CN114431849B - Aquatic animal heart rate detection method based on video image processing - Google Patents

Abstract

The invention discloses an aquatic animal heart rate detection method based on video image processing, which comprises the steps of acquiring an aquatic animal ventral video image sequence, selecting heart and background interested areas, keeping track, extracting an environmental noise component shared by the heart and the background interested areas by adopting a multi-dataset joint analysis method, and setting the environmental noise component to zero; then further eliminating static components in the model according to the bicolor reflection model, and selecting specular reflection components and diffuse reflection components; then, independent component analysis is carried out on a time signal sequence composed of specular reflection components and diffuse reflection components, and the independent component with the greatest heart rate information is obtained through correlation calculation; finally, carrying out frequency domain decomposition on the aquatic animal, screening out the subcomponent with the highest matching degree on the frequency domain according to the specific heart rate range of the aquatic animal, and outputting a cardiac waveform chart to output or calculating the heart rate value; the scheme provides a robust and accurate method for non-contact detection of the heart rate of the aquatic animals, and has wide application prospects in the aspects of aquaculture monitoring and biological stress resistance evaluation.

Description

A heart rate detection method for aquatic animals based on video image processing

technical field

The invention relates to the technical field of biological image information, in particular to a method for detecting the heart rate of aquatic animals based on video image processing.

Background technique

Aquaculture is an important part of my country's fishery. However, under the conditions of climate change and unreasonable farming methods, there will be problems such as disease outbreaks and high-temperature deaths. Breeding aquatic animals with stronger stress resistance or tolerance through selective breeding is one of the most effective countermeasures to deal with this problem.

As an important physiological indicator, heart rate is closely related to the metabolic state of aquatic animals and the stress to environmental changes. Therefore, technicians usually use heart rate as an evaluation index to evaluate the ability of aquatic animals to cope with internal or external environmental changes, and to assist their genetic breeding work. At present, the heart rate measurement of aquatic animals has certain limitations, mainly manifested in the need to restrict biological activities, continuous contact, and even surgical operations. Implantable electrode methods, Doppler ultrasound methods, etc. Heart rate is measured. Conventional remote non-contact heart rate measurement methods often lack high-precision signal denoising methods and signal separation methods. Due to the special physiological structure of aquatic animals, most of them do not have rich capillaries distributed under the superficial epidermis, resulting in very weak physiological information contained in video images, often requiring more accurate and robust technical means to extract heart rate information. Therefore, in the existing breeding research, there is a lack of methods to conveniently observe the individual heart rate of aquatic animals under non-contact conditions, so it is imminent to provide an accurate and robust method to detect the heart rate of aquatic animals.

Contents of the invention

In view of this, the purpose of the present invention is to propose a heart rate detection method for aquatic animals based on video image processing, in order to broaden the range of stress resistance evaluation indicators for genetic breeding researchers, thereby increasing the evaluation dimension in the genetic breeding process.

In order to realize above-mentioned technical purpose, the technical scheme that the present invention adopts is:

A kind of aquatic animal heart rate detection method based on video image processing, it comprises:

1) Obtain t frames of video images of the ventral surface of aquatic animals, then select a region of interest according to preset conditions, track and maintain it, and obtain a time signal sequence of the heart and the background region, wherein the region of interest includes the heart region of interest and background regions of interest;

2) According to the joint analysis method of multiple data sets, the common environmental noise shared by the heart region of interest and the background region of interest is screened out and removed;

3) According to the two-color reflectance model, the time signal sequence of the heart region of interest after removing the common environmental noise in step 2) is regarded as a linear combination of illumination change components, specular reflection components, and diffuse reflection components, and the static components are eliminated and Project to an orthogonal plane to remove the illumination variation component;

4) Perform independent component analysis on the time signal sequence of the heart region of interest that only contains specular reflection components and diffuse reflection components obtained through step 3), and filter out the strongest correlation with the green channel in the original heart region of interest signal sequence independent components of

5) decompose the independent components obtained in step 4) from the frequency domain, and screen out the subcomponents with the highest matching degree with the unique heart rate distribution range of aquatic animal species in the frequency domain;

6) Process the obtained subcomponents according to step 5) to further output cardiac waveform or heart rate value.

As a possible implementation, further, in step 1), the video image is a digital image illustrating colors in different color spaces; the aquatic animal has a heart organ and its blood tissue components contain red blood cells aquatic.

As a possible implementation, further, in step 1), the heart region of interest and the background region of interest are selected by manual selection or target detection algorithm and are tracked and maintained by the target detection algorithm; the obtained heart and The time signal sequence in the background area is the time signal sequence obtained after eliminating the quantization noise by spatial averaging.

As a possible implementation, further, in step 2), the multi-dataset joint analysis method is a joint blind source separation method based on multiple data sets.

As a possible implementation manner, further, in step 3), the static component is eliminated by performing time-domain normalization processing on the signal.

As a possible implementation, further, in step 5), the independent components obtained in step 4) are decomposed in the frequency domain according to the ensemble empirical mode decomposition method.

As a possible implementation, further, in step 6), according to the subcomponents obtained in step 5), the heart rate value corresponding to the signal is calculated by the frequency domain or time domain method, and then the heart rate value is output or converted into a cardiac waveform output.

Based on the above solution, the present invention also provides a computer-readable storage medium, wherein at least one instruction, at least one section of program, code set or instruction set is stored in said storage medium, and said at least one instruction, at least one section of program, The code set or instruction set is loaded by the processor and executed to realize the above-mentioned method for detecting heart rate of aquatic animals based on video image processing.

Adopt above-mentioned technical scheme, the present invention compares with prior art, and the beneficial effect that it has is:

1. The scheme of the present invention utilizes the background content that was originally redundant information: for the removal of background noise, no traditional and general-purpose processing method is used, but a background region of interest is added, and the heart is separated by joint blind source separation. The source signal component vector shared by the region of interest and the background region of interest is extracted and regarded as background noise removal. The advantage is that it can perform adaptive denoising according to the background part of different video images.

2. The scheme of the present invention uses a two-color reflection model to separate the signal containing heart rate information: according to the two-color reflection model, the signal is regarded as composed of three parts: the illumination change component, the specular reflection component and the diffuse reflection component. The distance between the light source, the subject and the camera changes, the specular reflection component is the specular reflection component produced by the direct reflection of the skin surface, and the diffuse reflection component is the diffuse reflection component after being transmitted through the skin surface and reabsorbed by the subcutaneous tissue and blood. Its advantage is that it explains the formation source of heart rate information and the way it is combined with other components from the physiological and physical level, which can provide more accurate and robust extraction of components containing heart rate information.

3. The solution of the present invention fully considers the physiological structure of aquatic animals when further screening the components containing heart rate information, and uses the green channel signal of the original signal as the screening standard. Hemoglobin is a component of the blood of aquatic animals. Hemoglobin is colored and has an absorption spectrum in the visible light region. The maximum absorption value of hemoglobin in aquatic animals is between 540 and 575 nanometers for the oxygen (Oxy) type, and between 538 and 568 nanometers for the carbonyl (carbonyl) type, and there is no significant difference compared with mammals. Therefore, using the green channel signal as a standard can be a more theoretically supported and more accurate screening method.

4. The scheme of the present invention adopts the ensemble empirical mode decomposition method to further separate the independent components after independent component analysis and screening in the frequency domain. After ensemble empirical mode decomposition, the selected independent components are divided into eigenmode functions in multiple frequency bands. According to the theoretical heart rate distribution range unique to aquatic animal species, the eigenmodes containing the most expected frequencies can be selected. The function performs cardiac waveform output or heart rate calculation. Its advantage is that it further improves the signal-to-noise ratio in the frequency domain, eliminates component interference other than the desired frequency, and improves the reliability of the calculation results.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is the schematic flow chart of the non-contact aquatic animal heart rate detection method based on video image processing in the present invention;

Fig. 2 is a schematic diagram of heart region of interest and background region of interest selection involved in the embodiment of the present invention;

FIG. 3 is a signal sequence diagram of a heart region of interest involved in an embodiment of the present invention;

FIG. 4 is a signal sequence diagram of a background region of interest involved in an embodiment of the present invention;

5 is a vector diagram of source signal components after joint blind source separation involved in an embodiment of the present invention;

FIG. 6 is a time signal sequence diagram of a cardiac region of interest reconstructed after removal of environmental noise involved in an embodiment of the present invention;

Fig. 7 is a time signal sequence diagram of a cardiac region of interest that only includes specular reflection components and diffuse reflection components after projection eliminates time-varying illumination variation components involved in an embodiment of the present invention;

Figure 8 is an independent component diagram after independent component analysis involved in the embodiment of the present invention;

FIG. 9 is a diagram of the eigenmode functions of each frequency band after the ensemble empirical mode decomposition method involved in the embodiment of the present invention;

Fig. 10 is a cardiac waveform diagram involved in the embodiment of the present invention;

FIG. 11 is a fast Fourier transform spectrum diagram involved in an embodiment of the present invention;

Fig. 12 is a peak and trough diagram in the peak detection method involved in the embodiment of the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments. In particular, the following examples are only used to illustrate the present invention, but not to limit the scope of the present invention. Likewise, the following embodiments are only some but not all embodiments of the present invention, and all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

In this embodiment, after the large yellow croaker is anesthetized, an implantable electrocardiographic device is connected around the pericardium through a surgical operation. While the electrocardiographic device is collecting data, a video of the ventral surface of the large yellow croaker is captured by a camera to obtain a video image sequence.

As shown in Figure 1, this solution provides a method for detecting the heart rate of aquatic animals based on video image processing. The general technical idea is: to obtain t frames of video images of aquatic animals, and then select the heart region of interest and the background region of interest respectively. After obtaining the signal sequence of the average pixel intensity of the two over time, the joint blind source separation is used to extract the environmental noise component shared by the two and set to zero to obtain the time signal sequence of the heart region of interest after the environmental noise is removed; then according to the two-color The reflectance model regards the new time signal sequence of the cardiac region of interest as a linear combination of illumination change components, specular reflection components, and diffuse reflection components, and performs temporal normalization on the time signal sequence to eliminate the static components, and then normalizes the time signal sequence The normalized time signal sequence is projected onto a plane orthogonal to the illumination change component to screen out the specular reflection component and the diffuse reflection component; then the independent component analysis is performed on the time signal sequence composed of the specular reflection component and the diffuse reflection component, and each The correlation between the independent components and the green channel signal in the time signal sequence of the original heart region of interest, the independent component containing the most diffuse reflection components is obtained; finally, the ensemble empirical mode decomposition is performed on the independent components containing the most diffuse reflection components, according to the aquatic animal species The unique heart rate range screens out the eigenmode function with the highest matching degree in the frequency domain, and the sliding average of the eigenmode function in the time domain can be output as a heartbeat waveform or further calculate the heart rate value.

Specifically, this program includes the following implementation steps:

1) Obtain t-frame video images of the ventral surface of the large yellow croaker, convert them into RGB video images, and combine them with those shown in Figure 2. In the first frame of video images, the heart in the video images can be manually selected. region and background region of interest, and then use the target tracking algorithm to complete the tracking and maintenance of the region of interest in the subsequent t-1 frame video images, so as to complete the selection of the heart region of interest and the background region of interest and complete it by the target detection algorithm The work of tracking and maintaining the region of interest; then, for each region of interest, calculate the average pixel intensity of the three color channels of each frame of RGB video image, and obtain the cardiac region of interest signal sequence shown in Figure 3 and Figure 4 The signal sequence of the background region of interest shown, the formula of the heart region of interest signal sequence is as follows:

X _hr (t) = [R _hr (t); G _hr (t); B _hr (t)] ^T ,

Wherein, R _hr (t), G _hr (t), B _hr (t) are respectively the average pixel intensity of the R, G, and B color channels of the heart region of interest video image;

The formula of the background region of interest signal sequence is as follows:

X _bg (t) = [R _bg (t); G _bg (t); B _bg (t)] ^T ;

Wherein, R _bg (t), G _bg (t), B _bg (t) are respectively the average pixel intensity of the R, G, and B color channels of the background region of interest video image;

2) According to the joint blind source separation method, using the mathematical formula SCV(t)=W*X(t), the heart region of interest signal sequence X _hr (t) and the background region of interest signal sequence X _bg (t ) to perform joint blind source separation processing to obtain an unmixing matrix W, where W is an unmixing matrix, and SCV (t) is a source signal component vector;

Further calculation can get the source signal component vector matrix shown in Figure 5

SCV(t)=[SCV _hr1 ; SCV _hr2 ; SCV _hr3 ; SCV _bg1 ; SCV _bg2 ; SCV _bg3 ] ^T ;

Wherein, the joint blind source separation method refers to a joint blind source separation framework realized based on Gaussian independent vector analysis;

Based on this, each source signal component vector SCV _hr(n) (t) of the heart region of interest signal sequence and each source signal component vector SCV _bg(n) of the background region of interest signal sequence can be calculated according to the source signal component vector matrix (t); Then each source signal component vector SCV _hr(n) (t) of the cardiac region of interest signal sequence obtained by calculation and each source signal component vector SCV _bg(n) (t) of the background region of interest signal sequence ), set the source signal component vector SCV _hr(m) (t) of the cardiac region of interest signal sequence corresponding to the maximum correlation coefficient to zero to obtain SCV _new (t), and then according to SCV(t)=W *X(t) can obtain the cardiac region of interest signal sequence and the background region of interest signal sequence X _new (t)=W\SCV _new (t) after removing the common environmental noise;

3) According to the two-color reflection model, the time signal sequence X _new (t)[1:3] of the heart region of interest obtained in step 2) after removing the environmental noise as shown in Figure 6 is regarded as the illumination change component, the mirror surface The linear combination of reflection component and diffuse reflection component, project X _new (t) normalized in the time domain to the plane P=[p ₁ ; p ₂ ] ^T orthogonal to the illumination change component, so as to obtain the Showing the cardiac region of interest time signal sequence X _rflct (t)=P*X _new (t) that only contains the specular reflection component and the diffuse reflection component;

4) Perform independent component analysis on the cardiac region of interest time signal sequence X _rflct (t) obtained in step 3) that only contains specular reflection components and diffuse reflection components, that is, X _rflct (t)=A*S(t) , where A is the mixing matrix, S(t) is the independent component of the source signal (as shown in Figure 8), and S(t)=

[s ₁ (t); s ₂ (t)] ^T , calculate the relationship between [s ₁ (t); s ₂ (t)] ^T and G _hr (t) in the signal sequence X _hr (t) of the original heart region of interest Correlation coefficient, select the independent component s(t) with the strongest correlation with the green channel signal of the original signal;

5) Use the ensemble empirical mode decomposition method to process the s(t) obtained in step 4), and get s(t)=∑imf(t)+r(t), as shown in Figure 9, where imf (t) is the eigenmode function of each frequency band, r(t) is the residual item, and then the imf(t) is screened, and the imf(t) with the highest matching degree with the heart rate distribution range of the large yellow croaker in the frequency domain is selected for Sliding average, calculate the average value of n consecutive items before and after to replace the current value in imf(t), and record the result as HR(t);

6) According to the HR (t) obtained in step 5), output directly as the cardiac waveform diagram; or further, adopt Fourier transform from the frequency domain to convert the HR (t) to the frequency domain, and obtain from the frequency spectrum The frequency value f _max with the largest amplitude, then the heart rate HR = f _max *60; using the peak detection method in the time domain, it is obtained that there are a peaks and b valleys coexisting on HR(t), and the heart rate Among them, t is the video duration, and its unit is second.

As shown in Figures 10, 11 and 12, they are respectively the cardiac waveform diagram, the fast Fourier transform spectrum and the peak and trough in the peak detection method of this scheme. The heart rate calculated by the frequency domain method is 65BPM, and the heart rate is calculated by the time domain method. The value is 68BPM, and the heart rate of the large yellow croaker is 67BPM measured synchronously by the electrode method.

The above results show that the results of the two heart rate calculation methods are consistent with the results of the verified method (electrode method) at the same time, which verifies the effectiveness and reliability of this method for heart rate detection of large yellow croaker to a certain extent.

In this embodiment, after the video of the target large yellow croaker is collected by a low-cost ordinary camera, the heart rate of the large yellow croaker can be obtained only by means of digital signal processing, so that there is no need to place sensors or electrodes on the large yellow croaker. Contact heart rate measurement realizes non-contact and non-invasive heart rate detection of large yellow croaker.

The scheme of the present invention can be widely used in the genetic breeding of large yellow croaker, the price is low, the realization method is simple, and the real-time is fast; the tolerance state to large yellow croaker can be effectively observed, and the large yellow croaker will not be caused by long-term contact measurement. stress response.

The scheme of this embodiment is based on two regions of interest (heart region of interest and background region of interest) for raw data acquisition, compared to the way of acquiring raw data with only a single region of interest (only heart region of interest). In other words (that is, the initial signal is the difference between multi-channel and single-channel), the detection results obtained by this scheme are more accurate, and there are fewer interference factors in the process calculation, making the results more valuable for reference; in addition, this scheme is based on the physical model— —The two-color reflection model decomposes the signal by projection. The two-color reflection model fully considers the interaction between the incident light and the biological skin surface, and can more accurately separate the reflected light signal containing the heart rate component; this scheme uses the independent component analysis method The signal is decomposed, and the independent component analysis can further separate the sub-components containing the stronger heart rate component according to the reflected light characteristics of the blood tissue; this scheme also uses the ensemble empirical mode decomposition to complete the frequency domain decomposition, and the ensemble empirical mode decomposition is The adaptive method does not need to set the cut-off frequency range, while the traditional frequency filter needs to set the cut-off frequency in advance. Animals living in water that contain red blood cells include aquatic animals with opaque skin and invisible heartbeat (such as large yellow croakers). The accuracy requirements of the methods required by the authors are completely different, resulting in different applicability to the target objects.

The above descriptions are only part of the embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent device or equivalent process conversion made by using the description of the present invention and the contents of the accompanying drawings, or directly or indirectly used in other related All technical fields are equally included in the scope of patent protection of the present invention.

Claims (5)

1. a kind of aquatic animal heart rate detection method based on video image processing, it is characterized in that, it comprises: 1) Obtain t frames of video images of the ventral surface of aquatic animals, then select a region of interest according to preset conditions, track and maintain it, and obtain a time signal sequence of the heart and the background region, wherein the region of interest includes the heart region of interest and background regions of interest; 2) According to the joint analysis method of multiple data sets, the common environmental noise shared by the heart region of interest and the background region of interest is screened out and removed; 3) According to the two-color reflectance model, the time signal sequence of the heart region of interest after removing the common environmental noise in step 2) is regarded as a linear combination of illumination change components, specular reflection components, and diffuse reflection components, and the static components are eliminated and Project to an orthogonal plane to remove the illumination variation component; 4) Perform independent component analysis on the time signal sequence of the heart region of interest that only contains specular reflection components and diffuse reflection components obtained through step 3), and filter out the strongest correlation with the green channel in the original heart region of interest signal sequence independent components of 5) decompose the independent components obtained in step 4) from the frequency domain, and screen out the subcomponents with the highest matching degree with the unique heart rate distribution range of aquatic animal species in the frequency domain; 6) according to step 5) process and obtain the sub-component to further output cardiac waveform or heart rate value; Among them, it includes: 1) Obtain t frames of video images of the ventral surface of aquatic animals, convert them into RGB video images, select heart ROIs and background ROIs, and use the target detection algorithm to track and maintain the ROIs; For the region of interest, calculate the average pixel intensity of the three color channels of each frame of RGB video image respectively, and obtain the cardiac region of interest signal sequence and the background region of interest signal sequence. The formula of the cardiac region of interest signal sequence is as follows: X _nr () = [ _hr (t); _hr (t); _hr ()] ^T , Wherein, R _hr (t), G _hr (t), B _hr () are respectively the average pixel intensity of the R, G, and B color channels of the heart region of interest video image; The formula of the background region of interest signal sequence is as follows: X _bg () = [ _bg (t); _bg (t); _bg ()] ^T ; Wherein, R _bg (t), G _bg (t), B _bg () are respectively the average pixel intensity of the R, G, and B color channels of the background region of interest video image; 2) According to the joint blind source separation method, using the mathematical formula SCV(t)=*X(t), the signal sequence X _hr () of the heart region of interest () and the signal sequence X _bg () of the background region of interest are combined Blind source separation is processed to obtain an unmixing matrix W, wherein W is an unmixing matrix, and SCV (t) is a source signal component vector; Further calculation to obtain the source signal component vector matrix SCV(t)=[SCV _hr1 ; CV _hr2 ; CV _hr3 ; CV _bg1 ; CV _bg2 ; CV _bg3 ] ^T ; Calculate and obtain each source signal component vector SCV hr() () of each source signal component vector SCV _hr() () of the cardiac region of interest signal sequence and each source signal component vector SCV _bg() () of the background region of interest signal sequence according to the source signal component vector matrix; According to the calculated correlation coefficient between each source signal component vector SCV _hr() () of the heart region of interest signal sequence and each source signal component vector SCV _bg() () of the background region of interest signal sequence, the maximum correlation The source signal component vector SCV _hr() () of the heart region of interest signal sequence corresponding to the coefficient is zeroed to obtain SCV _new (), and then according to SCV(t)=*X(t), the heart after removing the common environmental noise can be obtained ROI signal sequence and background ROI signal sequence X _new (t) = \SCV _new (); 3) According to the two-color reflectance model, the time signal sequence X _new (t) of the cardiac region of interest obtained in step 2) after removing the environmental noise is regarded as a linear combination of the illumination change component, the specular reflection component, and the diffuse reflection component, and the X _new () after time-domain normalization is projected onto the plane P=[p ₁ ; ₂ ] ^T orthogonal to the illumination change component, so as to obtain the time signal of the cardiac region of interest that only includes the specular reflection component and the diffuse reflection component sequence X _rflct (t) = *X _new (); 4) Perform independent component analysis on the cardiac region of interest time signal sequence X _rflct (t) obtained in step 3) that only contains specular reflection components and diffuse reflection components, namely X _rflct ()=A*S(t), wherein, A is the mixing matrix, S(t) is the independent component of the source signal, S(t)=[s ₁ (t); ₂ ()] ^T , Calculate the correlation coefficient between [ ₁ (T); ₂ ()] ^T and G _hr () in the signal sequence X _hr () of the original heart region of interest, and select the independent component s ( t); 5) Use the ensemble empirical mode decomposition method to process the s(t) obtained in step 4), and get s(t)=∑imf(t)+(t), where imf(t) is the Intrinsic mode function, r(t) is the residual item, and then filter the imf(t), select the imf(t) with the highest matching degree with the heart rate distribution range of aquatic animals in the frequency domain for sliding average, and calculate the continuous n The average value of the item is used to replace the current value in imf(t), and the result is recorded as HR(); 6) According to the HR () obtained in step 5), output directly as the cardiac waveform diagram; or further, adopt Fourier transform from the frequency domain to convert the HR () to the frequency domain, and obtain the amplitude from the frequency spectrum The maximum frequency value f _max , then the heart rate HR = f _max *60; using the peak detection method in the time domain, it is obtained that there are a peaks and b valleys coexisting on HR(), and the heart rate Among them, t is the video duration, and its unit is second. 2. the aquatic animal heart rate detection method based on video image processing as claimed in claim 1, is characterized in that, in step 1), described video image is the digital image that color is explained under different color spaces; An aquatic animal that has a heart organ and whose blood tissue components contain red blood cells. 3. the aquatic animal heart rate detection method based on video image processing as claimed in claim 1, it is characterized in that, in step 1), carry out choosing heart interest area and background interest area by manual selection or target detection algorithm and pass target The detection algorithm tracks and maintains it; The obtained temporal signal sequence of the heart and the background area is the temporal signal sequence obtained after the quantization noise is eliminated by spatial averaging. 4. the aquatic animal heart rate detection method based on video image processing as claimed in claim 1, is characterized in that, in step 1), in the 1st frame video image, adopts the manually selected mode to select respectively the heart feeling in the video image The region of interest and the background region of interest, and then use the target tracking algorithm to complete the tracking and maintenance of the region of interest in the subsequent t-1 frame video images; In step 2), the joint blind source separation method refers to a joint blind source separation framework implemented based on Gaussian independent vector analysis. 5. A computer-readable storage medium, characterized in that: at least one instruction, at least one section of program, code set or instruction set is stored in said storage medium, said at least one instruction, at least one section of program, code set Or the instruction set is loaded by the processor and executed to realize the method for detecting the heart rate of aquatic animals based on video image processing according to one of claims 1 to 4.