CN116491894B - Parkinson's disease identification method based on Euler image amplification algorithm - Google Patents
Parkinson's disease identification method based on Euler image amplification algorithm Download PDFInfo
- Publication number
- CN116491894B CN116491894B CN202211396417.5A CN202211396417A CN116491894B CN 116491894 B CN116491894 B CN 116491894B CN 202211396417 A CN202211396417 A CN 202211396417A CN 116491894 B CN116491894 B CN 116491894B
- Authority
- CN
- China
- Prior art keywords
- video
- function
- pyramid
- frequency
- signal
- 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
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000003321 amplification Effects 0.000 title claims abstract description 16
- 238000003199 nucleic acid amplification method Methods 0.000 title claims abstract description 16
- 208000018737 Parkinson disease Diseases 0.000 title claims description 9
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 23
- 208000027089 Parkinsonian disease Diseases 0.000 claims abstract description 19
- 206010034010 Parkinsonism Diseases 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims abstract description 8
- 238000012795 verification Methods 0.000 claims abstract description 7
- 230000002452 interceptive effect Effects 0.000 claims abstract description 6
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 19
- 238000005070 sampling Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 12
- 238000001228 spectrum Methods 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 9
- 239000013598 vector Substances 0.000 claims description 9
- 101100328518 Caenorhabditis elegans cnt-1 gene Proteins 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 4
- 230000001131 transforming effect Effects 0.000 claims description 4
- 238000000540 analysis of variance Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 210000001061 forehead Anatomy 0.000 claims description 3
- 210000003128 head Anatomy 0.000 claims description 3
- 230000006798 recombination Effects 0.000 claims description 3
- 238000005215 recombination Methods 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 4
- 201000010099 disease Diseases 0.000 abstract description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 2
- 238000011156 evaluation Methods 0.000 abstract description 2
- 239000000284 extract Substances 0.000 abstract 1
- 238000004891 communication Methods 0.000 description 15
- 238000001514 detection method Methods 0.000 description 10
- 238000007726 management method Methods 0.000 description 8
- 238000003745 diagnosis Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000013523 data management Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 230000033764 rhythmic process Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000001175 cerebrospinal fluid Anatomy 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002565 electrocardiography Methods 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 238000011049 filling Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000011068 loading method Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 230000002747 voluntary effect Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/40—Detecting, measuring or recording for evaluating the nervous system
- A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
- A61B5/4082—Diagnosing or monitoring movement diseases, e.g. Parkinson, Huntington or Tourette
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0002—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0002—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network
- A61B5/0015—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network characterised by features of the telemetry system
- A61B5/0022—Monitoring a patient using a global network, e.g. telephone networks, internet
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0077—Devices for viewing the surface of the body, e.g. camera, magnifying lens
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/02405—Determining heart rate variability
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4842—Monitoring progression or stage of a disease
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration by non-spatial domain filtering
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Abstract
The invention relates to the technical field of intelligent medical treatment, in particular to a parkinsonism identification method based on an Euler image amplification algorithm. The technical scheme comprises the following steps: s1: collecting videos of specific areas of hands through a mobile phone; s2: constructing a video pyramid for the acquired video in the step S1; s3: observing a video pyramid in S2, selecting a time domain signal, selecting a proper filter, processing to obtain a signal, amplifying in different proportions, reconstructing a video, and storing; s4: and S3, after the reconstructed video is obtained, pulse wave extraction is carried out by adopting a pulse wave extraction program. The invention analyzes the hand shake degree based on Euler image amplification algorithm, utilizes the video pulse wave technology, combines digital signal processing, extracts pulse rate variability, distinguishes different characteristic ranges of patients and normal persons through wave decomposition, adopts training sets to carry out interactive verification of models, greatly improves the recognition accuracy, and is convenient for early warning or disease course evaluation of parkinsonism.
Description
Technical Field
The invention relates to the technical field of intelligent medical treatment, in particular to a parkinsonism identification method based on an Euler image amplification algorithm.
Background
The current non-contact pulse rate detection technology comprises the following steps: electrocardiography (ECG), craniocerebral CT examination, cerebrospinal fluid examination: the heart rhythm detailed information can be provided, the accuracy is high, but the special knowledge and technical support of doctors are needed, expensive special equipment is needed, doctors and other special staff are needed to check in hospitals or related medical places, the heart rhythm detailed information is not suitable to carry about, the waiting time of patients is long, the contact checking process possibly has pain, and the checking cost is also high; the intelligent bracelet and the watch can provide more accurate data information, but the intelligent bracelet and the watch are in contact measurement, physical contact with a tested person is needed, voluntary and active coordination of the tested person is needed, and accurate data can be obtained.
Disclosure of Invention
Aiming at the problems in the background technology, the invention provides a parkinsonism identification method which utilizes non-contact pulse rate detection to measure at any time and any place, realizes noninductive and more comfortable long-term monitoring and meets the monitoring requirements of special environments and special objects.
The technical scheme of the invention is as follows: the parkinsonism identification method based on the Euler image amplification algorithm comprises the following steps:
s1: collecting videos of specific areas of hands through a mobile phone;
s2: constructing a video pyramid for the acquired video in the step S1;
s3: observing a video pyramid in S2, selecting a time domain signal, selecting a proper filter, processing to obtain a signal, amplifying in different proportions, reconstructing a video, and storing;
s4: s3, after a reconstructed video is obtained, pulse wave extraction is carried out by adopting a pulse wave extraction program;
in S4, the specific operation of extracting pulse waves from the reconstructed video obtained in S3 is as follows:
s4a: importing an original signal; calling experimental data to draw an original signal x graph;
s4b: removing baseline drift; drawing an original signal x1 graph;
s4c: performing band-pass filtering; setting the Wp passband cutoff frequency and the Ws stopband cutoff frequency, carrying out frequency and radian conversion, carrying out calculation on the order and the 3dB cutoff frequency of the filter in the button function, and carrying out calculation on the molecular denominator polynomial of the filter system function in the button function to obtain a digital filter coefficient;
s4d: calling cheby to remove 50Hz power frequency interference; setting a Wp1 passband cutoff frequency value and a Ws1 stopband cutoff frequency value, setting an rp passband ripple value of 3dB and an rs stopband attenuation value of 60dB, determining the order and stopband cutoff frequency of the filter through a cheb2ord command, and then designing an n 1-order low-pass and bandpass filter with a cutoff frequency Wn1 through a cheby2 function command;
s4e: comparing the signals; calculating filtered waveforms x2 and s21 through a filtfilt function command, and rapidly drawing a signal graph s21 by using a sublot command and a plot command; rapidly drawing a signal graph x2 by using the subslot and plot commands and displaying an axis grid line;
s4f: acquiring a spectrogram; obtaining amplitude-frequency response of the filtered waveform by means of the fft function, then drawing s21 signals, hiding the grid lines of the axes, drawing absolute amplitude-frequency response, and obtaining an amplitude-frequency diagram;
s4g: extracting R waves; s21 is given to A, a max function and a min function are used for respectively obtaining a maximum value and a minimum value in a vector A to PM and MM, if judgment is carried out twice, and a peak value P1 and a peak value cnt1 in corresponding time are obtained; invoking a subslot command drawing screen to output data of a peak value P1 and an array type of the peak value number cnt1 in a corresponding time t1, and then invoking a stem function to draw a peak value P1 data sequence from an x axis to a data value in a stem form to obtain an R wave sequence;
s4h: displaying the characteristics of the peak value; calling a mean function to obtain a mean value P1, solving a variance vp of a peak value P1 through a var function, solving a standard deviation sp of the peak value P1 through a std function, and calling a disp function to output all the standard deviation sp in a command window;
s4i: obtaining a period T; defining an initialization variable j=1, obtaining the number of time T1 elements through a length function by means of a for cycle, then obtaining an electrocardiosignal period T in the for cycle through a set limit m, drawing a corresponding graph, and sequentially solving a period mean value mT, a period T variance vT and a period T standard deviation sT;
s4j: the power spectrum of the signal is estimated.
S5: decomposing the pulse wave extracted in the step S4 to obtain required characteristics;
s6: and (5) extracting pulse waves of a plurality of groups of normal persons and parkinsonism persons, and identifying and verifying according to the characteristics obtained in the step (S5).
Preferably, the specific operation of S1 is:
s1a: firstly, placing a mobile phone stably, adjusting the mobile phone to a position where the mobile phone is collected in a view-finding frame and is located in the center, and collecting videos with 20s resolution of 320 x 240 and 20FPS frame rate by means of mobile phone APP;
s1b: the video is loaded into a four-dimensional matrix and the function returns an np_zeros (frame_count, height, width, 3), dtype= 'float') size matrix.
Preferably, the specific operation of S1 is:
s1a: firstly, placing a mobile phone stably, adjusting the mobile phone to a position where the mobile phone is collected in a view-finding frame and is located in the center, and collecting videos with 20s resolution of 320 x 240 and 20FPS frame rate by means of mobile phone APP;
s1b: the video is loaded into a four-dimensional matrix and the function returns an np_zeros (frame_count, height, width, 3), dtype= 'float') size matrix.
Preferably, the specific operation of S2 is:
s2a: sampling each frame in the video, and constructing an image pyramid code as follows:
for sampling, call pyrUp (src [, dst [, dstsize [, borderType ] ]) function;
s2b: restoring the image in the image pyramid, the process is as follows:
firstly, defining and establishing a video pyramid function, establishing a video pyramid to receive video frames, and presenting the whole video in a group of images and sub-image sets with different resolutions, wherein the sub-image sets comprise four parameters, namely frame number, video frame height, video frame width and channel count in sequence; defining pyr0 and realMaxLevel to store the first layer video frame number, the level number and the statistic pyr0 length corresponding to the pyramid respectively, defining an array resultList, establishing a for loop for the highest level and the frame length respectively, in the first for loop, calling an np.zeros function to create a float type matrix of ([ len (frames) ]+list (pyr 0 i. Shape)) dimension, returning the result to the defined array, defining a statistic frame number pyramid in the second for loop, returning the highest level number to the defined array resultList and finally returning the whole pyramid frame number value;
then, defining a reconstructed video pyramid function, inputting a video pyramid reconstructed video, defining a highest level number maxLevel for storing a video pyramid length, storing the frame number, the video frame height, the video frame width and the channel count of a first layer of the video pyramid in defined variables fNumber, H, W and chNum respectively, calling an np.zeros function to create a (pyrVideo [0]. Shape) size and float type matrix, newly creating a for loop to calculate the video size, defining a variable framePyr for storing the highest level number obtained by newly creating a loop, then re-reconstructing a video image by the obtained value, storing the obtained result in a defined array video result, and finally returning the image.
Preferably, the specific operation of S3 is:
s3a: observing time domain signals, selecting a 1 st layer and a 4 th layer in the video pyramid, and selecting a point on the forehead of the figure head to observe;
s3b: the filter is selected, and the process of constructing the video pyramid filter is as follows:
firstly, creating an ideal band-pass filter function, applying an ideal band-pass filter to input data on a specified axis, then calculating fast transformation on the signal by using a fftpack. Fft (tensor, axis=axis) function, changing a time domain signal into a frequency domain signal when processing an actual signal by means of a fftpack. Fftfreq (tensor. Shape [0], d=1.0/fps), generating a sampling frequency, namely, automatically generating a frequency range, judging through an if loop and fast transforming, and finally returning the absolute value of the data after transforming;
secondly, creating an ideal filter function of the video pyramid, inputting an ideal band-pass filter on the video pyramid, defining an initialized video pyramid result array, creating a new for loop to calculate the length in the range of the video pyramid layer, calculating the pyramid filter in the for loop by using an ideal time filter through the video pyramid layer number and the cutting frequency, adding the obtained value into the video pyramid result array, and returning the value;
s3c: amplifying the signals by amplifying each layer of signals to different degrees, namely a certain proportion of filtered video and an original video signal;
s3d: the video is reconstructed as S2b.
Preferably, the specific operation of S5 is:
s5a: the spectrum range of the pulse wave original signal acquired by wave decomposition is 0-160 Hz, and high-frequency noise and low-frequency baseline drift are filtered after the primary wave decomposition; reconstructing high-frequency and low-frequency electrocardiosignal components to obtain a recombined component for removing high-frequency noise; baseline drift of the recombination component is removed, and a pulse wave signal with the frequency band range of 0-80 Hz is obtained; performing wave decomposition on the frequency band signal again, and performing 5-scale wave decomposition on the pulse signal to read a picture added with noise, wherein the data are data points with the size of height-width;
performing line decomposition on the data, writing [ C, L ] = wavedec (X, N, 'wavename'), performing wave decomposition circulation, and reconstructing an X-wave signal, namely frequency band distribution, through a waverec (C, L, 'name') function to obtain 6 spectral energy components in total;
s5b: feature acquisition, by the formulaObtaining wave entropy values, wherein i is a defined independent variable, n represents the total number of states, H (X) represents entropy, (Xi) represents the ith value of a signal, and p (Xi) represents the ith value probability of the signal;
respectively calculating wave entropy values of pulse waves of a sufficient number of patients and normal persons to obtain an average value and a variance of the pulse waves; and performing ANOVA variance analysis on the frequency domain characteristics of the obtained energy probability distribution and wave entropy values of the frequency bands, and selecting a 98% confidence interval to obtain the frequency bands and wave entropy values with obvious difference to patients and normal people.
Preferably, the specific operation of S6 is:
the pulse wave frequency domain features of a sufficient number of patients and normal persons form a training set, and Bayes linear discriminant analysis is firstly carried out to obtain a projection matrix W; and then adopting a training set to carry out interactive verification of the model, and obtaining the identification accuracy.
The parkinsonism identification platform based on the Euler image amplification algorithm comprises a remote monitoring center and a remote medical service site, wherein the remote medical service site carries out identification detection on a patient through the parkinsonism identification method based on the Euler image amplification algorithm;
the remote monitoring center takes a central server as a carrier and provides site management, equipment management, data management, financial management and the like through a communication server, and is responsible for data communication and unified management of the whole platform;
the remote medical service site takes a mobile phone as a carrier, is provided with a wireless communication network, supports instant communication, can send data to the service site after user detection is completed, obtains diagnosis and analysis at the first time, and can also directly send detection data to a remote monitoring center.
Preferably, the remote medical service site accesses a wireless public communication network in a TCP/IP protocol and is in networking communication with a remote monitoring center.
Compared with the prior art, the invention has the following beneficial technical effects:
the method has the advantages that the non-contact pulse rate detection is adopted, the mobile phone is used for shooting videos, the shaking degree of hands is analyzed based on the Euler image amplification algorithm, the video pulse wave technology is used, the digital signal processing is combined, the pulse rate variability (heart rate variability) is extracted, different characteristic ranges of a patient and a normal person are distinguished through wave decomposition, the training set is adopted for carrying out interactive verification of a model, the recognition accuracy is greatly improved, and the method is convenient to use for early warning or disease course evaluation of the Parkinson disease;
and the mobile phone is used as a carrier, detection service can be provided at any time based on wireless communication capability and convenient operation of the mobile phone, the mobile phone can be used for carrying out network communication with a remote monitoring center through wireless communication after video is recorded by the mobile phone and flow operation, diagnosis and analysis are obtained, and the remote monitoring center provides various services.
Drawings
FIG. 1 is a flowchart of a parkinsonism identification method based on an Euler image amplification algorithm in the present invention;
FIG. 2 is an image pyramid construction and algorithm of the present invention;
FIG. 3 is a reconstructed original image construction and algorithm of the present invention;
FIG. 4 is a pulse wave spectrum of the present invention;
fig. 5 is a frequency band signal re-wave exploded view of the present invention.
Detailed Description
The technical scheme of the invention is further described below with reference to the attached drawings and specific embodiments.
Examples
As shown in fig. 1-3, the parkinson's disease identification method based on the euler image amplification algorithm provided by the invention comprises the following steps:
s1: collecting videos of specific areas of hands through a mobile phone;
s1a: firstly, placing a mobile phone stably, adjusting the mobile phone to a position where the mobile phone is collected in a view-finding frame and is located in the center, and collecting videos with 20s resolution of 320 x 240 and 20FPS frame rate by means of mobile phone APP;
s1b: the video is loaded into a four-dimensional matrix and the function returns an np_zeros (frame_count, height, width, 3), dtype= 'float') size matrix.
S2: constructing a video pyramid for the acquired video in the step S1;
s2a: sampling each frame in the video, and constructing an image pyramid code as follows:
for sampling, call pyrUp (src [, dst [, dstsize [, borderType ] ]) function; wherein the parameter src represents the input image; dst represents the output image; dstsize denotes the Size of the output image, defaulting to Size (); borderType represents the pixel extrapolation method (in other words, how sampled is in particular, the effect is not great, since there is a partial loss in detail, the principle of detail loss: since the real edge is an extreme point). The steps are as follows: (1) reading an original image; (2) the image is up-sampled; (3) displaying an image
In the process of constructing the image pyramid, minutiae: the data type of the stored result is short (range: 32767-32767), the image stored data type is unsigned char (8 bits) (range: 0-255), and the storage of the image subtracted residual value (range: 255-255) is facilitated.
The process can be summarized as follows: firstly, checking the layer number of the pyramid, secondly, preparing a result, then, constructing the pyramid from low resolution, defining a for loop in the pyramid, firstly, up-sampling to twice the size of the pyramid from low resolution in the for loop, filling the value, and finally, putting the pyramid into the image pyramid.
S2b: restoring the image in the image pyramid, the process is as follows:
firstly, defining and establishing a video pyramid function, wherein the video pyramid function comprises two parameters, namely, parameter 1 (frames) represents a video frame, parameter 2 (maxLevel) represents the highest layer, then establishing a video pyramid to receive the video frame, and after receiving the video frame, presenting the whole video in a group of images and sub-image sets with different resolutions, so that each layer of images has a specific shape and size, and the four parameters comprise a frame number, a video frame height, a video frame width and a channel count in sequence. Defining pyr0 and realMaxLevel to store the first layer video frame number, progression and statistic pyr0 length corresponding to pyramid, defining array resultList, establishing for the highest progression and frame length, creating a flow type matrix of ([ len (frames) ]+list (pyr 0 i. Shape)) dimension ([ len (frames) ]+list (pyr 0 i. Shape)) in the first for loop, defining statistic frame number pyramid in the second for loop, returning to the highest progression, adding to the defined array resultList and finally returning to the whole pyramid frame number value.
Then, defining a reconstructed video pyramid function, inputting a video pyramid reconstructed video, defining a highest level number maxLevel for storing a video pyramid length, storing the frame number, the video frame height, the video frame width and the channel count of a first layer of the video pyramid in defined variables fNumber, H, W and chNum respectively, calling an np.zeros function to create a (pyrVideo [0]. Shape) size and float type matrix, newly creating a for loop to calculate the video size, defining a variable framePyr for storing the highest level number obtained by newly creating a loop, then re-reconstructing a video image by the obtained value, storing the obtained result in a defined array video result, and finally returning the image. In this process, the number of pyramid layers is adjusted according to the actual video loading and the actual application requirements.
S3: s3a: observing time domain signals, selecting a 1 st layer and a 4 th layer in the video pyramid, and selecting a point on the forehead of the figure head to observe;
s3b: the filter is selected, and the process of constructing the video pyramid filter is as follows:
firstly, creating an ideal band-pass filter function, and then applying an ideal band-pass filter to input data on a specified axis, wherein parameters comprise tensor (tensor), namely signal length; the parameter (low, height) represents the cut-off frequency of the band-pass filter; fps represents the sampling frequency; axis represents the axis of the input data array along which each subarray applies a linear filter;
then, the fast transform of the signal is calculated using the fftpack. Fft (tensor, axis=axis) function, and when the actual signal is processed, the time domain signal is changed into the frequency domain signal to generate the sampling frequency, namely, the automatic generation frequency range, and the data absolute value after the inverse transform is finally returned through if loop judgment and fast inverse transform by means of the fftpack. Fftfreq (tensor [0], d=1.0/fps) function.
Secondly, creating an ideal filter function of the video pyramid, and applying an ideal band-pass filter on the input video pyramid, wherein the parameter video pyr represents the video pyramid; (low, height) represents the cut frequency of the band-pass filter; fps means units (frames/sec); roi means that if specified, only the roi of the frame is filtered. Defining an initialized video pyramid result array, creating a for loop to calculate the length in the range of the video pyramid layer, then calculating a pyramid filter by using an ideal time filter in the for loop through the video pyramid layer number and the cutting frequency, and finally adding the obtained value into the video pyramid result array and returning the value.
S3c: amplifying the signals by amplifying each layer of signals to different degrees, namely a certain proportion of filtered video and the original video signals.
S3d: firstly, defining and establishing a video pyramid function, wherein the video pyramid function comprises two parameters, namely, parameter 1 (frames) represents video frames, parameter 2 (maxLevel) represents the highest layer, then, establishing a video pyramid to receive video frames, and after receiving the video frames, presenting the whole video in a group of images and sub-image sets with different resolutions, so that each layer of images has a specific shape and size, and the four parameters comprise frame number, video frame height, video frame width and channel count in sequence. Defining pyr0 and realMaxLevel to store the first layer video frame number, progression and statistic pyr0 length corresponding to pyramid, defining array resultList, establishing for the highest progression and frame length, creating a flow type matrix of ([ len (frames) ]+list (pyr 0 i. Shape)) dimension ([ len (frames) ]+list (pyr 0 i. Shape)) in the first for loop, defining statistic frame number pyramid in the second for loop, returning to the highest progression, adding to the defined array resultList and finally returning to the whole pyramid frame number value.
Then, defining a reconstructed video pyramid function, inputting a video pyramid reconstructed video, defining a highest level number maxLevel for storing a video pyramid length, storing the frame number, the video frame height, the video frame width and the channel count of a first layer of the video pyramid in defined variables fNumber, H, W and chNum respectively, calling an np.zeros function to create a (pyrVideo [0]. Shape) size and float type matrix, newly creating a for loop to calculate the video size, defining a variable framePyr for storing the highest level number obtained by newly creating a loop, then re-reconstructing a video image by the obtained value, storing the obtained result in a defined array video result, and finally returning the image.
S4: s3, after a reconstructed video is obtained, pulse wave extraction is carried out by adopting a pulse wave extraction program;
in S4, the specific operation of extracting pulse waves from the reconstructed video obtained in S3 is as follows:
s4a: importing an original signal; firstly, the platform clears the screen and closes all, and starts to import the original signal: calling load command to load the txt file under the experimental data folder in D disk, setting sampling frequency 1k, creating the first new data window by figure and rapidly drawing the original signal x graph by using subslot and plot (x), displaying the grid line by grid command, calling label creation function xlabel, ylabel, title to add coordinate axis names "time (ms)" and "amplitude" for x axis and y axis, and the icon is named "original signal".
S4b: removing baseline drift; firstly setting a k=.7 cut-off value, firstly sampling an analog signal at a rate of 0.3, then giving a filter boundary frequency fc to realize digital signal processing, then inputting a zeros (size (x)) function in a command line window to generate a size (x) all-zero array, writing for circulation, calculating a filtered waveform y1, then subtracting process values x and y to obtain x1, rapidly drawing an original signal x1 graph by using a sublot and a plot (x 1), opening a grid by using a grid command, calling a label creation function xlabel, ylabel, title to add coordinate axis names of time (ms) and amplitude for an x axis and a y axis, and further obtaining an icon title of removing a baseline drift signal.
S4c: performing band-pass filtering; calling figure command to create a second new data window, setting Wp passband cutoff frequency and Ws stopband cutoff frequency, carrying out frequency and radian conversion on the two, then carrying out order and 3dB cutoff frequency of a calculation filter in a button function, carrying out calculation of a filter system function numerator-denominator polynomial in the button function to obtain a digital filter coefficient, returning a frequency vector under the sampling frequency by virtue of a freqz command, displaying an axis grid line by virtue of the grid command, naming a title to carry out bandpass filtering, and hiding the axis grid line.
S4d: calling cheby to remove 50Hz power frequency interference; the figure command creates a third new data window, sets a Wp1 passband cutoff frequency value and a Ws1 stopband cutoff frequency value, sets an rp passband ripple value of 3dB and an rs stopband attenuation value of 60dB, determines the order and stopband cutoff frequency of the filter through a cheb2ord command, designs an n 1-order low-pass and bandpass filter of the cutoff frequency Wn1 through a cheby2 function command, returns a frequency vector under the sampling frequency through a freqz command, displays the axial grid line through the grid command, and hides the axial grid line after naming the filter through a title command.
S4e: comparing the signals; calling figure command to create a fourth new data window, calculating filtered waveforms x2 and s21 through filtfilt function command, rapidly drawing a signal graph s21 by using sublot and plot commands, displaying an axis grid line to obtain a filtered signal, calling a label creation function xlabel, ylabel, title to add coordinate axis names of time (ms) and amplitude for an x axis and a y axis, and setting an icon title as the filtered signal; then, the original signal pattern x1 is quickly drawn by using the sub-lot and plot commands and the axis grid lines are hidden, the label creation function xlabel, ylabel, title is called to add the coordinate axis names "time (ms)" and "amplitude" for the x-axis and the y-axis, the icon title "original signal" is set, and the signal pattern x2 is quickly drawn by using the sub-lot and plot commands and the axis grid lines are displayed again.
S4f: acquiring a spectrogram; calling figure command to create a fifth new data window, obtaining amplitude-frequency response of the filtered waveform by means of the fft function, then drawing s21 signal and hiding axis grid lines, drawing absolute amplitude-frequency response, calling label creation function xlabel, ylabel, title to add coordinate axis names of frequency/HZ and amplitude for x axis and y axis, setting icon title of amplitude-frequency diagram, obtaining amplitude-frequency diagram, calling xlim function command to limit x axis lower limit to 0 and x axis upper limit to 100.
S4g: extracting R waves; s21 is given to A, the maximum value in the vector A is given to PM and the minimum value in the vector A is given to MM by using a max function and a min function respectively, the fluctuation range of the peak value is set to be not more than 0.3 times of the maximum waveform height, if selection condition judgment is established, the two vectors are compared with each other in element by using another usage of the max function, a larger value between the first two elements is returned, in the process, the peak value is larger than all values in 200 points around the peak value because the period is larger than 600, so the if judgment is performed again, and the peak value P1 and the peak value number cnt1 in the corresponding time are obtained; and calling a figure command to create a sixth new data window, calling a subtelot command to draw the data of the peak P1 and the peak number cnt1 array types in the corresponding time t1, then calling a stem function to draw the peak P1 data sequence from the x axis to the data value in a stem form, finally ending with a circle, using a grid on command to open a grid, calling a label creation function xlabel to add the coordinate axis name peak number and the y axis name as the peak value for the x axis, and setting the icon title as an R wave sequence.
S4h: displaying the characteristics of the peak value; calling a mean function to obtain a P1 average value, and then calling a disp function to directly output a peak value mean value mp in a command window; solving the variance vp of the peak value P1 through the var function, and calling the disp function again to directly output the peak value variance vp in the command window; the standard deviation sp of the peak value P1 is calculated through the std function, and the disp function is called again to directly output the peak value standard deviation sp in the command window.
S4i: obtaining a period T; defining an initialization variable j=1, obtaining the number of time T1 elements through a length function by means of a for loop, then obtaining an electrocardiosignal period T in the for loop through a set limit m, and then calling a disp function to directly output the electrocardiosignal period T in a command window; using a sublot command to rapidly draw 2*1 subgraphs, drawing the subgraphs in a 2 nd subgraph, then calling a stem function to draw a period T of an electrocardiosignal from an x axis to a data value in a stem form, ending with a circle, using a grid on command to open a grid, calling a label creation function xlabel to add a coordinate axis name of 'heart rate' and a y axis name of 'period value' for the x axis, setting a chart title of 'period sequence', calling a mean function to obtain a period mean value mT, and calling a disp function to directly output the period mean value mT in a command window; solving a period T variance vT through a var function, and calling a disp function to directly output the period variance vT in a command window; and solving the period T standard deviation sT by using the std function, and calling the disp function to directly output the period standard deviation sT in the command window.
S4j: estimating a power spectrum of the signal; calling figure command to create seventh new data window, calling hanning weighted cosine window function, producing a hanning window w according to length 128, setting point nfft of fft as 1024, overlapping part novelap as 64 when segmentation average, then outputting frequency-power spectrum by means of a spectrum function, then drawing curve by taking f element as horizontal coordinate value and p element as vertical coordinate value through plot function, calling label creating function xlabel as x axis and adding coordinate axis name "frequency (Hz)" and y axis name "power spectrum (dB)", setting power spectrogram "of picture title" signal ", calling xlim function command to limit x axis lower limit as 0 and x axis upper limit as 50; and obtaining a maximum value z and a corresponding frequency f1 of the power spectrum in the vector P by using a max function, calling a disp function to directly output the maximum value z and the corresponding frequency f1 of the maximum value in a command window, and obtaining a pulse wave spectrogram through a series of transformations, wherein the pulse wave spectrogram is shown in fig. 4.
S5: decomposing the pulse wave extracted in the step S4 to obtain required characteristics;
s5a: filtering high-frequency noise and low-frequency baseline drift based on the primary wave decomposition within the frequency spectrum range of the acquired pulse wave original signal within 0-160 Hz, namely determining the number of the decomposition stages of the previously acquired electrocardiosignals according to the frequency of the preset electrocardiosignal sampling and baseline drift, dividing the electrocardiosignal data in the hands into a plurality of groups to obtain preset noise values corresponding to each group, and decomposing the corresponding groups according to the preset noise values corresponding to each group and the decomposition stages according to the decomposition stages of the groups to obtain all high-frequency and low-frequency electrocardiosignal components; expanding the low-frequency electrocardiosignal component to obtain baseline drift; reconstructing high-frequency and low-frequency electrocardiosignal components to obtain a recombined component for removing high-frequency noise; and removing baseline drift of the recombination component to obtain a pulse wave signal with the frequency band range of 0-80 Hz. And the band signal is subjected to wave decomposition again, the result is shown in fig. 5;
then, 5 scale wave decomposition is performed on the pulse signal, namely, a picture added with noise is read, and the data is the data point of the size of the height by the width. Firstly, carrying out line decomposition on data, and writing [ C, L ] = wavedec (X, N, ' wavename '), wherein X represents an original signal, N represents a decomposition level, wavename ' represents a base function, a left return value C represents each decomposed coefficient, L represents a corresponding detail coefficient signal, and how many times above wave decomposition circulation is needed, and reconstructing an X wave signal, namely, frequency band distribution through waverec (C, L, ' wavename ') functions to obtain 6 spectral energy components in total;
TABLE 1
S5b: feature acquisition, by the formulaObtaining wave entropy values, wherein i is a defined independent variable, n represents the total number of states, H (X) represents entropy, (Xi) represents the ith value of a signal, and p (Xi) represents the ith value probability of the signal;
wave entropy values are calculated for pulse waves of 20 parkinsonism persons and 20 normal persons respectively, and an average value and a variance of the wave entropy values are obtained;
| type(s) | Average value of | Variance of |
| Parkinson's disease person | 0.1249 | 0.05153 |
| Normal person | 0.1438 | 0.03818 |
Performing ANOVA variance analysis on the obtained energy probability distribution and wave entropy values of 6 frequency bands and 7 frequency domain features, wherein the results are shown in Table 2;
note that: f (2 29) isA significance coefficient of the distribution, wherein->Two degrees of freedom of the F distribution respectively,
TABLE 2
The confidence interval of 98% is selected, and 4 parameters of d4, d3, d2 and wave entropy can be obtained, which have obvious differences for parkinsonism and normal people. The other 3 parameters were rejected without significant differences.
S6: and (5) extracting pulse waves of a plurality of groups of normal persons and parkinsonism persons, and identifying and verifying according to the characteristics obtained in the step (S5).
The pulse wave frequency domain characteristics of 20 parkinsonism persons and 20 normal persons form a training set, and Bayes linear discriminant analysis is firstly carried out to obtain a projection matrix W; then, the training set is still adopted to carry out interactive verification of the model, and the obtained recognition accuracy is shown in a table 3;
| type(s) | Identifying the correct number | Accuracy of identification |
| Parkinson's disease person | 18 | 90％ |
| Normal person | 15 | 75％ |
TABLE 3 Table 3
It can be seen that parkinsonism and normal people can be better distinguished according to the interactive verification and identification of the pulse wave frequency domain characteristic parameters of parkinsonism and normal people.
The parkinsonism identification platform based on the Euler image amplification algorithm comprises a remote monitoring center and a remote medical service site, wherein the remote medical service site carries out identification detection through the parkinsonism identification method based on the Euler image amplification algorithm;
the remote monitoring center takes a central server as a carrier and provides site management, equipment management, data management, financial management and the like through a communication server, and is responsible for data communication and unified management of the whole platform;
the remote medical service site takes the mobile phone as a carrier, is provided with a wireless communication network, supports instant communication, can send data to the service site after the user detection is completed, obtains diagnosis and analysis at the first time, and can also directly send detection data to a remote monitoring center.
The remote medical service site accesses a wireless public communication network in a TCP/IP protocol and is in networking communication with a remote monitoring center.
The above-described embodiment is only one preferred embodiment of the present invention, and many alternative modifications and combinations of the above-described embodiments can be made by those skilled in the art based on the technical solutions of the present invention and the related teachings of the above-described embodiments.
Claims (4)
1. The parkinsonism identification method based on the Euler image amplification algorithm is characterized by comprising the following steps of: the method comprises the following steps:
s1: collecting videos of specific areas of hands through a mobile phone;
s2: constructing a video pyramid for the acquired video in the step S1;
s3: observing a video pyramid in S2, selecting a time domain signal, selecting a proper filter, processing to obtain a signal, amplifying in different proportions, reconstructing a video, and storing;
s4: s3, after a reconstructed video is obtained, pulse wave extraction is carried out by adopting a pulse wave extraction program;
in S4, the specific operation of extracting pulse waves from the reconstructed video obtained in S3 is as follows:
s4a: importing an original signal; calling experimental data to draw an original signal x graph;
s4b: removing baseline drift; drawing a graph for removing the baseline drift signal x 1;
s4c: performing band-pass filtering; setting the Wp passband cutoff frequency and the Ws stopband cutoff frequency, carrying out frequency and radian conversion, carrying out calculation on the order and the 3dB cutoff frequency of the filter in the button function, and carrying out calculation on the molecular denominator polynomial of the filter system function in the button function to obtain a digital filter coefficient;
s4d: calling cheby to remove 50Hz power frequency interference; setting a Wp1 passband cutoff frequency value and a Ws1 stopband cutoff frequency value, setting an rp passband ripple value of 3dB and an rs stopband attenuation value of 60dB, determining the order and stopband cutoff frequency of the filter through a cheb2ord command, and then designing an n 1-order low-pass and bandpass filter with a cutoff frequency Wn1 through a cheby2 function command;
s4e: comparing the signals; calculating filtered waveforms x2 and s21 through a filtfilt function command, and rapidly drawing a signal graph s21 by using a sublot command and a plot command, wherein s21 is a new signal processed based on the filtered waveform x 2; rapidly drawing a signal waveform x2 by using a subslot command and a plot command, displaying an axis grid line, wherein the waveform x2 is obtained by filtering a waveform x1 signal;
s4f: acquiring a spectrogram; obtaining amplitude-frequency response of the filtered waveform by means of the fft function, then drawing s21 signals, hiding the grid lines of the axes, drawing absolute amplitude-frequency response, and obtaining an amplitude-frequency diagram;
s4g: extracting R waves; s21 is given to an A vector, a max function and a min function are used for respectively obtaining a maximum value and a minimum value in the A vector to PM and MM, if judgment is carried out twice, and a peak value P1 and a peak value cnt1 in corresponding time are obtained; invoking a subslot command drawing screen to output data of a peak value P1 and an array type of the peak value number cnt1 in a corresponding time t1, and then invoking a stem function to draw a peak value P1 data sequence from an x axis to a data value in a stem form to obtain an R wave sequence;
s4h: displaying the characteristics of the peak value; calling a mean function to obtain a mean value P1, solving a variance vp of a peak value P1 through a var function, solving a standard deviation sp of the peak value P1 through a std function, and calling a disp function to output all the standard deviation sp in a command window;
s4i: obtaining a period T; defining an initialization variable j=1, obtaining the number of time T1 elements through a length function by means of a for cycle, then obtaining an electrocardiosignal period T in the for cycle through a set limit m, drawing a corresponding graph, and sequentially solving a period mean value mT, a period T variance vT and a period T standard deviation sT;
s4j: estimating the power spectrum of the signal to obtain a pulse wave spectrum;
s5: decomposing the pulse wave extracted in the step S4 to obtain required characteristics;
s5a: the spectrum range of the pulse wave original signal acquired by the wave decomposition is 0-160 Hz, and the high-frequency noise and the low-frequency baseline drift are filtered after the primary wave decomposition; reconstructing high-frequency and low-frequency electrocardiosignal components to obtain a recombined component for removing high-frequency noise; baseline drift of the recombination component is removed, and a pulse wave signal with the frequency band range of 0-80 Hz is obtained; performing wave decomposition on the frequency band signal again, and performing 5-scale wave decomposition on the pulse signal to read a picture added with noise, wherein the data are data points with the size of height-width;
performing line decomposition on the data, writing [ C, L ] = wavedec (X, N, 'wavename'), performing wave decomposition circulation, and reconstructing an X-wave signal, namely frequency band distribution, through a waverec (C, L, 'name') function to obtain 6 spectral energy components in total;
s5b: feature acquisition, by the formulaObtaining wave entropy values, wherein i is a defined independent variable, n represents the total number of states, H (X) represents entropy, xi represents the ith value of a signal, and p (Xi) represents the ith value probability of the signal;
respectively calculating wave entropy values of pulse waves of a sufficient number of patients and normal persons to obtain an average value and a variance of the pulse waves; performing ANOVA variance analysis on the frequency domain characteristics of the obtained energy probability distribution and wave entropy values of the frequency bands, and selecting a 98% confidence interval to obtain frequency bands and wave entropy values with obvious difference to patients and normal persons;
s6: pulse wave extraction is carried out on a plurality of groups of normal persons and parkinsonism persons, and identification and verification are carried out according to the characteristics obtained in the step S5;
the pulse wave frequency domain features of a sufficient number of patients and normal persons form a training set, and Bayes linear discriminant analysis is firstly carried out to obtain a projection matrix W; and then adopting a training set to carry out interactive verification of the model, and obtaining the recognition accuracy.
2. The parkinson's disease identification method based on the euler image amplification algorithm according to claim 1, wherein: the specific operation of S1 is as follows:
s1a: firstly, placing a mobile phone stably, adjusting an acquisition position to the center position of a mobile phone viewfinder, and acquiring a video with 20s resolution of 320 x 240 and 20FPS frame rate by means of mobile phone APP;
s1b: the video is loaded into a four-dimensional matrix and the function returns an np_zeros (frame_count, height, width, 3), dtype= 'float') size matrix.
3. The parkinson's disease identification method based on the euler image amplification algorithm according to claim 1, wherein: the specific operation of S2 is as follows:
s2a: sampling each frame in the video, and constructing an image pyramid code as follows:
for sampling, call pyrUp (src [, dst [, dstsize [, borderType ] ]) function;
s2b: restoring the image in the image pyramid, the process is as follows:
firstly, defining and establishing a video pyramid function, establishing a video pyramid to receive video frames, and presenting the whole video in a group of images and sub-image sets with different resolutions, wherein the sub-image sets comprise four parameters, namely frame number, video frame height, video frame width and channel count in sequence; defining pyr0 and realMaxLevel to store the first layer video frame number, the level number and the statistic pyr0 length corresponding to the pyramid respectively, defining an array resultList, establishing a for loop for the highest level and the frame length respectively, in the first for loop, calling an np.zeros function to create a float type matrix of ([ len (frames) ]+list (pyr 0 i. Shape)) dimension, returning the result to the defined array, defining a statistic frame number pyramid in the second for loop, returning the highest level number to the defined array resultList and finally returning the whole pyramid frame number value;
then, defining a reconstructed video pyramid function, inputting a video pyramid reconstructed video, defining a highest level number maxLevel for storing a video pyramid length, storing the frame number, the video frame height, the video frame width and the channel count of a first layer of the video pyramid in defined variables fNumber, H, W and chNum respectively, calling an np.zeros function to create a (pyrVideo [0]. Shape) size and float type matrix, newly creating a for loop to calculate the video size, defining a variable framePyr for storing the highest level number obtained by newly creating a loop, then re-reconstructing a video image by the obtained value, storing the obtained result in a defined array video result, and finally returning the image.
4. The parkinson's disease identification method based on the euler image magnification algorithm according to claim 3, wherein: the specific operation of S3 is as follows:
s3a: observing time domain signals, selecting a 1 st layer and a 4 th layer in the video pyramid, and selecting a point on the forehead of the figure head to observe;
s3b: the filter is selected, and the process of constructing the video pyramid filter is as follows:
firstly, creating an ideal band-pass filter function, applying an ideal band-pass filter to input data on a specified axis, then calculating fast transformation on the signal by using a fftpack. Fft (tensor, axis=axis) function, changing a time domain signal into a frequency domain signal when processing an actual signal by means of a fftpack. Fftfreq (tensor. Shape [0], d=1.0/fps), generating a sampling frequency, namely, automatically generating a frequency range, judging through an if loop and fast transforming, and finally returning the absolute value of the data after transforming;
secondly, creating an ideal filter function of the video pyramid, inputting an ideal band-pass filter on the video pyramid, defining an initialized video pyramid result array, creating a new for loop to calculate the length in the range of the video pyramid layer, calculating the pyramid filter in the for loop by using an ideal time filter through the video pyramid layer number and the cutting frequency, adding the obtained value into the video pyramid result array, and returning the value;
s3c: amplifying the signals by amplifying each layer of signals to different degrees, namely a certain proportion of filtered video and an original video signal;
s3d: the video is reconstructed as S2b.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211396417.5A CN116491894B (en) | 2022-11-09 | 2022-11-09 | Parkinson's disease identification method based on Euler image amplification algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211396417.5A CN116491894B (en) | 2022-11-09 | 2022-11-09 | Parkinson's disease identification method based on Euler image amplification algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116491894A CN116491894A (en) | 2023-07-28 |
| CN116491894B true CN116491894B (en) | 2024-02-27 |
Family
ID=87321938
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211396417.5A Active CN116491894B (en) | 2022-11-09 | 2022-11-09 | Parkinson's disease identification method based on Euler image amplification algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116491894B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018168909A1 (en) * | 2017-03-16 | 2018-09-20 | 株式会社カオテック研究所 | Parkinson's disease diagnosing device |
| CN110353646A (en) * | 2019-07-29 | 2019-10-22 | 苏州市高事达信息科技股份有限公司 | Contactless heart rate detection method |
| CN110916631A (en) * | 2019-12-13 | 2020-03-27 | 东南大学 | Student classroom learning state evaluation system based on wearable physiological signal monitoring |
| CN112949349A (en) * | 2019-12-09 | 2021-06-11 | 南宁莲现健康科技有限公司 | Method and system for displaying pulse condition waveform in real time based on face video |
| CN113378798A (en) * | 2021-07-16 | 2021-09-10 | 福建师范大学 | Pulse video pulse extraction method and system based on video micro-motion amplification |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108471961B (en) * | 2016-06-15 | 2021-02-09 | 深圳迈瑞生物医疗电子股份有限公司 | Physiological parameter calculation method and corresponding medical equipment |
-
2022
- 2022-11-09 CN CN202211396417.5A patent/CN116491894B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018168909A1 (en) * | 2017-03-16 | 2018-09-20 | 株式会社カオテック研究所 | Parkinson's disease diagnosing device |
| CN110353646A (en) * | 2019-07-29 | 2019-10-22 | 苏州市高事达信息科技股份有限公司 | Contactless heart rate detection method |
| CN112949349A (en) * | 2019-12-09 | 2021-06-11 | 南宁莲现健康科技有限公司 | Method and system for displaying pulse condition waveform in real time based on face video |
| CN110916631A (en) * | 2019-12-13 | 2020-03-27 | 东南大学 | Student classroom learning state evaluation system based on wearable physiological signal monitoring |
| CN113378798A (en) * | 2021-07-16 | 2021-09-10 | 福建师范大学 | Pulse video pulse extraction method and system based on video micro-motion amplification |
Non-Patent Citations (2)
| Title |
|---|
| 基于欧拉视频放大技术的视频图像微变化分析与研究;胡聘聘;中国优秀硕士学位论文全文数据库;20150415(第04期);全文 * |
| 基于腕部脉搏波的脉率变异性分析系统研究与设计;罗明;中国优秀硕士学位论文全文数据库;20200215(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116491894A (en) | 2023-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107529646B (en) | Non-contact heart rate measurement method and device based on Euler image amplification | |
| CN112120694B (en) | Motor imagery electroencephalogram signal classification method based on neural network | |
| KR20140139564A (en) | Systems and methods for ecg monitoring | |
| US11406304B2 (en) | Systems and methods for physiological sign analysis | |
| CN112070785A (en) | Medical image analysis method based on computer vision | |
| CN114241077B (en) | CT image resolution optimization method and device | |
| CN112587153A (en) | End-to-end non-contact atrial fibrillation automatic detection system and method based on vPPG signal | |
| CN105139443A (en) | Three-dimensional imaging system and method of diagnosis result | |
| CN105433931A (en) | Processing device and method for describing waveform by light volume change | |
| Saini et al. | Effective automated method for detection and suppression of muscle artefacts from single‐channel EEG signal | |
| CN114041786A (en) | Ballistocardiogram signal detection method, ballistocardiogram signal detection device and ballistocardiogram signal detection equipment | |
| CN111789629A (en) | Breath sound intelligent diagnosis and treatment system and method based on deep learning | |
| US20190313929A1 (en) | System and method for non-invasive extraction of fetal electrocardiogram signals | |
| Sujadevi et al. | Denoising of phonocardiogram signals using variational mode decomposition | |
| CN116491894B (en) | Parkinson's disease identification method based on Euler image amplification algorithm | |
| Moya-Albor et al. | Heart rate estimation using Hermite transform video magnification and deep learning | |
| Jain et al. | An adaptive method for shrinking of wavelet coefficients for phonocardiogram denoising | |
| CN110956192A (en) | Method and device for classifying non-reconstruction compressed sensing physiological data | |
| CN116687409B (en) | Emotion recognition method and system based on digital twin and deep learning | |
| CN116881762A (en) | Emotion recognition method based on dynamic brain network characteristics | |
| CN110537907A (en) | Electrocardiosignal compression and identification method based on singular value decomposition | |
| Unser | Wavelets, statistics, and biomedical applications | |
| CN106343989A (en) | Image processing-based blood pressure monitoring method | |
| Akhbardeh et al. | Evaluation of heart condition based on ballistocardiogram classification using compactly supported wavelet transforms and neural networks | |
| Christensen et al. | Breath sounds analysis for asthma monitoring: a method for automated detection of flow events from tracheal recordings |
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 |