Fluorescence-image registration method, computer device and computer-readable storage medium

Abstract

Provided in the present disclosure are a fluorescence-image registration method, a computer device and a computer-readable storage medium. The method comprises: determining a normalized cross-power spectrum matrix between a reference fluorescence image and a fluorescence image to be registered, and performing form conversion on the normalized cross-power spectrum matrix, so as to obtain a phase angle sequence; determining slope information corresponding to each of a plurality of phase angles, and on the basis of the slope information respectively corresponding to the plurality of phase angles, determining weight values respectively corresponding to the plurality of phase angles, wherein the slope information corresponding to each phase angle comprises the slope between each phase angle and at least some of the other phase angles in the phase angle sequence; on the basis of the weight values respectively corresponding to the plurality of phase angles, performing straight-line fitting on the phase angles, so as to obtain a target slope of a straight line formed by the phase angles, and on the basis of the target slope, determining the sub-pixel offset of the fluorescence image to be registered that is relative to the reference fluorescence image; and on the basis of the sub-pixel offset, performing image registration. The method can improve the registration precision.

Description

Method for registering fluorescent images, computer device and computer-readable storage medium

[0001] The present disclosure claims priority from a Chinese patent application No. 2024104881971 filed on April 19, 2024, and entitled "Method for registering fluorescent images, computer device and computer-readable storage medium", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present disclosure relates to the technical field of computer image processing, in particular, to a method for registering fluorescent images, a computer device and a computer-readable storage medium. BACKGROUND

[0003] At present, high-throughput sequencing technology is the main means for identifying gene sequences. High-throughput sequencing technology uses optical imaging method to capture fluorescent signals that can reflect the types of bases on the gene sequence, and obtains fluorescent images. In multiple sequencing cycles, multiple fluorescent signals that can reflect the types of part or all of the bases on the gene sequence are captured to obtain multiple fluorescent images. By base recognition on the multiple fluorescent images, the types of part or all of the bases on the gene sequence are obtained. In multiple sequencing cycles, relative movement between the camera and the test platform occurs, which causes the positions of the fluorescent signals corresponding to the same gene sequence to deviate in different sequencing cycles, thereby reducing the accuracy of identifying the gene sequence. Moreover, the image registration process needs to be processed and analyzed in real time during the sequencing process, so real-time and accurate registration of the fluorescent images is an important sequencing link.

[0004] SUMMARY

[0005] The present disclosure provides at least a method for registering fluorescent images, a computer device and a computer-readable storage medium.

[0006] In a first aspect, the embodiments of the present disclosure provide a computer device, comprising: a processor, a memory, and a bus, the memory storing machine readable instructions executable by the processor, when the computer device is running, the processor and the memory communicate through the bus, the machine readable instructions are executed by the processor to perform a registration process of a fluorescence image, the fluorescence image comprising an image obtained by photographing a fluorescence signal generated by a sequencing object after a chemical reaction; the registration process of the fluorescence image comprises: determining a normalized cross power spectrum matrix between a reference fluorescence image and a fluorescence image to be registered, performing a form conversion on the normalized cross power spectrum matrix to obtain a phase angle sequence; the phase angle sequence comprises a plurality of phase angles and position information of the plurality of phase angles in a frequency domain coordinate system; determining slope information corresponding to each of the plurality of phase angles, and determining weight values corresponding to the plurality of phase angles based on the slope information corresponding to the plurality of phase angles respectively; the slope information corresponding to each of the plurality of phase angles comprises a slope between each of the plurality of phase angles and at least part of other phase angles in the phase angle sequence; performing a linear fitting on the phase angles based on the weight values corresponding to the plurality of phase angles respectively to obtain a target slope of a straight line formed by the phase angles, and determining a sub-pixel offset of the fluorescence image to be registered relative to the reference fluorescence image based on the target slope; performing image registration based on the sub-pixel offset.

[0007] In an optional implementation, when determining the slope information corresponding to each of the plurality of phase angles, the processor is configured to: traverse the plurality of phase angles in the phase angle sequence, and for a current phase angle traversed, perform the following determination process: determining a target phase angle corresponding to the current phase angle from the phase angle sequence; the target phase angle corresponding to the current phase angle comprises: a phase angle in the phase angle sequence other than the current phase angle; or, a phase angle in a neighboring region of the current phase angle and a phase angle in a center region of the phase angle sequence; determining a slope between the current phase angle and the corresponding target phase angle.

[0008] In an optional implementation, the processor, when determining the weight value corresponding to each of the plurality of phase angles based on the slope information corresponding to each of the plurality of phase angles, is configured to: determine a slope standard deviation corresponding to each of the plurality of phase angles based on the slope between each of the plurality of phase angles and other phase angles in the sequence of phase angles except for the phase angle; normalize the slope standard deviations corresponding to the plurality of phase angles to obtain a first weight component corresponding to each of the plurality of phase angles; the slope standard deviation corresponding to each of the plurality of phase angles is negatively correlated with the first weight component corresponding to each of the plurality of phase angles; and / or determine a second weight component based on first slope information of each of the plurality of phase angles relative to phase angles in a neighboring region and second slope information of each of the plurality of phase angles relative to phase angles in a center region; the difference between the first slope information and the second slope information is negatively correlated with the second weight component; and determine the weight value of each of the plurality of phase angles based on the first weight component and / or the second weight component.

[0009] In an optional implementation, the processor, when determining the second weight component based on the first slope of each of the plurality of phase angles relative to phase angles in a neighboring region and the second slope of each of the plurality of phase angles relative to phase angles in a center region, is configured to: determine a first slope mean value based on the first slope of each of the plurality of phase angles relative to phase angles in a neighboring region; determine a second slope mean value based on the second slope of each of the plurality of phase angles relative to phase angles in a center region; determine a mean difference between the first slope mean value and the second slope mean value; normalize the mean differences corresponding to the plurality of phase angles to obtain a second weight component corresponding to each of the plurality of phase angles.

[0010] In an optional implementation, the processor, when determining the normalized cross-power spectrum matrix between the reference fluorescent image and the to-be-registered fluorescent image, is configured to: perform windowing processing on the reference fluorescent image to obtain a windowed reference fluorescent image, and perform Fourier transform on the windowed reference fluorescent image to obtain a reference fluorescent image in a frequency domain; perform windowing processing on the to-be-registered fluorescent image to obtain a windowed to-be-registered fluorescent image, and perform Fourier transform on the windowed to-be-registered fluorescent image to obtain a to-be-registered fluorescent image in a frequency domain; and determine the normalized cross-power spectrum matrix based on the reference fluorescent image in the frequency domain and the to-be-registered fluorescent image in the frequency domain.

[0011] In an alternative implementation, when performing the form conversion on the normalized cross-power spectrum matrix to obtain the phase angle sequence, the processor is configured to: perform rank-1 matrix approximation on the normalized cross-power spectrum matrix to obtain a maximum singular value vector of the normalized cross-power spectrum matrix; wherein the maximum singular value vector comprises a left singular value vector and a right singular value vector; and perform phase unwrapping on the left singular value vector and the right singular value vector respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector.

[0012] In an alternative implementation, after performing the phase unwrapping on the left singular value vector and the right singular value vector respectively to obtain the first phase angle sequence corresponding to the left singular value vector and the second phase angle sequence corresponding to the right singular value vector, the processor is further configured to: filter each phase angle in the first phase angle sequence and the second phase angle sequence based on noise distribution characteristics and phase angle jump point distribution characteristics of each phase angle included in the first phase angle sequence and the second phase angle sequence respectively to obtain filtered first phase angle sequence and filtered second phase angle sequence.

[0013] In an alternative implementation, when performing the linear fitting on the phase angles based on the weight values respectively corresponding to the phase angles to obtain the target slope of the straight line formed by the phase angles, the processor is configured to: perform importance sorting on the phase angles according to the weight values respectively corresponding to the phase angles to obtain an importance sorting result; and perform the linear fitting on the phase angles using a target fitting method according to the importance sorting result to obtain the target slope.

[0014] In a second aspect, the embodiments of the present disclosure provide a method for registering fluorescent images, which comprises: determining a normalized cross-power spectrum matrix between a reference fluorescent image and a fluorescent image to be registered, and performing form conversion on the normalized cross-power spectrum matrix to obtain a phase angle sequence; the phase angle sequence comprises a plurality of phase angles and position information of the plurality of phase angles in a frequency domain coordinate system; determining slope information corresponding to each of the plurality of phase angles, and determining weight values respectively corresponding to the plurality of phase angles based on the slope information respectively corresponding to the plurality of phase angles; the slope information corresponding to each of the plurality of phase angles comprises a slope between each of the plurality of phase angles and at least part of other phase angles in the phase angle sequence; performing linear fitting on the phase angles based on the weight values respectively corresponding to the phase angles to obtain a target slope of a straight line formed by the phase angles, and determining a sub-pixel offset of the fluorescent image to be registered relative to the reference fluorescent image based on the target slope; and performing image registration based on the sub-pixel offset.

[0015] In an optional implementation, the determining of the slope information corresponding to each of the plurality of phase angles comprises: traversing the plurality of phase angles in the sequence of phase angles, and for a current phase angle in the traversal, performing the following determination process: determining a target phase angle corresponding to the current phase angle from the sequence of phase angles; the target phase angle corresponding to the current phase angle comprises: a phase angle other than the current phase angle in the sequence of phase angles; or a phase angle in a neighboring region of the current phase angle and a phase angle in a center region of the sequence of phase angles; and determining a slope between the current phase angle and the corresponding target phase angle.

[0016] In an optional implementation, the determining of the weight value corresponding to each of the plurality of phase angles based on the slope information corresponding to each of the plurality of phase angles comprises: determining a slope standard deviation corresponding to each of the plurality of phase angles based on a slope between each of the plurality of phase angles and a phase angle other than the phase angle in the sequence of phase angles; performing normalization processing on the slope standard deviations corresponding to the plurality of phase angles to obtain a first weight component corresponding to each of the plurality of phase angles; the slope standard deviation corresponding to each of the plurality of phase angles is negatively correlated with the first weight component corresponding to each of the plurality of phase angles; and / or determining a second weight component based on first slope information of each of the plurality of phase angles relative to a phase angle in a neighboring region and second slope information of each of the plurality of phase angles relative to a phase angle in a center region; a difference between the first slope information and the second slope information is negatively correlated with the second weight component; and determining the weight value corresponding to each of the plurality of phase angles based on the first weight component and / or the second weight component.

[0017] In an optional implementation, the determining of the second weight component based on the first slope of each of the plurality of phase angles relative to a phase angle in a neighboring region and the second slope of each of the plurality of phase angles relative to a phase angle in a center region comprises: determining a first slope mean value based on the first slope of each of the plurality of phase angles relative to a phase angle in a neighboring region; determining a second slope mean value based on the second slope of each of the plurality of phase angles relative to a phase angle in a center region; determining a mean difference between the first slope mean value and the second slope mean value; and performing normalization processing on the mean differences corresponding to the plurality of phase angles to obtain the second weight component corresponding to each of the plurality of phase angles.

[0018] In an optional implementation, the determining the normalized cross power spectrum matrix between the reference fluorescence image and the fluorescence image to be registered comprises: performing windowing processing on the reference fluorescence image to obtain a windowed reference fluorescence image, and performing Fourier transform on the windowed reference fluorescence image to obtain a reference fluorescence image in a frequency domain; and performing windowing processing on the fluorescence image to be registered to obtain a windowed fluorescence image to be registered, and performing Fourier transform on the windowed fluorescence image to be registered to obtain a fluorescence image to be registered in a frequency domain; and determining the normalized cross power spectrum matrix according to the reference fluorescence image in the frequency domain and the fluorescence image to be registered in the frequency domain.

[0019] In an optional implementation, the performing form conversion on the normalized cross power spectrum matrix to obtain a phase angle sequence comprises: performing rank 1 matrix approximation on the normalized cross power spectrum matrix to obtain a maximum singular value vector of the normalized cross power spectrum matrix; wherein the maximum singular value vector comprises a left singular value vector and a right singular value vector; and performing phase unwrapping on the left singular value vector and the right singular value vector respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector.

[0020] In an optional implementation, after the performing phase unwrapping on the left singular value vector and the right singular value vector respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector, the method further comprises: filtering the phase angles in the first phase angle sequence and the second phase angle sequence respectively based on noise distribution characteristics and phase angle jump point distribution characteristics of each phase angle included in the first phase angle sequence and the second phase angle sequence, to obtain filtered first phase angle sequence and second phase angle sequence.

[0021] In an optional implementation, the performing linear fitting on the phase angles based on the weight values corresponding to the phase angles respectively to obtain a target slope of a straight line formed by the phase angles comprises: performing importance sorting on the phase angles according to the weight values corresponding to the phase angles respectively to obtain an importance sorting result; and performing linear fitting on the phase angles by using a target fitting method according to the importance sorting result to obtain the target slope.

[0022] In a third aspect, the optional implementation of the present disclosure further provides a computer readable storage medium, and the computer readable storage medium stores a computer program. When the computer program is run, the steps in the second aspect or any possible implementation of the second aspect are performed.

[0023] In order to make the above objectives, characteristics and advantages of the present disclosure more obvious and easy to understand, the following preferred embodiments are specifically described below, and the accompanying drawings are described in detail as follows. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following will briefly introduce the drawings needed to be used in the embodiments. The drawings herein are incorporated into the specification and form a part of the specification, which show the embodiments consistent with the present disclosure, and are used to explain the technical solutions of the present disclosure together with the specification. It should be understood that the following drawings only show some embodiments of the present disclosure, and therefore should not be considered as a limitation to the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor.

[0025] FIG. 1 shows a schematic diagram of a gene sequencing system according to some embodiments of the present disclosure;

[0026] FIG. 2 shows a structural schematic diagram of an optical detection system according to some embodiments of the present disclosure;

[0027] FIG. 3 shows a flowchart of a registration method of a fluorescence image according to some embodiments of the present disclosure;

[0028] FIG. 4 shows a schematic diagram of a normalized cross-power spectrum matrix represented by a phase angle according to some embodiments of the present disclosure;

[0029] FIG. 5a shows an example diagram of a wrapped phase angle corresponding to a left singular value vector according to some embodiments of the present disclosure;

[0030] FIG. 5b shows an example diagram of an unwrapped phase angle corresponding to a left singular value vector according to some embodiments of the present disclosure;

[0031] FIG. 5c shows an example diagram of a wrapped phase angle corresponding to a right singular value vector according to some embodiments of the present disclosure;

[0032] FIG. 5d shows an example diagram of an unwrapped phase angle corresponding to a right singular value vector according to some embodiments of the present disclosure;

[0033] FIG. 6a shows an example diagram of a wrapped phase angle corresponding to a left singular value vector according to some embodiments of the present disclosure;

[0034] FIG. 6b shows an example diagram of a wrapped phase angle corresponding to a right singular value vector according to some embodiments of the present disclosure;

[0035] FIG. 6c shows an example diagram of an unwrapped phase angle corresponding to a left singular value vector according to some embodiments of the present disclosure;

[0036] FIG. 6d shows an example diagram of the unwrapped phase angle corresponding to the right singular value vector according to some embodiments of the present disclosure;

[0037] FIG. 6e shows an example diagram of the unwrapped phase angle corresponding to the left singular value vector after median filtering and slope denoising according to some embodiments of the present disclosure;

[0038] FIG. 6f shows an example diagram of the unwrapped phase angle corresponding to the right singular value vector after median filtering and slope denoising according to some embodiments of the present disclosure;

[0039] FIG. 7 shows a flowchart of determining the weight value of the phase angle according to some embodiments of the present disclosure;

[0040] FIG. 8a shows an example diagram of the weight of the phase angle corresponding to the left singular value vector according to some embodiments of the present disclosure;

[0041] FIG. 8b shows an example diagram of the weight of the phase angle corresponding to the right singular value vector according to some embodiments of the present disclosure;

[0042] FIG. 9a shows an example diagram of the registration error in the X-axis direction according to some embodiments of the present disclosure;

[0043] FIG. 9b shows an example diagram of the registration error in the Y-axis direction according to some embodiments of the present disclosure;

[0044] FIG. 9c shows an example diagram of the time of straight line fitting according to some embodiments of the present disclosure;

[0045] FIG. 10 shows a schematic diagram of a computer device according to some embodiments of the present disclosure. DETAILED DESCRIPTION

[0046] In order to make the objects, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure but not all of the embodiments. The components of the embodiments of the present disclosure described and shown herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure is not intended to limit the scope of the claimed present disclosure, but only represents selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present disclosure.

[0047] High-Throughput Sequencing (HTS) is also called Next Generation Sequencing (NGS) technology because it can obtain hundreds of thousands to millions of gene sequences at a time, greatly increasing the sequencing efficiency compared with traditional first-generation sequencing technology. However, high-throughput sequencing relies on high-precision sequencing equipment, and at present, high-throughput sequencing is still based on an optical imaging method, using a camera to capture a fluorescence signal that can reflect the type of base on the gene sequence, obtaining a fluorescence image, in multiple sequencing cycles, obtaining multiple fluorescence signals that can reflect the type of part or all of the bases on the gene sequence, obtaining multiple fluorescence images, and obtaining the type of part or all of the bases on the gene sequence by base recognition on the multiple fluorescence images.

[0048] Referring to FIG. 1, FIG. 1 is a whole architecture diagram of a gene sequencing system provided by an embodiment of the present disclosure. As shown in FIG. 1, the gene sequencing system 100 provided by the embodiment of the present disclosure includes a chip 10, a chip platform 20, a reagent storage container 30, a flow guide system 40, an optical detection system 50, and a computer system 60.

[0049] Among them, the chip 10 is attached with one or more sequencing objects; the sequencing object here can be a gene fragment, which includes a gene sequence of a certain length, for example, 150 bp in length; the sequencing object can also be a gene molecule.

[0050] The chip platform 20 is configured to fix and support the chip 10.

[0051] The reagent storage container 30 is configured to store one or more reagents for sequencing; here, the reagent can exemplarily include a Polymerase Chain Reaction (PCR) fluorescent reagent.

[0052] The flow guide system 40 is configured to controllably deliver the one or more reagents for sequencing from the reagent storage container 30 to the chip 10, so as to contact the sequencing object and occur a chemical reaction;

[0053] The optical detection system 50 is configured to excite a fluorescent label carried by the sequencing object after a chemical reaction with the reagent for sequencing occurs, and detect a fluorescence signal generated by exciting the fluorescent label, to generate a fluorescence image;

[0054] The computer system 60 is configured to acquire the fluorescence image from the optical detection system 50, and identify the gene sequence of the sequencing object according to the acquired fluorescence image.

[0055] Referring to FIG. 2, FIG. 2 is a structural schematic diagram of an optical detection system 50 provided by an embodiment of the present disclosure, the optical detection system 50 at least comprising a light source component 501 and an imaging assembly 502.

[0056] The imaging assembly 502 at least comprises a diaphragm sheet 5021, a dichroic mirror 5022-1, a microscope 5023 and an image sensor 5024, and the diaphragm sheet 5021 is provided with a plurality of light transmission holes 5025. Here, the image sensor 5024 may, for example, be an industrial camera.

[0057] The light source component 501 emits scattered excitation light, the excitation light passes through at least part of the light transmission holes on the diaphragm sheet 5021, forms a plurality of laser light beams and irradiates the dichroic mirror 5022-1; the dichroic mirror 5022-1 reflects the excitation light beams passing through the light transmission holes to the chip 10, the microscope 5023 collects the fluorescent signal generated by the sequencing object on the chip 10 after a chemical reaction; and the image sensor 5024 senses the fluorescent signal and generates a fluorescent image.

[0058] In an implementation manner, the light source component 501 may, for example, be a light-emitting diode, or may also be a semiconductor laser (LD), or simultaneously contain both. When the light source component 501 is a semiconductor laser, it can well assist the microscope 5023 in focusing.

[0059] In an implementation manner, the dichroic mirror 5022-1 may, for example, be fixedly inclined at a certain angle, and may, for example, be fixed by dispensing.

[0060] In an implementation manner, the diaphragm sheet 5021 may, for example, be a light-proof material, and may, for example, be a metal sheet. The scattered excitation light emitted by the light source component 501 can only pass through the light transmission hole part, and the non-light transmission hole part of the diaphragm sheet 5021 will shield the excitation light.

[0061] In a possible implementation manner, the imaging assembly 502 further comprises an attenuation sheet 5026, the attenuation sheet 5026 is arranged in parallel with the diaphragm sheet 5021 and is configured to attenuate the intensity of the excitation light beams. Referring to FIG. 2, the attenuation sheet 5026 may, for example, be arranged behind the diaphragm sheet 5021, and the excitation light beams can directly irradiate the attenuation sheet 5026 after passing through the light transmission hole.

[0062] In a possible implementation manner, the imaging assembly 502 further comprises a convex lens 5027, the convex lens 5027 is arranged in parallel with the attenuation sheet 5026 and is located between the attenuation sheet 5026 and the dichroic mirror 5022-1, and is configured to collimate the excitation light beams into parallel light beams.

[0063] In a possible implementation, the microscope 5023 can include an objective lens 50231 and a tube lens 50232, the objective lens 50231 is located below the dichroic mirror 5022-1, the tube lens 50232 is located above the dichroic mirror 5022-1, and the image sensor 5024 is located above the tube lens 50232.

[0064] One side of the objective lens 50231 is configured with a motor, for example, a voice coil motor. Through the motor on the objective lens 50231, the objective lens 50231 can be controlled to move up and down to achieve focusing of the objective lens 50231.

[0065] When sequencing, the image sensor 5024 (for example, an industrial camera) in the imaging assembly 502 needs to image the fluorescent signals generated by different regions of the chip 10 through the relative displacement between the chip 10 and the image sensor 5024 in different sequencing cycles. Due to the existence of mechanical errors, the image sensor 5024 may generate positional deviations between images after imaging the fluorescent signals generated by the same sequencing object in different sequencing cycles. The relative displacement amount between the image sensor 5024 and the chip 10 in each sequencing cycle needs to be calculated through image registration, so that the images after imaging the fluorescent signals generated by the same sequencing object in different sequencing cycles can be spatially aligned, so that there is no positional deviation in the imaging position of the same sequencing object in different sequencing cycles. The accuracy of image registration affects the accuracy of base sequence detection, and the image registration process needs to be processed and analyzed in real time during sequencing. Therefore, real-time and accurate registration of fluorescent images is a very important sequencing link.

[0066] In some current image registration schemes, special technical means are usually used to increase markers with obvious features such as marker points or marker lines on the chip 10, and the markers on the chip 10 are captured through optical imaging. The markers are used for registration between different fluorescent images. However, this method increases the manufacturing cost of the chip, and is limited by the performance of the image sensor 5024. The obtained fluorescent images have low contrast and no clear texture features, which leads to the inability to achieve high-precision image registration.

[0067] Based on the above research, the disclosure provides a fluorescent image registration method, which uses the phase features between a reference fluorescent image and a to-be-registered fluorescent image to determine the sub-pixel offset amount of the to-be-registered fluorescent image relative to the reference fluorescent image, and then registers the to-be-registered fluorescent image according to the sub-pixel offset amount, thereby improving the image registration accuracy.

[0068] The above-mentioned defects are the results of the inventors' practice and careful research, and thus the discovery process of the above-mentioned problems and the solutions proposed by the present disclosure to solve the above-mentioned problems should be the contributions of the inventors to the present disclosure.

[0069] It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0070] To facilitate the understanding of the present embodiment, first, a fluorescent image registration method disclosed by the present embodiment is introduced in detail, and the execution subject of the fluorescent image registration method provided by the present embodiment is generally a computer device with certain computing power, which for example includes a computer system in a gene sequencing system or a computer device for processing fluorescent images obtained by gene sequencing.

[0071] The fluorescent image registration method provided by the present embodiment is described below.

[0072] Referring to FIG. 3, a flowchart of the fluorescent image registration method provided by the present embodiment is shown, and the method includes steps S301-S304, wherein:

[0073] S301, determine the normalized cross-power spectrum matrix between the reference fluorescent image and the fluorescent image to be registered, and perform a form conversion on the normalized cross-power spectrum matrix to obtain a phase angle sequence; the phase angle sequence includes a plurality of phase angles and position information of the plurality of phase angles in a frequency domain coordinate system.

[0074] In a specific implementation, the reference fluorescent image may be any one of a plurality of frames of fluorescent images taken during the sequencing process of a target gene fragment, for example. The frame of fluorescent image is taken as a reference, and other fluorescent images are taken as images to be registered. The registration method provided by the present embodiment is used to perform image registration processing on the fluorescent images to be registered, so that all the fluorescent images to be registered are aligned with the reference fluorescent image.

[0075] The phase angle sequence of the fluorescent image to be registered and the reference fluorescent image is obtained by performing rank 1 matrix approximation on the normalized cross-power spectrum matrix of the fluorescent image to be registered and the reference fluorescent image, obtaining a singular value vector, and calculating the unwrapped phase angle according to the singular value vector. Then, the unwrapped phase angle is unwrapped to obtain the phase angle sequence.

[0076] The phase angle sequence includes a plurality of phase angles, and the number of phase angles is usually determined according to the size of the reference fluorescent image and the to-be-registered fluorescent image; here, the size of the reference fluorescent image and the to-be-registered fluorescent image is consistent. In the phase angle sequence, each phase angle corresponds to position information in a frequency domain coordinate system, and the position information indicates a coordinate point of the phase angle in the frequency domain coordinate system; the frequency domain coordinate system is a common rectangular coordinate system, in which the horizontal axis represents the ID of each phase angle, and the vertical axis represents the radian value of the phase angle.

[0077] The embodiment of the present disclosure provides a specific way of determining the normalized cross power spectrum matrix between the reference fluorescent image and the to-be-registered fluorescent image, including: performing windowing processing on the reference fluorescent image to obtain a reference fluorescent image after windowing processing, and performing Fourier transform on the reference fluorescent image after windowing processing to obtain a reference fluorescent image in the frequency domain; and performing windowing processing on the to-be-registered fluorescent image to obtain a to-be-registered fluorescent image after windowing processing, and performing Fourier transform on the to-be-registered fluorescent image after windowing processing to obtain a to-be-registered fluorescent image in the frequency domain; and determining the normalized cross power spectrum matrix according to the reference fluorescent image in the frequency domain and the to-be-registered fluorescent image in the frequency domain.

[0078] Exemplarily, due to the periodicity of Fourier transform, when performing Fourier transform on the image, in order to reduce the edge effect of the image, a two-dimensional window function can be used to weight the reference fluorescent image and the to-be-registered fluorescent image, so as to reduce the edge effect of the image when the image is expressed in the frequency domain. Commonly used window functions include Hanning window, Hamming window, Blackman window, Flat-top window, raised-cosine window, etc. The present disclosure does not particularly limit the use of which window function.

[0079] Exemplarily, the following formula (1) represents the spatial position relationship between the reference fluorescent image and the to-be-registered fluorescent image in the time domain.

[0080] g(x,y) = f(x-x0,y-y0) (1)

[0081] In the above formula (1), x and y represent the position coordinates of the pixel points in the reference fluorescent image and the to-be-registered fluorescent image, and x0 and y0 represent the offset amount in the row direction and the column direction between the reference fluorescent image and the to-be-registered fluorescent image. Wherein, g(x,y) and f(x,y) respectively represent the original reference fluorescent image and the to-be-registered fluorescent image in the time domain or after windowing processing.

[0082] The reference fluorescence image and the fluorescence image to be registered are subjected to Fourier transform to obtain a reference fluorescence image in the frequency domain and a fluorescence image to be registered in the frequency domain. Alternatively, the reference fluorescence image subjected to windowing processing and the fluorescence image to be registered subjected to windowing processing are subjected to Fourier transform, and according to the shift property in the Fourier transform property, the following formula (2) is obtained.

[0083] wherein G(x, y) represents the reference fluorescence image in the frequency domain, F(x, y) represents the fluorescence image to be registered in the frequency domain. M and N represent the size of the two-dimensional image. u and v correspond to a certain point on the frequency domain image, and the position of the point on the frequency domain image.

[0084] The expression of the normalized cross-power spectrum matrix of the reference fluorescence image and the fluorescence image to be registered in the frequency domain is shown in the following formula (3):

[0085] wherein * represents conjugation.

[0086] For example, referring to the schematic diagram of the normalized cross-power spectrum matrix represented by the phase angle (represented in the form of phase angle) shown in FIG. 4, each point in the schematic diagram represents the phase angle corresponding to each complex number in the cross-power spectrum matrix Q(u, v).

[0087] In addition, after obtaining the cross-power spectrum shown in FIG. 4, considering that aliasing and noise artifacts mainly interfere with high frequencies in the frequency domain, in order to effectively eliminate high-frequency interference, the cross-power spectrum matrix can also be subjected to low-pass filtering processing. Commonly used low-pass filters include ideal low-pass filters, Gaussian low-pass filters, Butterworth low-pass filters, trapezoidal low-pass filters, etc., and the present disclosure does not make any limitation on which filter is used to filter the cross-power spectrum matrix.

[0088] After obtaining the normalized cross-power spectrum matrix, the normalized cross-power spectrum matrix can be subjected to formal conversion in the following manner to obtain a phase angle sequence: the normalized cross-power spectrum matrix is subjected to rank 1 matrix approximation to obtain a maximum singular value vector of the normalized cross-power spectrum matrix; wherein the maximum singular value vector includes a left singular value vector and a right singular value vector; the left singular value vector and the right singular value vector are subjected to phase unwrapping respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector.

[0089] Specifically, the normalized cross-power spectrum matrix can be subjected to rank 1 matrix approximation using SVD decomposition, and in addition, the normalized power method mentioned in the scheme provided in the patent application with publication number “CN115937282A” can also be used to subject the normalized cross-power spectrum matrix to rank 1 matrix approximation.

[0090] Taking the rank-1 matrix approximation of the normalized cross-power spectrum matrix by SVD decomposition as an example, the normalized cross-power spectrum is shown in the above formula (3), and the normalized cross-power spectrum matrix can be decomposed into the following form:

[0091] Q(u,v) = exp(-j2πux0)exp(-j2πvy0)

[0092] Further, the normalized cross-power spectrum matrix is expressed in the form of multiplication of a row vector and a column vector:

[0093] wherein:

[0094] Since the component with the largest singular value is less affected by noise, the rank-1 matrix approximation can play a role in suppressing noise.

[0095] The rank-1 matrix approximation of Q is obtained as follows:

[0096] wherein σ1 is the largest singular value of Q, u1 and v1 are the left singular vector and the right singular vector corresponding to σ1 respectively, and thus and

[0097] Then, the phase unwrapping is performed on the phase angles corresponding to the left singular vector and the right singular vector respectively to obtain a first phase angle sequence corresponding to the left singular vector and a second phase angle sequence corresponding to the right singular vector.

[0098] Referring to some phase unwrapping examples shown in FIGS. 5a-5d, wherein: the horizontal axis of each example graph represents the serial number ID of each phase angle in the phase angle sequence. The vertical axis represents the radian of the unwrapped true phase angle.

[0099] In an ideal state, the unwrapped phase angle is theoretically a straight line, which can be fitted with a straight line equation in a least squares fitting manner to obtain the target slope of the straight line. According to the expression relationship between the target slope and the offset, the sub-pixel offset of the to-be-registered fluorescent image relative to the reference fluorescent image can be known. The expression relationship between the target slope and the offset is shown in the following formula (5):

[0100] wherein k u represents the straight line slope corresponding to the unwrapped phase angle of the left principal singular vector u1; kv represents the straight line slope corresponding to the unwrapped phase angle of the right principal singular vector v1.

[0101] In real-time sequencing, the fluorescence signal of the base is weakened with the increase of the sequencing cycle, which reduces the signal-to-noise ratio of the fluorescence image and causes noise in the image. In addition, for high-concentration samples, more noise will appear in the corresponding fluorescence image. The noise will cause a large error in the phase angle in the phase unwrapping process, which will further affect the accuracy of the straight line fitting.

[0102] Therefore, when performing phase angle unwrapping, a phase angle denoising method can be used to remove some phase angle noise. In some implementations, median filtering, mean filtering, Gaussian filtering, etc. can be used for phase angle denoising. The following describes a phase angle denoising process provided by the present disclosure in detail.

[0103] In some possible embodiments provided by the present disclosure, the phase angles corresponding to the left singular value vector and the right singular value vector of the cross-power spectrum matrix are respectively subjected to phase unwrapping; after obtaining the first phase angle sequence and the second phase angle sequence, the method further includes: based on the noise distribution characteristics and the phase jump point distribution characteristics of each phase angle included in the first phase angle sequence and the second phase angle sequence, respectively filtering the phase angles in the first phase angle sequence and the second phase angle sequence to obtain the first phase angle sequence and the second phase angle sequence after filtering.

[0104] Here, by analyzing the phase angles in the phase angle sequence, the phase angle sequence can be divided into a phase angle jump region and a phase angle non-jump region in a frequency domain coordinate system. According to the distribution of the phase angle jump region and the non-jump region, the phase jump point distribution characteristics are determined. That is, the phase jump points are distributed in the jump region. Relative to the distribution characteristics of the phase jump points, the noise points existing in the non-jump region are likely to be considered as phase jump points. Therefore, when filtering the phase feature points, not only the noise but also the true jump points should be retained.

[0105] For example, the present disclosure uses median filtering and slope denoising to denoise the phase angles in the phase angle sequence. Specifically, the wrapped phase angles are first subjected to median filtering, and then the phase angles after median filtering are subjected to slope denoising. An initial slope is set. In multiple iterations, each phase angle is unwrapped, and the phase angles satisfying the following formula (6) are marked as 1, and the phase angles not satisfying the formula are marked as 0. The phase angle set composed of the phase angles marked as 1 is found out.

[0106] |θ i+1 -(k×slope+D)|≤ε (6)

[0107] wherein D is an initial value of the i th unwrapped phase angle, denoted as θ i; i = 0, 1 …… n; k is the interval between two phase angles, k is initially 1; slope: initial slope; θ i+1 : represents the i+1 th unwrapped phase angle; when the unwrapped phase angle satisfies formula (6), D = θ i+1 , k = 1; otherwise k = k + 1; ε is a threshold value, in radians.

[0108] For example, see some filtered noise examples shown in Figures 6a-6f, wherein: in Figures 6e and 6f, not only are the true jump points in the 0-100, 400-411 regions (jump regions) retained, but the noise in the 100-400 region (non-jump region) is also filtered. Moreover, the obtained phase angles satisfy the linear distribution characteristics, while the phase angles obtained in Figures 6c and 6d do not satisfy the linear distribution characteristics.

[0109] For example, taking Figure 6e as the first phase angle sequence and Figure 6f as the second phase angle sequence, participating in the linear fitting process can improve the speed of linear fitting compared to Figures 6c and 6d.

[0110] After obtaining the phase angle sequence in the above S301 step, the method further comprises the following steps:

[0111] S302, determining the slope information corresponding to each of the plurality of phase angles, and determining the weight values corresponding to the plurality of phase angles based on the slope information corresponding to each of the plurality of phase angles; the slope information corresponding to each of the plurality of phase angles includes the slope between each of the plurality of phase angles and at least part of the other phase angles in the phase angle sequence.

[0112] As shown in Figures 6e and 6f, there are gross errors at both ends of the phase angle, and directly using the least squares linear fitting method will reduce the accuracy of linear fitting, so it is necessary to use a certain method to remove gross errors before linear fitting. Commonly used methods for removing gross errors include RANSAC, PROSAC, and robust estimation based on weight selection iteration.

[0113] RANSAC and PROSAC both have a selection strategy when calculating the model. Before calculating the model, a quality function is constructed to sort the phase angles, so that the phase angles with high quality are given priority to participate in the model calculation, which can speed up the convergence of the model.

[0114] The phase angle distribution after denoising filtering is approximately a straight line, and the slope between the phase angles should be approximately consistent. The phase angle deviating from the straight line distribution more, the greater the change of the slope distance from all other phase angles. The slope sequence of each phase angle distance from all other phase angles can be calculated, m phase angles correspond to m slope sequences, the standard deviation of the m slope sequences is calculated, and the standard deviation is used as one of the phase angle weights. The greater the standard deviation of the slope sequence corresponding to the phase angle, the higher the possibility of noise of the phase angle. By observing the distribution of the phase angle, the noise of the phase angle is more distributed at both ends of the phase angle, and the phase angle distributed in the middle region is more stable. The slope distribution of the phase angle and the surrounding region points should be approximately consistent with the slope distribution of the phase angle distance from the middle region of the phase angle. Therefore, the slope distribution rule can also be used as one of the weights of the phase angle.

[0115] Specifically, the embodiment of the present disclosure provides a specific way to determine the slope information corresponding to each phase angle in a plurality of phase angles, comprising: traversing a plurality of phase angles in the phase angle sequence, and for the current phase angle traversed, performing the following determination process: determining the corresponding target phase angle for the current phase angle from the phase angle sequence; the target phase angle corresponding to the current phase angle includes: the phase angle in the phase angle sequence except the current phase angle; or, the phase angle in the adjacent region corresponding to the current phase angle, and the phase angle in the center region of the phase angle sequence; determining the slope between the current phase angle and the corresponding target phase angle.

[0116] In a specific implementation, in the process of traversing a plurality of phase angles, for the current phase angle traversed, assuming that the phase angle sequence includes k phase angles, respectively represented as: M1, M2, M3, …, M k , in the case that the target phase angle includes other phase angles in the phase angle sequence except the current phase angle, assuming that the current phase angle is M i , the other phase angles include: M1, M2, …, M (i-1) , M (i+1) , …, M k .

[0117] In the case that the target phase angle includes the phase angle in the adjacent region corresponding to the current phase angle, and the phase angle in the center region of the phase angle sequence, the phase angle in the adjacent region corresponding to the current phase angle, for example, includes each phase angle in the adjacent region determined according to a preset step n, with the position of the current phase angle in the phase angle sequence as the center. Illustratively, assuming that the phase angle sequence includes k phase angles, respectively represented as: M1, M2, M3, …, M k , for the i-th phase angle M i traversed, the phase angle in the adjacent region corresponding to the i-th phase angle M iThe corresponding adjacent regions include, for example, i-n~i+n; here, because there are cases in which the current phase angle is at the beginning or end of the phase angle sequence, the adjacent phase angles include, for example, M1, …, M

[0118] If i is less than or equal to n, the adjacent phase angles include, for example, M1, …, M (i+1) , …, M (i+n) .

[0119] If i is greater than n and less than (k-n), the adjacent phase angles include, for example, M (i-n) , M (i-n+1) , …, M (i-1) , M (i+1) , …, M (i+n) .

[0120] If i is greater than or equal to (k-n), the adjacent phase angles include, for example, M (i-n) , M (i-n+1) , …, M (i-1) , M (i+1) , …, M k .

[0121] The center region phase angle is, for example, a center region determined according to k, that is, a region located at the center of the phase angle sequence, and the size of the center region is, for example, s, that is, the center region includes, for example, M 25 and M 26 . If k is 50 and s is 2, the center region phase angle includes, for example, M 25 , M 26 and M 27 .

[0122] After the target phase angle is determined, the slope between the current phase angle and the corresponding target phase angle can be determined.

[0123] The slope between the current phase angle and the corresponding target phase angle is determined as the slope information corresponding to the current phase angle.

[0124] The disclosure also provides a specific manner for determining a weight value corresponding to each of the plurality of phase angles based on the slope information corresponding to each of the plurality of phase angles, including:

[0125] determining a slope standard deviation corresponding to each of the plurality of phase angles based on the slope between each of the plurality of phase angles and other phase angles in the phase angle sequence except for each of the plurality of phase angles; performing normalization processing on the slope standard deviations corresponding to the plurality of phase angles to obtain a first weight component corresponding to each of the plurality of phase angles; and the slope standard deviation corresponding to each of the plurality of phase angles is negatively correlated with the first weight component corresponding to each of the plurality of phase angles;

[0126] and / or,

[0127] Determining a second weight component based on first slope information of each phase angle relative to the phase angle of the adjacent area and second slope information of each phase angle relative to the phase angle of the central area; wherein the degree of difference between the first slope information and the second slope information is negatively correlated with the second weight component;

[0128] A weight value for each phase angle is determined based on the first weight component and / or the second weight component.

[0129] Here, the weight value of the phase angle may be composed of only the first weight component, only the second weight component, or the first weight component and the second weight component in a certain ratio, which is selected according to the actual application.

[0130] For example, referring to the flowchart of determining the weight value of the phase angle shown in FIG7 , FIG7 includes steps S701 to S707, wherein:

[0131] S701: Obtain a phase angle sequence.

[0132] Here, the phase angle sequence may be either the first phase angle sequence or the second phase angle sequence. The phase angle sequence includes m denoised phase angles.

[0133] S702: Determine each phase angle relative to its slope sequence.

[0134] In the above step S701, m phase angles are obtained. In this step, the m slope sequences correspond to the m phase angles respectively. Each phase angle corresponds to a slope sequence.

[0135] S703: Calculate the standard deviation of each slope sequence and determine a first weight component.

[0136] The present disclosure provides some possible embodiments, in which, for each phase angle, a slope sequence corresponding to each phase angle is obtained based on the slopes corresponding to each phase angle and other phase angles except itself; and a standard deviation of the slope sequence of each phase angle is calculated to obtain the standard deviation of the slope corresponding to each phase angle.

[0137] Here, the slope standard deviation represents the similarity of each data point in the slope sequence, where each data point is the slope of each phase angle relative to the other phase angles in the slope sequence. This way, the similarity of each phase angle to the other phase angles is determined. A larger slope standard deviation indicates a lower similarity, meaning it deviates further from the straight line.

[0138] Thus, according to the slope standard deviation, a first weight component of each phase angle is determined, and the greater the slope standard deviation, the smaller the first weight component.

[0139] S704, first slope information of each phase angle relative to phase angles of adjacent regions is calculated.

[0140] Here, the first slope information is represented as a slope mean value of each phase angle relative to phase angles of adjacent regions in the surrounding region, and the phase angles of adjacent regions can be one or more.

[0141] S705, second slope information of each phase angle relative to the phase angle of the center region is calculated.

[0142] Here, the second slope difference information is represented as a slope mean value of each phase angle relative to the phase angle of the center region distributed in the center region, and the phase angle of the center region can also be one or more.

[0143] S706, the difference between the slope mean value indicated by the first slope difference information and the slope mean value indicated by the second slope difference information is calculated to determine the second weight component.

[0144] Here, when determining the second weight component, for example, a first slope mean value can be determined based on the first slope of each phase angle relative to the phase angle of the adjacent region; and

[0145] a second slope mean value is determined based on the second slope of each phase angle relative to the phase angle of the center region;

[0146] The mean difference between the first slope mean value and the second slope mean value is determined, and the mean difference corresponding to each of the plurality of phase angles is normalized to obtain the second weight component corresponding to each of the plurality of phase angles.

[0147] The greater the difference between the slope mean value indicated by the first slope information and the slope mean value indicated by the second slope information, the higher the possibility of being noise, that is, the greater the distance of the target feature point relative to other feature points. The second weight component is also lower.

[0148] S707, based on the proportion of the normalized first weight component and the second weight component in the weight value, the weight value of each phase angle is determined.

[0149] Preferably, the proportion of the first weight component and the second weight component in the weight value can be set to 50%, and then linear fitting is performed according to the obtained weight value.

[0150] The S702-S703 steps and the S704-S706 steps do not have a sequence of execution, that is, the first weight component and the second weight component can be executed simultaneously or without a sequence.

[0151] After the weight value of the phase angle is determined in the S302, the method further includes the following steps:

[0152] S303, based on the weight value corresponding to each of the plurality of phase angles, performing linear fitting on the phase angles to obtain a target slope of a straight line formed by the phase angles, and determining a sub-pixel offset of the to-be-registered fluorescent image relative to the reference fluorescent image based on the target slope.

[0153] In some possible embodiments provided by the present disclosure, when performing linear fitting on the phase angles based on the weight value corresponding to each of the plurality of phase angles, for example, the plurality of phase angles are sorted in importance according to the weight value corresponding to each of the plurality of phase angles to obtain an importance sorting result; and the linear fitting on the phase angles is performed by using a target fitting method according to the importance sorting result to obtain the target slope.

[0154] Specifically, after the linear fitting on the phase angles is performed based on the position information of each of the plurality of phase angles in the phase angle sequence and the weight value of the phase angle to obtain the target slope, the sub-pixel offset of the to-be-registered image relative to the reference image can be obtained based on the expression relationship between the target slope and the sub-pixel offset.

[0155] When performing linear fitting, the least square method can be used to perform linear fitting, or the PROSAC algorithm can be used to perform linear fitting.

[0156] When the least square method is used to perform linear fitting, the phase angle with a weight value greater than a preset threshold value can be determined as a target phase angle, and the least square method is used to process the target phase angle to obtain a target straight line.

[0157] When the PROSAC algorithm is used to perform linear fitting, the phase angle with a large weight value is preferentially selected to participate in linear fitting according to the weight value of the phase angle in the phase angle sequence.

[0158] For example, the least square method performs linear fitting only once, and the phase angle with a weight value less than a preset threshold value is abandoned, and when the PROSAC algorithm is used to perform linear fitting, the phase angle with a large weight value is preferentially selected to perform linear fitting in each iteration until a convergence condition is reached.

[0159] For example, referring to the phase angle weight example diagrams shown in FIGS. 8a and 8b, wherein:

[0160] In FIGS. 8a and 8b, the horizontal axis represents the serial number ID of each phase angle in the phase angle sequence. The vertical axis represents the radian corresponding to each phase angle. Among them, the larger the area of each phase angle (dot) in the figure represents the lower the weight value, and the smaller the area represents the higher the weight value. The center position of the circle is mapped to the vertical axis to represent the radian value of the phase angle.

[0161] After determining the sub-pixel offset of the image to be registered relative to the reference image in the above S303 step, the method further comprises the following steps.

[0162] S304, image registration based on the sub-pixel offset.

[0163] For example, after obtaining the sub-pixel offset of the fluorescence image to be registered relative to the reference fluorescence image, the fluorescence image to be registered is registered according to the sub-pixel offset to align the fluorescence image to be registered with the reference fluorescence image.

[0164] In addition, in order to more intuitively show the advantages of the present disclosure when performing linear fitting, the present disclosure also provides a specific example, which compares the linear fitting of the PROSAC algorithm with the linear fitting of the existing algorithm RANSAC in the above embodiment. Specifically:

[0165] In both methods, SVD phase decomposition is used to obtain a singular value vector, and the phase angle corresponding to the singular value vector is unwrapped to obtain a phase angle sequence.

[0166] The unwrapped phase angle sequence is fitted with the RANSAC algorithm, and the phase angle sequence is fitted with the PROSAC algorithm after the phase angle is weighted in the above example of the present disclosure.

[0167] It should be noted that in order to control the variables of the two methods, the phase angle sequence corresponding to the two methods is subjected to the phase angle filtering process in the above example of the present disclosure.

[0168] The test hardware is: Intel(R) Core(TM) i7-10750H CPU @ 2.60GHz 32G memory.

[0169] During the test, first, 2000 salt and pepper noises are randomly added to the original fluorescence image with a size of 500x500, and the noise intensity is valued at 0 or 255. Then, Gaussian noise with a mean of 1000 and a standard deviation of 3 is added, and then 100 sub-pixel offsets are randomly set within 0-11 pixels to perform affine transformation on the original fluorescence image containing noise (salt and pepper noise and Gaussian noise), obtaining 100 images to be registered containing sub-pixel offsets.

[0170] Respectively, two methods are used to sub-pixel registration of 100 images to be registered, each known sub-pixel offset as a result value, respectively, to register the error. Get the linear fitting experiment comparison example as shown in Figure 9a~9c; Wherein:

[0171] In Figure 9a, Err_X_PROSAC represents the X-axis direction registration error distribution curve when the present disclosure is used to register 100 images to be registered; Err_X_RANSAC represents the X-axis direction registration error distribution curve when the existing RANSAC algorithm is used to register 100 images to be registered.

[0172] In Figure 9b, Err_Y_PROSAC represents the Y-axis direction registration error distribution curve when the present disclosure is used to register 100 images to be registered; Err_Y_RANSAC represents the Y-axis direction registration error distribution curve when the existing RANSAC algorithm is used to register 100 images to be registered.

[0173] In Figure 9a and Figure 9b, the horizontal axis represents the corresponding serial number of 100 images to be registered; The vertical axis represents the pixel error, that is, the real sub-pixel offset minus the sub-pixel offset obtained by the registration algorithm. Unit: pixel.

[0174] In Figure 9c, Line FitCost_PROSAC represents the processing time distribution curve when the present disclosure is used to perform linear fitting on 100 images to be registered; Line FitCost_RANSAC represents the processing time distribution curve when the existing RANSAC algorithm is used to perform linear fitting on 100 images to be registered. The horizontal axis represents the corresponding serial number of 100 images to be registered; The vertical axis represents the linear fitting time, unit: millisecond.

[0175] By calculation, the comparison experiment results are as follows: using the existing RANSAC algorithm to process 100 images to be registered, the following processing results are obtained: the mean value of the absolute value of the registration error in the X-axis direction: 0.0563795; The mean value of the absolute value of the registration error in the Y-axis direction: 0.0602355; Linear fitting time: 10.3ms.

[0176] When the present method is used to process 100 images to be registered, the following processing results are obtained: the mean value of the absolute value of the registration error in the X-axis direction: 0.0459316; The mean value of the absolute value of the registration error in the Y-axis direction: 0.0510514; Linear fitting time: 1.8ms. It can be seen that the present method further accelerates the convergence time of linear fitting under the condition of the same matching accuracy compared with the existing RANSAC algorithm, thereby improving the image registration speed.

[0177] The method for registering fluorescent images provided by the embodiments of the present disclosure determines a normalized cross-power spectrum matrix between a reference fluorescent image and a fluorescent image to be registered, performs a form conversion on the normalized cross-power spectrum matrix to obtain a phase angle sequence; the phase angle sequence includes a plurality of phase angles and position information of the plurality of phase angles in a frequency domain coordinate system; slope information corresponding to each of the plurality of phase angles is determined, and weight values corresponding to the plurality of phase angles are determined based on the slope information corresponding to the plurality of phase angles; the slope information corresponding to each of the phase angles includes a slope between each of the phase angles and at least some other phase angles in the phase angle sequence; a straight line fitting is performed on the phase angles based on the weight values corresponding to the plurality of phase angles to obtain a target slope of a straight line formed by the phase angles, and a sub-pixel offset of the fluorescent image to be registered relative to the reference fluorescent image is determined based on the target slope; image registration is performed based on the sub-pixel offset, so that the phase characteristics between the reference fluorescent image and the fluorescent image to be registered are used to describe the specific situation of the displacement between the reference fluorescent image and the fluorescent image to be registered caused by mechanical movement, the fluorescent images can be registered at a sub-pixel level, and the registration accuracy is improved.

[0178] Those skilled in the art can understand that, in the above method of the specific implementation, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process, and the specific execution order of each step should be determined by its function and possible internal logic.

[0179] The embodiments of the present disclosure also provide a computer device, as shown in FIG. 10, which is a structural schematic diagram of the computer device provided by the embodiments of the present disclosure, and includes a processor 11, a memory 12 and a bus 13.

[0180] The memory 12 stores machine readable instructions executable by the processor 12, when the computer device is running, the processor 11 and the memory 11 communicate through the bus 13, the machine readable instructions are executed by the processor 11 to perform a registration process of a fluorescent image, the fluorescent image includes an image obtained by photographing a fluorescent signal generated after a chemical reaction of a sequencing object, and the sequencing object has a genetic sequence; the registration process of the fluorescent image includes:

[0181] A normalized cross-power spectrum matrix between a reference fluorescent image and a fluorescent image to be registered is determined, a form conversion is performed on the normalized cross-power spectrum matrix to obtain a phase angle sequence; the phase angle sequence includes a plurality of phase angles and position information of the plurality of phase angles in a frequency domain coordinate system;

[0182] determining slope information corresponding to each of the plurality of phase angles, and determining weight values corresponding to the plurality of phase angles based on the slope information corresponding to the plurality of phase angles respectively; the slope information corresponding to each of the plurality of phase angles comprises slopes between each of the plurality of phase angles and at least some other phase angles in the sequence of phase angles;

[0183] performing linear fitting on the plurality of phase angles based on the weight values corresponding to the plurality of phase angles respectively, to obtain a target slope of a straight line formed by the plurality of phase angles, and determining a sub-pixel offset of the to-be-registered fluorescent image relative to the reference fluorescent image based on the target slope;

[0184] performing image registration based on the sub-pixel offset.

[0185] In an optional implementation, the processor 11, when determining the slope information corresponding to each of the plurality of phase angles, is configured to:

[0186] traversing the plurality of phase angles in the sequence of phase angles, and performing the following determination process for a current phase angle traversed:

[0187] determining a target phase angle corresponding to the current phase angle from the sequence of phase angles;

[0188] the target phase angle corresponding to the current phase angle comprises phase angles in the sequence of phase angles other than the current phase angle, or phase angles of a neighboring region of the current phase angle and a central region of the sequence of phase angles;

[0189] determining a slope between the current phase angle and the target phase angle.

[0190] In an optional implementation, the processor 11, when determining the weight values corresponding to the plurality of phase angles based on the slope information corresponding to the plurality of phase angles respectively, is configured to:

[0191] determining a slope standard deviation corresponding to each of the plurality of phase angles based on slopes between each of the plurality of phase angles and other phase angles in the sequence of phase angles other than each of the plurality of phase angles, performing normalization processing on the slope standard deviations corresponding to the plurality of phase angles respectively to obtain first weight components corresponding to the plurality of phase angles respectively, and the slope standard deviation corresponding to each of the plurality of phase angles is negatively correlated with the first weight component corresponding to each of the plurality of phase angles;

[0192] and / or,

[0193] determine a second weight component based on first slope information of each of the phase angles relative to phase angles of adjacent regions and second slope information of each of the phase angles relative to a phase angle of a center region; the difference between the first slope information and the second slope information is negatively correlated with the second weight component;

[0194] determine a weight value of each of the phase angles based on the first weight component and / or the second weight component.

[0195] In an optional implementation, the processor 11, when determining a second weight component based on first slopes of each of the phase angles relative to phase angles of adjacent regions and second slopes of each of the phase angles relative to a phase angle of a center region, is configured to:

[0196] determine a first slope mean value based on the first slopes of each of the phase angles relative to the phase angles of the adjacent regions; and

[0197] determine a second slope mean value based on the second slopes of each of the phase angles relative to the phase angle of the center region;

[0198] determine a mean difference of the first slope mean value and the second slope mean value, and normalize the mean difference corresponding to each of the phase angles to obtain the second weight component corresponding to each of the phase angles.

[0199] In an optional implementation, the processor 11, when determining a normalized cross power spectrum matrix between a reference fluorescent image and a fluorescent image to be registered, is configured to:

[0200] perform windowing processing on the reference fluorescent image to obtain a reference fluorescent image after windowing processing, and perform Fourier transform on the reference fluorescent image after windowing processing to obtain a reference fluorescent image in a frequency domain; and

[0201] perform windowing processing on the fluorescent image to be registered to obtain a fluorescent image to be registered after windowing processing, and perform Fourier transform on the fluorescent image to be registered after windowing processing to obtain a fluorescent image to be registered in a frequency domain; and

[0202] determine the normalized cross power spectrum matrix according to the reference fluorescent image in the frequency domain and the fluorescent image to be registered in the frequency domain.

[0203] In an optional implementation, the processor 11, when performing form conversion on the normalized cross power spectrum matrix to obtain a phase angle sequence, is configured to:

[0204] perform rank 1 matrix approximation on the normalized cross power spectrum matrix to obtain a maximum singular value vector of the normalized cross power spectrum matrix; wherein the maximum singular value vector includes a left singular value vector and a right singular value vector.

[0205] phase-unwrap the left singular value vector and the right singular value vector respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector.

[0206] In an optional implementation, the processor 11, after phase-unwrapping the left singular value vector and the right singular value vector respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector, is further configured to:

[0207] filter each phase angle in the first phase angle sequence and the second phase angle sequence based on noise distribution characteristics and phase angle jump point distribution characteristics of each phase angle included in the first phase angle sequence and the second phase angle sequence respectively, to obtain the first phase angle sequence and the second phase angle sequence after filtering.

[0208] In an optional implementation, when the processor 11 performs linear fitting on the phase angles based on the weight values corresponding to the phase angles respectively to obtain a target slope of a straight line formed by the phase angles, the processor 11 is configured to:

[0209] perform importance sorting on the phase angles according to the weight values corresponding to the phase angles respectively to obtain an importance sorting result.

[0210] perform linear fitting on the phase angles using a target fitting method according to the importance sorting result to obtain the target slope.

[0211] The embodiments of the present disclosure further provide a computer readable storage medium, which stores a computer program. The computer program is run by a processor to perform the steps of the registration method of the fluorescent image described in the above method embodiments. The storage medium can be a volatile or non-volatile computer readable storage medium.

[0212] The embodiments of the present disclosure further provide a computer program product, which carries a program code. The program code includes instructions that can be used to perform the steps of the registration method of the fluorescent image described in the above method embodiments. For details, refer to the above method embodiments, which will not be described here.

[0213] The computer program product can be implemented by hardware, software or a combination thereof. In an optional embodiment, the computer program product is embodied in a computer storage medium. In another optional embodiment, the computer program product is embodied in a software product, such as a software development kit (SDK) or the like.

[0214] It can be understood that, before using the technical solutions disclosed in the embodiments of the present disclosure, the type, use range, use scenario and the like of personal information involved in the present disclosure should be informed to the user and the authorization of the user should be obtained according to relevant laws and regulations.

[0215] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here. In several embodiments provided by the present disclosure, it should be understood that the disclosed system, device and method can be implemented by other ways. The device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there can be another division way in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some communication interface, device or unit, and can be electrical, mechanical or other forms.

[0216] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the present embodiment.

[0217] In addition, each functional unit in the various embodiments of the present disclosure can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0218] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a nonvolatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present disclosure essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present disclosure. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0219] Finally, it should be noted that: the above-described embodiments are only specific embodiments of the present disclosure, used to illustrate the technical solutions of the present disclosure, and not to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art who is familiar with the technology in the art can still make modifications or easily think of changes to the technical solutions described in the foregoing embodiments, or make equivalent replacements to some of the technical features; and these modifications, changes or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A computer device, characterized by, The processor, the memory and the bus, the memory stores machine readable instructions executable by the processor, when the computer equipment runs, the processor and the memory are communicated through bus, the machine readable instructions are executed when the processor executes the registration process of fluorescent image, the fluorescent image includes the image obtained by shooting the fluorescent signal generated after the chemical reaction of sequencing object; The registration process of the fluorescent image includes: Determine the normalized cross power spectrum matrix between the reference fluorescent image and the fluorescent image to be registered, and perform form conversion on the normalized cross power spectrum matrix to obtain a phase angle sequence;The phase angle sequence includes a plurality of phase angles, and the position information of a plurality of phase angles in the frequency domain coordinate system; Determine the slope information corresponding to each of the plurality of phase angles, and determine the weight value corresponding to each of the plurality of phase angles based on the slope information corresponding to each of the plurality of phase angles;The slope information corresponding to each of the phase angles includes the slope between each of the phase angles and at least part of the other phase angles in the phase angle sequence; Based on the weight value corresponding to each of the plurality of phase angles, the phase angle is linearly fitted to obtain the target slope of the straight line formed by the phase angle, and based on the target slope, the sub-pixel offset of the fluorescent image to be registered relative to the reference fluorescent image is determined; Based on the sub-pixel offset, the image registration is performed. The processor, when determining the slope information corresponding to each of the plurality of phase angles, is configured to:

2. The computer device according to claim 1, wherein: Traverse the plurality of phase angles in the phase angle sequence, and for the current phase angle traversed, execute the following determination process: From the phase angle sequence, determine the target phase angle corresponding to the current phase angle; The target phase angle corresponding to the current phase angle includes: the phase angle in the phase angle sequence except the current phase angle;Or, the adjacent region phase angle corresponding to the current phase angle, and the center region phase angle of the phase angle sequence;Determine the slope between the current phase angle and the corresponding target phase angle. The processor, when determining the weight value corresponding to each of the plurality of phase angles based on the slope information corresponding to each of the plurality of phase angles, is configured to:

3. The computer device of claim 1 or 2, wherein, Determine the slope standard deviation corresponding to each of the phase angles based on the slope between each of the phase angles and the other phase angles in the phase angle sequence except each of the phase angles;The slope standard deviation corresponding to each of the phase angles is negatively related to the first weight component of each of the phase angles; And / or, Determine the second weight component based on the first slope information of each of the phase angles relative to the adjacent region phase angle, and the second slope information of each of the phase angles relative to the center region phase angle;The difference degree of the first slope information and the second slope information is negatively related to the second weight component; Determine the weight value of each phase angle based on the first weight component and / or the second weight component. ​ 4. The computer device of claim 3, wherein, The processor is configured to: determine a first slope mean value based on the first slope of each of the phase angles relative to the phase angles of the adjacent regions; determine a second slope mean value based on the second slope of each of the phase angles relative to the phase angles of the central region; and determine a mean difference between the first slope mean value and the second slope mean value, and normalize the mean difference corresponding to each of the phase angles to obtain a second weight component corresponding to each of the phase angles. The processor is configured to:

5. The computer device of any of claims 1-4, wherein, perform windowing processing on the reference fluorescence image to obtain a windowed reference fluorescence image, and perform Fourier transform on the windowed reference fluorescence image to obtain a reference fluorescence image in a frequency domain; perform windowing processing on the to-be-registered fluorescence image to obtain a windowed to-be-registered fluorescence image, and perform Fourier transform on the windowed to-be-registered fluorescence image to obtain a to-be-registered fluorescence image in a frequency domain; and determine the normalized cross power spectrum matrix based on the reference fluorescence image in the frequency domain and the to-be-registered fluorescence image in the frequency domain. The processor is configured to: perform rank-1 matrix approximation on the normalized cross power spectrum matrix to obtain a maximum singular value vector of the normalized cross power spectrum matrix, wherein the maximum singular value vector includes a left singular value vector and a right singular value vector; 6. The computer device of claim 5, wherein, perform phase unwrapping on the left singular value vector and the right singular value vector respectively to obtain a first phase angle sequence corresponding to the left singular value vector and a second phase angle sequence corresponding to the right singular value vector. The processor is further configured to: filter each of the phase angles in the first phase angle sequence and the second phase angle sequence based on noise distribution characteristics and phase angle jump point distribution characteristics of each of the phase angles included in the first phase angle sequence and the second phase angle sequence, to obtain filtered first phase angle sequence and filtered second phase angle sequence.

7. The computer device of claim 6, wherein, The processor is configured to: perform importance sorting on the phase angles according to the weight values corresponding to the phase angles to obtain an importance sorting result; 8. The computer device of claim 1, wherein, perform linear fitting on the phase angles by using a target fitting method according to the importance sorting result to obtain a target slope of a straight line formed by the phase angles. The method comprises: ​ 9. A method of registering fluorescent images, characterized by, ​ determining a normalized cross power spectrum matrix between the reference fluorescent image and the fluorescent image to be registered, performing a form conversion on the normalized cross power spectrum matrix to obtain a phase angle sequence; the phase angle sequence includes a plurality of phase angles and position information of the plurality of phase angles in a frequency domain coordinate system; determining slope information corresponding to each of the plurality of phase angles, and determining weight values corresponding to the plurality of phase angles based on the slope information corresponding to the plurality of phase angles respectively; the slope information corresponding to each of the phase angles includes a slope between each of the phase angles and at least part of other phase angles in the phase angle sequence; performing linear fitting on the phase angles based on the weight values corresponding to the plurality of phase angles respectively to obtain a target slope of a straight line formed by the phase angles, and determining a sub-pixel offset of the fluorescent image to be registered relative to the reference fluorescent image based on the target slope; performing image registration based on the sub-pixel offset.

10. The method of claim 9, wherein, The determination of the slope information corresponding to each of the plurality of phase angles includes: traversing the plurality of phase angles in the phase angle sequence, and performing the following determination process for a current phase angle traversed: determining a target phase angle corresponding to the current phase angle from the phase angle sequence; the target phase angle corresponding to the current phase angle includes: phase angles in the phase angle sequence other than the current phase angle; or, phase angles in a neighboring region of the current phase angle and phase angles in a central region of the phase angle sequence; and determining a slope between the current phase angle and the target phase angle.

11. The method according to claim 9 or 10, characterized in that, The determination of the weight values corresponding to the plurality of phase angles based on the slope information corresponding to the plurality of phase angles respectively includes: determining, based on the slope between each of the phase angles and other phase angles in the phase angle sequence other than each of the phase angles, a slope standard deviation corresponding to each of the phase angles; performing normalization processing on the slope standard deviations corresponding to the plurality of phase angles respectively to obtain first weight components corresponding to the plurality of phase angles respectively; the slope standard deviation corresponding to each of the phase angles is negatively correlated with the first weight component of each of the phase angles; and / or determining second weight components based on first slope information of each of the phase angles relative to phase angles in the neighboring region and second slope information of each of the phase angles relative to phase angles in the central region; the difference degree of the first slope information and the second slope information is negatively correlated with the second weight components; determining the weight value of each of the phase angles based on the first weight component and / or the second weight component.

12. The method of claim 11, wherein, The determination of the second weight components based on the first slope of each of the phase angles relative to the phase angles in the neighboring region and the second slope of each of the phase angles relative to the phase angles in the central region includes: determining a first slope mean value based on the first slope of each of the phase angles relative to the phase angles in the neighboring region; and determining a second slope mean value based on the second slope of each of the phase angles relative to the phase angles in the central region. Determine the mean difference of the first slope mean value and the second slope mean value, and normalize the mean difference corresponding to each of the plurality of phase angles to obtain a second weight component corresponding to each of the plurality of phase angles.

13. The method according to any one of claims 9-12, characterized in that, The method further comprises: The method further comprises: The method further comprises: The method further comprises:

14. The method of claim 13, wherein, The method further comprises: The method further comprises: The method further comprises:

15. The method of claim 14, wherein, The computer readable storage medium stores a computer program, and when the computer program is run by a computer device, the computer device performs the steps of the registration method of the fluorescence image according to any one of claims 9 to 16. ​ 16. The method of claim 9, wherein, ​ ​ ​ 17. A computer-readable storage medium, characterized in that, ​

Images