized and the standard deviation is initialized to a constant c, the variance is c 2 ；
Step 22, background modeling is carried out on the video image sequence V by using a formula (2.1),
where g (x, y, t) is the gray value at the image coordinate (x, y) at time t, P (g (x, y, t)) is the probability of this gray value occurring, μ t And σ t Respectively the expectation and variance of the Gaussian distribution of the pixel at the time t;
step 23, detecting background pixels according to the formula (2.2),
|g(x,y,t)-μ t-1 (x,y)|<λσ t-1 (2.2)
wherein λ is a gaussian background parameter;
step 24, respectively updating the expectation, the standard deviation and the variance of the Gaussian model by using a formula (2.3), a formula (2.4) and a formula (2.5);
μ t (x,y)=(1-α)*μ t-1 (x,y)+α*g(x,y,t) (2.3)
and step 25, repeating the step 23 and the step 24 until the background pixels are detected, and obtaining a background model GM.
And step 3: and segmenting the blood vessels in each frame of image to obtain the outlines and the central lines of the blood vessels.
The process of segmenting the blood vessel in each frame of image to obtain the contour and the midline of the blood vessel comprises the following steps:
step 31, selecting a proper threshold value Th, binarizing the gray value GM (x, y) of any point in the scene model GM image according to a formula (3.1) to obtain a binary image B (x, y), and obtaining a blood vessel region set H by taking a white connected region, wherein the blood vessel region set H comprises a plurality of segmented blood vessels H;
and 32, selecting a target blood vessel H from the blood vessel set H, refining by using a mathematical morphology framework extraction method, and extracting a framework to obtain a centerline E of the blood vessel H.
And 4, step 4: and detecting and tracking the moving target in the blood vessel by adopting a background difference method, calculating to obtain the position of the moving target, and generating a motion trail diagram according to the position.
The process of detecting and tracking the moving target in the blood vessel by adopting a background difference method and calculating to obtain the position and the motion trail diagram of the moving target comprises the following steps:
step 41, differentiating the current frame image f (x, y) and the background image GM (x, y) by using a formula (4.1) to obtain a moving target image D (x, y);
D(x,y)＝|f(x,y)-GM(x,y)| (4.1)
step 42, setting a threshold value T, binarizing the moving target image, if D (x, y) > T, then D (x, y) =255, otherwise, D (x, y) =0, and obtaining a white pixel area in the new binary image D (x, y) as the moving target;
step 43, performing and operation on the central line obtained in step 32 and the moving target D (x, y) to obtain an intersection point set HJ, and calculating the distance from each point in the HJ to one end of the central line to obtain a distance set Hd;
step 44, generating a motion trail graph W according to the distance set Hd of each frame, setting the initial pixel value of W to be 0, the abscissa represents the number of video frames, the ordinate represents the distance from each point on the central line to the central line end point, and for the ith frame image, the distance set is Hd i Then to Hd i Let W (i, a) =255, and the motion trajectory graph is to convert the measurement of blood flow velocity in the microcirculation into two-dimensional imageThe blood flow velocity can be obtained by measuring the trajectory direction of the middle moving object, that is, calculating the slope of the trajectory in the moving trajectory diagram, and the obtained moving trajectory diagram is shown in fig. 3.
And 5: and projecting the motion trail diagram of the moving target, and calculating to obtain the speed of the moving target.
The process of projecting the motion trail diagram of the moving target and calculating the speed of the moving target comprises the following steps:
step 51, projecting the motion trail graph W in the S direction by using a formula (5.1), wherein the value of a projection angle theta is rotated to pi from 0 to obtain a projection set Hw;
wherein M and N are the height and width of the image, respectively, and the operator is ┚ with a person who grasps it, e.g., X ┚ performs the round-down operation on X; the correspondence between the projection coordinates (t, s) and the original coordinates (x, y) is shown in fig. 4;
step 52, calculating the variance of each histogram in the set Hw and putting the variance into the set H sigma;
step 53, selecting the largest element from the variance set H σ, and finding the corresponding projection histogram from the set Hw according to the element, thereby finding the projection angle θ of the projection histogram;
at step 54, the complementary cut of the angle θ is calculated to obtain the blood flow velocity v, i.e., v = cot θ. :
the invention applies the background difference and projection method to the detection of the blood velocity of the human microcirculation, and can quickly and accurately realize the detection of the blood velocity of the microcirculation.
FIG. 5 is a schematic diagram of a human microcirculation blood flow velocity detection system based on background difference and projection according to an exemplary embodiment. As shown in fig. 5, the system for detecting blood flow velocity of human microcirculation based on background difference and projection provided in this embodiment includes:
the video image sequence acquisition module is used for carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V;
the background modeling module is used for carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model;
the image blood vessel segmentation module is used for segmenting blood vessels in each frame of image to obtain the outlines and the central lines of the blood vessels;
the motion trail module is used for detecting and tracking a motion target in a blood vessel by adopting a background difference method, calculating to obtain the position of the motion target and generating a motion trail map according to the position;
and the projection module is used for projecting the motion trail diagram of the moving target and calculating to obtain the speed of the moving target.
In a possible implementation manner of this embodiment, the video image sequence obtaining module includes: the device comprises a video image file module, an image graying module, an image projection module, a correlation calculation module, a motion vector calculation module and an image translation module;
the video image file module acquires a video image file of the microcirculation of the nail folds of the ring finger through the micro-camera; the image graying module performs graying processing on an image G (x, y) of each frame of the fold microcirculation video shown in the figure 2 by using a formula (1.1) to obtain an image G (x, y);
Gray＝0.114×B+0.587×G+0.299×R (1.1)
b, G, R is the three color components of each pixel point in the image;
the image projection module selects a first frame of a video image sequence as a reference frame image, performs offset correction on each subsequent frame with reference to the first frame, respectively projects the images in the row direction and the column direction by using a formula (1.2) to obtain two independent one-dimensional data,
wherein the content of the first and second substances,
respectively represent the k frame image g k (i, j) the ith row and the jth column of gray projection values, M and N are the height and width of the image, respectively, M and N represent the projection interval length, respectively, and M and N satisfy the following conditions: m is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to N;
the correlation calculation module calculates the correlation between the line gray level projection curve of the k frame image and the line projection curve of the reference frame image according to the formula (1.3),
wherein, P k (i) Is the gray projection value, P, of the ith line of the kth frame image 1 (i) The gray projection value of the ith line of the first frame image is taken as q, the offset is a search range of one side relative to the reference frame, M-2q represents the height of the image after the search ranges on two sides are removed, namely the height of a matching area, and omega is the search offset;
let omega be min The motion vector calculation module can calculate the motion vector delta of the k frame image relative to the reference frame image in the vertical direction according to the formula (1.4) for the value of omega at the minimum value of rho (omega) k ，
δ k ＝m+1-ω min (1.4)
Wherein, delta k Indicating for positive a movement of | δ of the frame to be matched upwards k I pixels, negative indicates a downward movement of delta k L pixels;
by the same method, the image translation module can calculate the motion vector delta of the current frame image in the horizontal direction k For the horizontal direction shift-delta of the current frame image k Is translated by-delta in the vertical direction k And obtaining the stable video image sequence V.
In a possible implementation manner of this embodiment, the background modeling module includes: the device comprises a background model initialization module, a modeling module, a background pixel detection module and an updating module;
the background model initialization module selects a first frame image from VInitializing a background model, initializing the expectation of a Gaussian function to the gray value of an image pixel, initializing a standard deviation to a constant c, and then setting a variance to c 2 ；
The modeling module models V against the background using equation (2.1),
where g (x, y, t) is the gray value at the image coordinate (x, y) at time t, P (g (x, y, t)) is the probability of this gray value occurring, μ t And σ t Respectively the expectation and variance of the Gaussian distribution of the pixel at the time t;
the background pixel detection module detects background pixels according to equation (2.2),
|g(x,y,t)-μ t-1 (x,y)|<λσ t-1 (2.2)
wherein λ is a gaussian background parameter;
the updating module respectively updates the expectation, the standard deviation and the variance of the Gaussian model by using formulas (2.3), (2.4) and (2.5);
μ t (x,y)=(1-α)*μ t-1 (x,y)+α*g(x,y,t) (2.3)
and the background pixel detection module and the updating module are mutually alternated, detect background pixels and update the expectation, the standard deviation and the variance of the Gaussian model until the background pixels are detected completely, so as to obtain a background model GM.
In a possible implementation manner of this embodiment, the image vessel segmentation module includes: a background model binarization module and a skeleton extraction module;
the background model binarization module selects a proper threshold value Th for the gray value GM (x, y) of any point in the obtained background model GM image, then binarizes the image according to a formula (3.1) to obtain a binary image B (x, y), and a white connected region is selected to obtain a blood vessel region set H which comprises a plurality of segmented blood vessels H;
and the skeleton extraction module selects a target blood vessel H from the blood vessel set H, refines the target blood vessel H by using a mathematical morphology skeleton extraction method, and extracts a skeleton to obtain a midline E of the blood vessel H.
In a possible implementation manner of this embodiment, the motion trajectory module includes: the device comprises a difference module, a moving target binarization module, a distance collection module and a moving track map generation module;
the difference module uses a formula (4.1) to carry out difference on the current frame image f (x, y) and the background image GM (x, y) to obtain a moving target image D (x, y);
D(x,y)＝|f(x,y)-GM(x,y)| (4.1)
a moving target binarization module sets a threshold value T, binarizes the moving target image, if D (x, y) > T, D (x, y) =255, otherwise, D (x, y) =0, and obtains a white pixel area in a new binary image D (x, y) which is the moving target;
the distance collection module performs AND operation on the obtained central line and the moving target D (x, y) to obtain an intersection point collection HJ, and calculates the distance from each point in the HJ to one end of the central line to obtain a distance collection Hd;
the motion trail picture generation module generates a motion trail picture W according to the distance set Hd of each frame, the initial pixel value of W is set to be 0, the horizontal coordinate represents the number of video frames, the vertical coordinate represents the distance from each point on the central line to the central line end point, and for the ith frame of image, the distance set is Hd i Then to Hd i Let W (i, a) =255, that is, a motion trace map is a map for converting the measurement of the blood flow velocity in the microcirculation into the measurement of the track direction of a moving object in a two-dimensional image, that is, a motion trace mapThe blood flow velocity can be obtained by calculating the slope of the trajectory, and a motion trajectory diagram is obtained, as shown in fig. 3.
In a possible implementation manner of this embodiment, the projection module includes: the system comprises an S direction projection module, a variance calculation module, a projection angle module and an angle cotangent calculation module;
the S direction projection module projects the motion trail diagram W in the S direction by using a formula (5.1), and the value of a projection angle theta is rotated to pi from 0 to obtain a projection set Hw;
where M and N are the height and width of the image, respectively, an operator ┚ denotes lower rounding, and the correspondence between the projected coordinates (t, s) and the original coordinates (x, y) is as shown in fig. 4;
the variance calculation module calculates the variance of each histogram in the set Hw and puts the variance into the set H sigma;
the projection angle module selects the largest element from the variance set H sigma, and finds a corresponding projection histogram from the set Hw according to the element, so as to find the projection angle theta of the projection histogram;
the angular cotangent calculation module calculates the cotangent of the angle theta to obtain the blood flow velocity v, namely v = cot theta.
The method is mainly used for automatically detecting the speed of the microcirculation blood flow of the nail folds, applies the methods of background difference and projection to the detection of the speed of the microcirculation blood flow of the human body, can quickly and accurately realize the detection of the speed of the microcirculation blood flow, has important reference value and auxiliary diagnosis function for the clinical diagnosis of certain diseases, and has great value for the detection of the health condition of the human body, the judgment of the curative effect of medication and the like.
The foregoing is only a preferred embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the principle of the present invention, and these modifications and improvements are also considered to be within the scope of the present invention.
Claims (2)
1. A method for detecting the blood flow velocity of human microcirculation is characterized by comprising the following steps:
step 1: carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V;
step 2: performing background modeling on the video image sequence V by adopting a Gaussian mixture model;
and step 3: segmenting the blood vessel in each frame of image to obtain the contour and the central line of the blood vessel;
and 4, step 4: detecting and tracking a moving target in a blood vessel by adopting a background difference method, calculating to obtain the position of the moving target, and generating a motion trail diagram according to the position;
and 5: projecting the motion trail diagram of the moving target, and calculating to obtain the speed of the moving target;
in step 1, the process of debouncing the nail fold microcirculation video image to obtain the stable video image sequence V includes the following steps:
step 11: acquiring a video image file of microcirculation of the anonymous nail plica by a microscopic camera;
step 12, performing graying processing on each frame image G (x, y) of the nail plica microcirculation video by using a formula (1.1) to obtain an image G (x, y);
Gray＝0.114×B+0.587×G+0.299×R (1.1)
b, G, R is the three color components of each pixel point in the image;
step 13, selecting the first frame image of the nail fold microcirculation video image sequence as the reference frame image, performing offset correction on each frame later with reference to the first frame image, projecting the images in the row direction and the column direction respectively by using a formula (1.2) to obtain two independent one-dimensional data,
wherein the content of the first and second substances,
respectively represent the k frame image g k (i, j) the ith row and the jth column of the image, wherein M and N are the height and width of the image respectively, M and N respectively represent the length of a projection interval, M is more than or equal to 1 and less than or equal to M, and N is more than or equal to 1 and less than or equal to N;
step 14, calculating the correlation ρ (ω) between the line gray level projection curve of the k frame image and the line projection curve of the reference frame image according to the formula (1.3):
wherein, P k (i) Is the gray projection value of the ith line of the k frame image, P 1 (i) The gray projection value of the ith line of the first frame image is obtained, q is the search range of the offset on one side relative to the reference frame, M-2q represents the height of the image after the search ranges on two sides are removed, and omega is the search offset;
step 15, assume ω min The value of omega when the rho (omega) is minimum, then the motion vector delta of the k frame image relative to the reference frame image in the vertical direction is calculated according to the formula (1.4) k ，
δ k ＝m+1-ω min (1.4)
Wherein, delta k Indicating for positive a movement of | δ of the frame to be matched upwards k L pixels, δ k Negative indicates a downward movement of | δ k L pixels;
step 16, calculating the motion vector Δ of the current frame image in the horizontal direction by the same method as step 15 k For the horizontal direction shift-delta of the current frame image k Is translated by-delta in the vertical direction k Obtaining a stable video image sequence V;
in step 2, the process of background modeling the video image sequence V by using the gaussian mixture model includes the following steps:
step 21, selecting a first frame image from the video image sequence V to initialize a background model, wherein the expectation of the gaussian function is initialized toThe gray value of the image pixel, the standard deviation is initialized to be constant c, and the variance is c 2 ；
Step 22, background modeling is carried out on the video image sequence V by using a formula (2.1),
where g (x, y, t) is the gray value at the image coordinate (x, y) at time t, P (g (x, y, t)) is the probability of this gray value occurring, μ t And σ t Respectively the expectation and variance of the Gaussian distribution of the pixel at the time t;
step 23, detecting background pixels according to the formula (2.2),
|g(x,y,t)-μ t-1 (x,y)|<λσ t-1 (2.2)
wherein λ is a gaussian background parameter;
step 24, respectively updating the expectation, the standard deviation and the variance of the Gaussian model by using a formula (2.3), a formula (2.4) and a formula (2.5);
μ t (x,y)＝(1-α)*μ t-1 (x,y)+α*g(x,y,t) (2.3)
step 25, repeating the step 23 and the step 24 until the background pixel detection is finished, and obtaining a background model GM;
in step 3, the process of segmenting the blood vessel in each frame of image to obtain the contour and centerline of the blood vessel includes the following steps:
step 31, selecting a proper threshold value Th, binarizing the gray value GM (x, y) of any point in the scene model GM image according to a formula (3.1) to obtain a binary image B (x, y), and obtaining a blood vessel region set H by taking a white connected region, wherein the blood vessel region set H comprises a plurality of segmented blood vessels H;
step 32, selecting a target blood vessel H from the blood vessel set H, and extracting a skeleton by using a mathematical morphology skeleton extraction method to obtain a centerline E of the blood vessel H;
in step 4, the process of detecting and tracking the moving target in the blood vessel by adopting a background difference method and calculating to obtain the position and the motion trail diagram of the moving target comprises the following steps:
step 41, differentiating the current frame image f (x, y) and the background image GM (x, y) by using a formula (4.1) to obtain a moving target image D (x, y);
D(x,y)＝|f(x,y)-GM(x,y)|(4.1)
step 42, setting a threshold value T, binarizing the moving target image, if D (x, y) > T, then D (x, y) =255, otherwise D (x, y) =0, and obtaining a white pixel area in the new binary image D (x, y) as the moving target;
step 43, performing and operation on the central line obtained in step 32 and the moving target D (x, y) to obtain an intersection point set HJ, and calculating the distance from each point in the HJ to one end of the central line to obtain a distance set Hd;
step 44, generating a motion trail graph W according to the distance set Hd of each frame, setting the initial pixel value of W to be 0, the abscissa represents the number of video frames, the ordinate represents the distance from each point on the central line to the central line end point, and for the ith frame image, the distance set is Hd i Then to Hd i Let W (i, a) =255, resulting motion trajectory graph;
in step 5, projecting the motion trail diagram of the moving object, and calculating the speed of the moving object includes the following steps:
step 51, projecting the motion trail graph W in the S direction by using a formula (5.1), and rotating the value of a projection angle theta from 0 to pi to obtain a projection set Hw;
where M and N are the height and width of the image, respectively, and the operator is nucleotide ┚ to denote rounding-down;
step 52, calculating the variance of each histogram in the set Hw and putting the variance into the set H sigma;
step 53, selecting the largest element from the variance set H σ, and finding the corresponding projection histogram from the set Hw according to the element, thereby finding the projection angle θ of the projection histogram;
at step 54, the complementary cut of the angle θ is calculated to obtain the blood flow velocity v, i.e., v = cot θ.
2. A system for detecting the velocity of human microcirculation blood flow, comprising:
the video image sequence acquisition module is used for carrying out debouncing operation on the nail fold microcirculation video image to obtain a stable video image sequence V;
the background modeling module is used for carrying out background modeling on the video image sequence V by adopting a Gaussian mixture model;
the image blood vessel segmentation module is used for segmenting blood vessels in each frame of image to obtain the outlines and the central lines of the blood vessels;
the motion trail module is used for detecting and tracking a motion target in a blood vessel by adopting a background difference method, calculating to obtain the position of the motion target and generating a motion trail map according to the position;
the projection module is used for projecting a motion trail diagram of the moving target and calculating to obtain the speed of the moving target;
the video image sequence acquisition module comprises:
the video image file module is used for acquiring a video image file of the microcirculation of the nameless nail folds through the microscopic camera;
the image graying module is used for graying each frame image G (x, y) of the nail fold microcirculation video to obtain an image G (x, y);
the image projection module is used for selecting a first frame image of the nail fold microcirculation video image sequence as a reference frame image, performing offset correction on each frame later by referring to the first frame image, and projecting the images in the row direction and the column direction respectively to obtain two independent one-dimensional data;
a correlation calculation module for calculating the correlation ρ (ω) between the line gray level projection curve of the kth frame image and the line projection curve of the reference frame image;
a motion vector calculation module for calculating a motion vector delta of the k frame image in a vertical direction with respect to the reference frame image k ；
An image translation module for calculating the motion vector delta of the current frame image in the horizontal direction k And horizontally shifting the current frame image by-delta k Translating by-delta in the vertical direction k Obtaining a stable video image sequence V;
the background modeling module includes:
a background model initialization module for selecting a first frame image from the video image sequence V to initialize a background model, wherein the expectation initialization of the Gaussian function is the gray value of the image pixel, the standard deviation is initialized to a constant c, and the variance is c 2 ；
The modeling module is used for carrying out background modeling on the video image sequence V;
the background pixel detection module is used for detecting background pixels;
the updating module is used for updating the expectation, the standard deviation and the variance of the Gaussian model;
the image vessel segmentation module comprises:
the background model binarization module is used for selecting a proper threshold Th, binarizing gray values GM (x, y) of any point in the scene model GM image to obtain a binary image B (x, y), and taking a white connected region to obtain a blood vessel region set H which comprises a plurality of segmented blood vessels H;
the skeleton extraction module is used for selecting a target blood vessel H from the blood vessel set H, refining the target blood vessel H by using a mathematical morphology skeleton extraction method, and extracting a skeleton to obtain a central line E of the blood vessel H;
the motion trail module comprises:
the difference module is used for carrying out difference on the current frame image f (x, y) and the background image GM (x, y) to obtain a moving target image D (x, y);
a moving target binarization module, configured to set a threshold T, binarize a moving target image, if D (x, y) > T, then D (x, y) =255, otherwise, D (x, y) =0, and a white pixel area in the new binary image D (x, y) is the moving target;
the distance collection module is used for performing and operation on the central line and the moving target D (x, y) to obtain an intersection point collection HJ, and calculating the distance from each point in the HJ to one end of the central line to obtain a distance collection Hd;
a motion trail graph generating module for generating a motion trail graph W according to the distance set Hd of each frame, setting the initial pixel values of the motion trail graph W to be 0, the abscissa represents the number of video frames, the ordinate represents the distance from each point on the central line to the central line end point, and for the ith frame image, the distance set is Hd i Then to Hd i Let W (i, a) =255 for each element a in (a), resulting in a motion trajectory graph;
the projection module includes:
the S direction projection module is used for projecting the motion trail diagram W in the S direction, and the value of the projection angle theta is rotated to pi from 0 to obtain a projection set Hw;
the variance calculation module is used for calculating the variance of each histogram in the set Hw and putting the variance into the set H sigma;
the projection angle module is used for selecting the largest element from the variance set H sigma and finding the corresponding projection histogram from the set Hw according to the element so as to find the projection angle theta of the projection histogram;
and the angle cotangent calculation module is used for calculating the cotangent of the angle theta to obtain the blood flow velocity v, namely v = cot theta.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910288803.4A CN110060275B (en) | 2019-04-11 | 2019-04-11 | Method and system for detecting human body microcirculation blood flow velocity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910288803.4A CN110060275B (en) | 2019-04-11 | 2019-04-11 | Method and system for detecting human body microcirculation blood flow velocity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110060275A CN110060275A (en) | 2019-07-26 |
| CN110060275B true CN110060275B (en) | 2022-12-20 |
Family
ID=67318618
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910288803.4A Active CN110060275B (en) | 2019-04-11 | 2019-04-11 | Method and system for detecting human body microcirculation blood flow velocity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110060275B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110833393A (en) * | 2019-10-18 | 2020-02-25 | 广东药科大学 | Intelligent microcirculation detector and nail fold microcirculation detection method |
| CN112419329A (en) * | 2020-06-03 | 2021-02-26 | 中煤华晋集团有限公司王家岭矿 | Bulk similarity simulation top coal migration monitoring method based on MATLAB |
| CN111956200B (en) * | 2020-09-01 | 2023-07-21 | 徐州利华电子科技发展有限公司 | Microcirculation high-speed blood flow measuring and calculating system and measuring and calculating method |
| CN112164046B (en) * | 2020-09-24 | 2022-12-16 | 中山大学中山眼科中心 | Automatic analysis method for ocular microvascular hemodynamic parameters |
| CN113450329B (en) * | 2021-06-29 | 2023-01-10 | 广州医软智能科技有限公司 | Microcirculation image blood vessel branch erythrocyte flow rate calculation method and system |
| CN114219780A (en) * | 2021-12-10 | 2022-03-22 | 东北大学 | Deep learning-based method for detecting and segmenting nail fold microcirculation blood vessel |
| CN114882065B (en) * | 2022-07-12 | 2023-03-14 | 深圳市瑞图生物技术有限公司 | Method and device for judging fluidity of detection object, analyzer and storage medium |
| CN115998275B (en) * | 2022-12-26 | 2023-10-24 | 广东省新黄埔中医药联合创新研究院 | Blood flow velocity detection calibration method, device, equipment and readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102346911A (en) * | 2010-07-28 | 2012-02-08 | 北京集翔多维信息技术有限公司 | Method for segmenting blood vessel in digital subtraction angiography (DSA) image sequence |
| CN106846359A (en) * | 2017-01-17 | 2017-06-13 | 湖南优象科技有限公司 | Moving target method for quick based on video sequence |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7664329B2 (en) * | 2006-03-02 | 2010-02-16 | Honeywell International Inc. | Block-based Gaussian mixture model video motion detection |
| US8805051B2 (en) * | 2009-04-07 | 2014-08-12 | Virginia Commonwealth University | Image processing and machine learning for diagnostic analysis of microcirculation |
| CN104183127B (en) * | 2013-05-21 | 2017-02-22 | 北大方正集团有限公司 | Traffic surveillance video detection method and device |
-
2019
- 2019-04-11 CN CN201910288803.4A patent/CN110060275B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102346911A (en) * | 2010-07-28 | 2012-02-08 | 北京集翔多维信息技术有限公司 | Method for segmenting blood vessel in digital subtraction angiography (DSA) image sequence |
| CN106846359A (en) * | 2017-01-17 | 2017-06-13 | 湖南优象科技有限公司 | Moving target method for quick based on video sequence |
Non-Patent Citations (1)
| Title |
|---|
| 一种基于ST图轨迹方向的微循环流速测量算法;吕菲等;《生物医学工程学杂志》;20150425(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110060275A (en) | 2019-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110060275B (en) | Method and system for detecting human body microcirculation blood flow velocity | |
| CN105488815B (en) | A kind of real-time objects tracking for supporting target size to change | |
| Mukherjee et al. | Level set analysis for leukocyte detection and tracking | |
| Li et al. | Robust visual tracking based on convolutional features with illumination and occlusion handing | |
| CN109949341B (en) | Pedestrian target tracking method based on human skeleton structural features | |
| CN103810491B (en) | Head posture estimation interest point detection method fusing depth and gray scale image characteristic points | |
| CN106875425A (en) | A kind of multi-target tracking system and implementation method based on deep learning | |
| CN109767454B (en) | Unmanned aerial vehicle aerial video moving target detection method based on time-space-frequency significance | |
| CN102222341B (en) | Motion characteristic point detection method and device, moving target detecting method and device | |
| CN109816673A (en) | A kind of non-maxima suppression, dynamic threshold calculates and method for detecting image edge | |
| CN108875586B (en) | Functional limb rehabilitation training detection method based on depth image and skeleton data multi-feature fusion | |
| CN113744337B (en) | Synchronous positioning and mapping method integrating vision, IMU and sonar | |
| CN105279769B (en) | A kind of level particle filter tracking method for combining multiple features | |
| CN110930411B (en) | Human body segmentation method and system based on depth camera | |
| CN105787943B (en) | SAR image registration method based on multi-scale image block feature and rarefaction representation | |
| CN110555868A (en) | method for detecting small moving target under complex ground background | |
| CN109708658B (en) | Visual odometer method based on convolutional neural network | |
| CN113743385A (en) | Unmanned ship water surface target detection method and device and unmanned ship | |
| CN111709982B (en) | Three-dimensional reconstruction method for dynamic environment | |
| Zhong et al. | Seeing through the occluders: Robust monocular 6-DOF object pose tracking via model-guided video object segmentation | |
| CN112489088A (en) | Twin network visual tracking method based on memory unit | |
| CN115760893A (en) | Single droplet particle size and speed measuring method based on nuclear correlation filtering algorithm | |
| Li et al. | Moving object detection in dynamic scenes based on optical flow and superpixels | |
| CN103093481B (en) | A kind of based on moving target detecting method under the static background of watershed segmentation | |
| CN112435345B (en) | Human body three-dimensional measurement method and system based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |