A visual rapid screening method for microbial contamination of plant extracts

Abstract

This invention discloses a rapid visual screening method for microbial contamination in plant extracts, belonging to the field of rapid microbial detection technology. The method includes the following steps: First, the plant extract to be tested is introduced into a microfluidic colorimetric cartridge and mixed with a pre-placed ZIF-8@LPL core-shell probe to form a micro-confined liquid film. Then, a mobile terminal controls an external pulsed LED to flash and acquires the first frame image. Based on the single-column pixel array with the white border printed on the cartridge, a one-dimensional fast Fourier transform is performed to invert the single-row exposure readout time and the grating's initial phase. Subsequently, a phase-shifting drive command with a slight frequency offset is issued to control CMOS continuous shooting to acquire a sequence of frames. Further, pure dark-phase pixel rows during the LED off-state period and after delayed background attenuation are extracted, mapped to a unified holographic canvas according to physical absolute coordinates, and maximum projection reconstruction is performed to output the CFU equivalent concentration. This invention is applicable to the screening of dark-colored, high-viscosity, and high-polyphenol plant extracts.

Description

Technical Field

[0001] This invention relates to the field of rapid microbial detection technology, specifically a rapid visual screening method for microbial contamination in plant extracts. Background Technology

[0002] It is mainly used for production release, in-process sampling inspection, warehouse verification, and on-site rapid testing of samples such as extracts of traditional Chinese medicine, plant-derived functional beverages, food, and raw materials for health products. The existing testing process generally involves an optical report object, a solid carrier, and a portable terminal for reading. The device first contacts the surface of the report object or carrier, and then collects light excitation, delayed acquisition, and reads the image or light intensity for qualitative or quantitative detection.

[0003] Chinese patent document CN105705938A discloses a phosphorescent reporter. The technical concept of this phosphorescent reporter involves inorganic phosphorescent particles as the luminescent body, with an outer shell that further provides molecular recognition. The phosphorescent reporter can be used for the detection of various analytes. The document states that the detector can be placed on a porous membrane, a surface-immobilized carrier, a microfluidic chip, a paper-based microfluidic chip, a membrane, or a microplate. By providing a light source to excite the phosphorescent reporter, the reporter is excited. After excitation is stopped and a certain time is waited, the luminescence signal of the phosphorescent reporter is collected by light to determine the presence or content of the analyte. Readout can also be performed using portable electronic devices such as mobile phones with accessories. These accessories can shield ambient light, fix the test card, provide the excitation light source, and cooperate with a terminal camera to collect luminescence images.

[0004] While existing technologies can reduce background by using delayed luminescence in general analyte detection, they still have limitations in dark, viscous plant extracts rich in polyphenols. Firstly, plant extracts or food samples are complex matrices, and interfering substances within the matrix can affect absorbance measurements; furthermore, changes in optical path length significantly impact the results. Secondly, plant polyphenols have strong fluorescence quenching or interaction effects on optical sensing systems, making the optical and chemical environment surrounding the reporter analyte susceptible to influence. Thirdly, existing time-gated phosphorescence readouts require very accurate and short excitation-measurement delays. Literature also indicates that such time-resolved luminescence is not easily and stably implemented in the built-in optoelectronic systems of consumer-grade mobile phones and tablets. When these factors are considered in the screening of microbial contamination in plant extracts, problems such as insufficient background suppression, unstable image readout, decreased repeatability, and increased false negatives arise.

[0005] Therefore, the core technical problem faced by existing technologies is how to simultaneously achieve rapid visual screening of microbial contamination read by portable terminals under complex plant extract sample conditions. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a rapid visual screening method for microbial contamination in plant extracts. The method involves a mobile terminal controlling an external pulsed LED to flash and acquire the first frame image. A one-dimensional fast Fourier transform is performed on a single-column pixel array with a white border printed on a card, inverting the single-row exposure readout time and the grating's initial phase. Subsequently, a slightly frequency-biased out-of-phase drive command is issued to control CMOS continuous shooting to acquire a sequence of frames. Furthermore, rows of pure dark-phase pixels in the LED-off period and after delayed background attenuation are extracted, mapped to a unified holographic canvas using physical absolute coordinates, and subjected to maximum projection reconstruction to output the CFU equivalent concentration. This method solves the technical problems described in the background art.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: A rapid visual screening method for microbial contamination of plant extracts includes: introducing the original plant extract to be tested into a microfluidic colorimetric cartridge and mixing it with a pre-placed ZIF-8@LPL core-shell probe to form a micro-confined liquid film; controlling an external pulsed LED to flash and acquiring the first frame image via a mobile terminal; performing a one-dimensional fast Fourier transform on a single-column pixel array of the white edge printed on the cartridge to invert the single-line exposure reading time and the grating initial phase; Based on the single-line exposure readout time, an out-of-phase drive command with a slight frequency offset is sent to the external pulse LED to control the CMOS continuous shooting to obtain a sequence of frames carrying the physical space sliding slice coordinates; pure dark phase pixel rows that are in the LED off period and have delayed background decay are extracted from the sequence of frames, mapped to a unified holographic canvas according to physical absolute coordinates and maximum value projection reconstruction is performed to output CFU equivalent concentration.

[0008] Furthermore, the microfluidic colorimetric cartridge includes a detection port, a flow channel capillary injection channel, a mixing and release chamber, a detection window, a tail physical airbag, and an exhaust end connected in sequence. The plant extract to be tested enters the detection window through the flow channel capillary injection channel and is flattened into a micro-confined liquid film. When capillary stagnation occurs, the tail physical airbag provides micro-positive pressure to push it forward.

[0009] Furthermore, the ZIF-8@LPL core-shell probe is pre-placed in the transition area between the mixing and release chamber and the detection window, and is fixed in the form of a dry film layer. After the plant extract solution enters, it first wets and releases the ZIF-8@LPL core-shell probe, and then forms a uniformly dispersed state in the detection window before participating in subsequent image acquisition.

[0010] Furthermore, when the mobile terminal captures the first frame image, it outputs the original frame and locks the exposure parameters, focal length parameters, and white balance parameters, and turns off the automatic anti-flicker processing chain; the single-column pixel array is formed by the continuous narrow band area of ​​the white edge of the card box after column mean synthesis and mean removal, and then a one-dimensional fast Fourier transform is performed.

[0011] Furthermore, when the main peak identification of the one-dimensional fast Fourier transform does not meet the preset signal-to-noise conditions, the mobile terminal sends a wideband sweep handshake signal to the external pulse LED, collects stripe images for different candidate driving frequencies, and determines the phase-locked drive frequency and the start delay based on the prominence of the main peak and the phase continuity.

[0012] Furthermore, the mobile terminal sends out out-of-phase drive commands with slight frequency offsets to the external pulse LED based on the phase-locked drive frequency, and keeps the exposure parameters, focal length parameters, and white balance parameters unchanged during continuous acquisition; each frame in the sequence records the frame number, grating start phase, and physical space sliding slice coordinates.

[0013] Furthermore, the mobile terminal combines the positioning features of the gyroscope output and the white edge of the card box printing to perform inter-frame affine registration. When the inter-frame matching error or attitude disturbance exceeds the preset threshold, the current continuous shooting buffer is discarded, and the acquisition of the first frame image, the inversion of the single-line exposure reading time, and the inversion of the grating start phase are re-executed.

[0014] Furthermore, the mobile terminal determines the sampling time of each row of pixels in the sequence frame based on the single-line exposure reading time, the starting delay amount, the grating starting phase and the slight frequency offset, retains only the pure dark phase pixel rows that are in the LED off period and after the delayed background decay, and sets the pixel rows that do not meet the conditions to zero.

[0015] Furthermore, the mobile terminal projects the pure dark phase pixel rows onto a unified holographic canvas based on the physical space sliding slice coordinates and the row coordinates of each pure dark phase pixel row; when multiple pure dark phase pixel rows from different sequence frames are mapped to the same canvas coordinates, the pixel values ​​with higher brightness are retained, and the canvas coordinates that are not covered remain blank.

[0016] Furthermore, the mobile terminal performs threshold separation and connected component labeling on the reconstructed unified holographic canvas, removes connected components located in the edge area of ​​the card box and isolated noise points that do not meet the preset area conditions, and outputs the CFU equivalent concentration based on the number of retained connected components.

[0017] (III) Beneficial Effects This invention provides a rapid visual screening method for microbial contamination in plant extracts, which has the following beneficial effects: The plant extract to be tested is liquid-pressed into a micro-confined liquid film using a microfluidic colorimetric cartridge, which then contacts a pre-installed zIF-8@LPL core-shell probe within the confined space. This allows high-absorbance, high-viscosity, and high-polyphenol samples to be preferentially connected to the corresponding detection system, avoiding the impact of sample morphology fluctuations on subsequent visual acquisition. A single-column pixel array is formed by printing white edges on the cartridge, and a one-dimensional fast Fourier transform is performed to directly invert the single-line exposure time and grating initial phase from the first frame image. The mobile terminal and the external pulsed LED form a soft phase-locked reference, avoiding the impact of terminal differences on the uniformity of the screening process. A small frequency offset is injected into the external pulsed LED, changing the rolling stripe from a fixed occlusion state to a continuous sliding slice state. Sequence frames obtain different spatial sampling positions within the same acquisition window, transforming temporal sampling opportunities into an accumulative spatial coverage process.

[0018] By extracting only the pure dark-phase pixel rows after the LED extinction period and background decay, the background of the plant extract itself and the long-persistence probe response are temporally separated, ensuring that the data entering the unified holographic canvas originates from the same source and preventing outdated data from being mixed into subsequent reconstructions. The pure dark-phase pixel rows are mapped to the unified holographic canvas according to physical absolute coordinates. Multi-frame data at the same coordinates are subjected to maximum projection, and multiple frames are realistically swept into a continuous dark-phase map, maintaining the coordinate relationship between the sequence frames, which facilitates the subsequent determination of connected components and clear object boundaries. Attached Figure Description

[0019] Figure 1 This is a diagram illustrating the overall collaborative architecture of the visual rapid screening system for microbial contamination in plant extracts according to the present invention. Figure 2 This is a schematic diagram of the structure and unidirectional liquid path of the microfluidic colorimetric card box described in this invention; Figure 3 This is a schematic diagram illustrating the principle of dual insulation shielding and micro-confined liquid film formation described in this invention. Figure 4 This is a schematic diagram of the pressure compensation, correction, and detection window filling and shaping state described in this invention; Figure 5 This is a flowchart of the spatial stripe inversion and soft phase-locked loop based on the white edge of card box printing as described in this invention; Figure 6 This is a schematic diagram of the spatiotemporal slice formation of rolling stripes under micro-frequency offset driving as described in this invention; Figure 7 This is a flowchart of the pure dark phase pixel row extraction, physical coordinate projection, and interpolation-free reconstruction output process described in this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figures 1-7 This invention provides a rapid visual screening method for microbial contamination in plant extracts, comprising: Step 1: The original extract of the plant to be tested is transformed into a micro-confined liquid film with light transmittance, anti-polyphenol quenching ability and continuous spreading morphology without decolorization, centrifugation and magnetic bead elution. This provides a stable, single and repeatable detection starting point for subsequent heterogeneous terminal imaging synchronization and dark phase reconstruction.

[0022] Plant extract stock solutions differ from conventional culture media in that they are dark in color, have high viscosity resistance, and high polyphenol loading. These three properties can superimpose to form a tandem failure chain within the same detection window. If only the liquid layer thickness is shortened, although bulk absorption is reduced, polyphenols will still approach the luminescent core and trigger photoinduced electron transfer quenching. If only the probe coating is strengthened, the probe will not be poisoned, but the entire liquid layer will still prevent effective luminescence from passing through the sample due to excessive light absorption. If only capillary injection is relied upon, non-Newtonian flow will stagnate at the leading edge of the detection window, eventually forming bands of varying thickness.

[0023] Based on this field constraint, this step integrates thinning, sieving, and shaping into a single processing chain, and collectively refers to the resulting state as double insulation shielding. The effect of thinning in weakening bulk absorption is called optical path insulation, and the effect of sieving in blocking polyphenols from approaching the luminescent core is called quenching insulation. Only when both are established simultaneously within the same detection window can subsequent long afterglow readout have a feasible basis.

[0024] The system consists of a sample inlet, a leading capillary channel, a mixing and release chamber, a detection window, a rear physical gasbag, and an exhaust terminal, all connected in series to form a unidirectional liquid path. After the plant extract is dripped into the sample inlet, the leading capillary channel uses surface energy difference to draw the liquid into the cartridge. Then, in the mixing and release chamber, the previously fixed ZIF-8@LPL core-shell probe dry film layer is wetted. Subsequently, it enters the detection window, whose thickness is limited by a spacer layer, where it is flattened into a continuous liquid layer. When the leading edge of the liquid layer reaches the exhaust terminal, excess gas is expelled, leaving a uniform, micro-confined liquid film within the detection window. The liquid sequentially completes four actions—mixing, sieving, thickness limiting, and pressure replenishment—along its path, avoiding subsequent steps that deal with non-uniform and unstable imaging targets.

[0025] The core-shell probe refers to a core-shell structure probe comprising a ZIF-8 mesoporous shell and a long-afterglow luminescent core; for the sake of brevity, it will be referred to as the ZIF-8@LPL core-shell probe in the following text.

[0026] The microfluidic colorimetric cassette preferably employs a sandwich structure consisting of an upper light-transmitting cover, a lower supporting substrate, and a spacer layer sandwiched between them. The upper light-transmitting cover is made of cyclic olefin copolymer, quartz glass, or a similar light-transmitting material; the lower supporting substrate is made of black polymer board, black-coated glass, or a similar low-reflection material; and the spacer layer is made of polyester film or polyimide film, the thickness of which is the optical path thickness. .

[0027] The detection window is not a simple container, but a confined flat plate cavity. Once the liquid enters this cavity, it is forcibly flattened by the upper and lower walls, causing the dark-colored plant matrix to exist as a thin liquid film rather than a thick liquid column. To ensure that the transmission margin after entering the terminal field of view is not completely consumed by plant pigments, the detection window is calibrated using the following formula: ; Where: transmission margin Detection window at wavelength The remaining luminous flux margin relative to the lowest identifiable transmission threshold of the terminal, taking a real number. The time indicates that the band can be used for further acquisition; absorption coefficient The plant extract to be tested at a wavelength The equivalent absorption intensity at the point is a positive real number; the wavelength is pre-scanned. The transmittance or absorbance at that point is calculated by combining the calibration curve of the blank matrix. Chromium equivalent concentration : The effective chromophore concentration causing bulk absorption, taken as a positive real number; the matrix transmittance obtained from the pre-scan after sample introduction, converted according to a pre-established plant matrix mapping model; optical path thickness. To detect the restricted distance between the upper and lower walls of the window, the value range is: The preferred value range is Determined by the thickness of the spacer layer, and confirmed by a contact thickness gauge, white light interferometer, or manufacturing specifications; minimum transmission threshold. The lower limit of the difference between dark areas and the background that the mobile terminal imaging device can still distinguish, and the value satisfies ;wavelength : The target spectral position used for subsequent excitation and readout, with a value of a positive real number; In engineering, absorption coefficient The chromophore equivalent concentration is provided by the matrix calibration library of the plant extract category. The lowest transmission threshold is obtained from a single pre-scan before sample introduction. The optical path thickness is given in the mobile terminal camera calibration table. The reason for locking it within the above range is because Continuing to increase the size will amplify the plant's own absorption, and Further reduction will significantly increase the injection resistance and amplify incomplete coverage defects.

[0028] In one embodiment, the operator places a drop of the plant extract to be tested at the injection inlet. The liquid first wets the lead capillary channel, then passes through the mixing and release chamber, and finally spreads rapidly within the detection window. At this point, when viewed from the front of the cartridge, the previously nearly opaque dark droplet no longer appears as a bulge, but rather as a continuous thin layer. No thick-edged liquid accumulation appears at the edge of the detection window, and no visible voids appear in the center. For plant extracts resembling donkey-hide gelatin, extracts, or those containing colloidal fibers, the cross-section of the lead capillary channel can be changed from rectangular to a tapered section that is wider at the front and narrower at the back, allowing the liquid front to undergo a stretching and rectification process before entering the detection window.

[0029] Optical path insulation alone cannot guarantee luminescence of the long-afterglow probe, because polyphenol aggregates and polyphenol-colloidal complexes, pigment aggregates and some oxidation byproducts in the plant extract will still directly approach the luminescent core and induce non-radiative inactivation.

[0030] For this purpose, a ZIF-8@LPL core-shell probe is used, in which the luminescent core is responsible for releasing a long afterglow signal in the subsequent dark-field stage, and the ZIF-8 mesoporous shell is responsible for establishing molecular size gating. This core-shell probe is preferably fixed in the transition area between the mixing release chamber and the detection window as one of the following: a dry film layer, a dot matrix coating layer, or a pre-dispersed particle layer. The plant extract to be tested is wetted and released after entering the probe.

[0031] To allow small bacterial metabolites to penetrate the mesoporous shell and reach the luminescent core, while blocking larger polyphenol molecules with larger kinetic diameters, the following penetration gating relationship is adopted: ; Where: Penetration probability Components The probability of entering the neighborhood of the luminescent core through the ZIF-8 mesoporous shell ranges from [value missing]. Gating steepness The degree to which mesoporous shells amplify dimensional differences, taken as a positive real number: dynamic diameter. Components The equivalent size in the solution migration state is a positive real number. Mesopore diameter The effective pore diameter of the ZIF-8 mesoporous shell, with a range of values ​​being: : Obtained by nitrogen adsorption-desorption testing combined with NLDFT or BJH methods, and stored as probe factory parameters; Component index : The number of the migrating component being analyzed, with a value of a positive integer; In engineering, this gating relationship does not require online solving for every molecule, but is used to determine the correspondence between shell pore size and sample type: for plant extracts with known polyphenol spectra, a component size library is pre-established; for unknown categories, a conservative mesopore size is selected based on the size of the largest polyphenol component in the sample. When the kinetic diameter corresponding to bacterial small molecule metabolites Smaller than mesopore diameter At that time, the probability of penetration Approaching 1; when the kinetic diameter corresponding to the plant polyphenol macromolecule is close to 1. When the wavelength is greater than 1.8nm, the penetration probability The aperture is rapidly reduced to near zero. This allows the question of whether molecules are allowed to approach the luminescent core to be specifically addressed through aperture gating, enabling molecules that can be excited and recognized to enter, while preventing molecules that can cause quenching from entering.

[0032] In one embodiment, the ZIF-8@LPL core-shell probe is initially laid dry at the end of the mixing and release chamber. Upon arrival of the front of the plant extract to be tested, the dry film layer is wetted and detaches from the substrate, subsequently forming a uniform dispersion within the detection window. At this point, no visible coarse particles settle within the cartridge, nor do they aggregate into bright clumps near the entrance of the detection window. For plant extracts with high polysaccharide content and strong adhesion, the core-shell probe can be arranged in a dual-zone configuration instead of a single-layer dry film: a pre-release layer is placed near the mixing and release chamber, and a compensation layer is placed near the center of the detection window. The former ensures initial dispersion, while the latter ensures that new probes are added after the liquid front reaches the center of the detection window, thus avoiding a concentration gradient that is concentrated at the entrance and dilute at the end. For plant extracts with particularly high polyphenol loading, core-shell probe particles with more continuous shells are preferred. The above alternative paths change the probe arrangement but do not alter the principle of double insulation and shielding.

[0033] Plant extracts often exhibit shear-thinning or localized yielding properties. Relying solely on the guide capillary channel, the liquid front may suddenly decelerate in the high-viscosity section, even stopping in the middle of the detection window, resulting in simultaneous areas of full liquid, thin liquid, and empty windows within the same field of view. To prevent subsequent steps from mistaking this geometric defect for a dark phase feature, this step incorporates a physical airbag at the tail end, using the completeness of the liquid coverage as the trigger criterion. ; In the formula: completion rate The volume fraction of the detection window actually occupied by liquid, with a value range of [value missing]. ; Filled volume : The volume of liquid that enters the detection window and remains stable, a non-negative real number; Detection window width : Horizontal dimension of the detection window, taken as a positive real number; Length of the detection window : Vertical dimension of the detection window, taken as a positive real number; optical path thickness Consistent with the previous definition, the range of values ​​remains unchanged; In engineering terms, the volume already filled It can be calculated from the volume corresponding to the engraving lines on the card case, or by multiplying the area of ​​the wetted window region in the mobile terminal preview image by the optical path thickness mentioned earlier. Obtained. Locally measured completion rate. At this time, the operator presses the physical airbag, which applies one or more micro-positive pressures to the rear of the detection window, continuing to push the stagnant liquid front to the exhaust end; the degree of complete coverage is measured locally. When the pressure is reached, it indicates that the detection window has been continuously filled and no further pressurization is needed. Using a physical airbag instead of an external pump is to maintain the cartridge's one-time sealed configuration.

[0034] Preferably, a microbial localization capture layer is pre-formed on the bottom surface of the detection window. This microbial localization capture layer is selected from one of the following: a poly-L-lysine layer, a chitosan layer, a hydrophilic thin layer with quaternary ammonium groups, or a surface structure with a microporous retention array. After the plant extract to be tested enters the detection window and forms a micro-confined liquid film, it is initially maintained for a preset retention time. This allows microbial cells to form dispersed residence sites on the bottom surface of the detection window. The uniformly dispersed ZIF-8@LPL core-shell probe then senses changes in the local metabolic microenvironment around these residence sites, thereby forming spatially separated bright spot regions in the subsequent reconstructed map, providing a spatial basis for connected domain counting.

[0035] Duration of stay The term refers to the waiting time from the formation of the liquid film to the start of sequential data acquisition; its purpose is to allow microbial cells to complete their residence on the basal surface; it is obtained through pre-experimental calibration, generally by testing standard bacterial suspensions within the same cartridge. The effective pore size of the ZIF-8 mesoporous shell is preferably limited to 0.8 to 1.5 nm. This shell creates size exclusion for polyphenol aggregates and polyphenol-colloidal complexes with a hydration kinetic diameter greater than 1.8 nm, making it difficult for them to enter the vicinity of the luminescent core; for free phenolic components with a hydration kinetic diameter not greater than 1.8 nm, their diffusion rate is limited and the subsequent dark-phase acquisition window is time-gated, preventing them from constituting a dominant PET quenching path during the sampling period.

[0036] In one embodiment, the operator observes the liquid front stopping at one-third of the way down the detection window. After pressing the tail-end physical airbag once, the liquid front continues to advance, expelling residual gas at the end, until the entire detection window is continuously filled without any broken bubbles. If the sample is a high-colloidal plant extract, the tail-end physical airbag can be used in conjunction with a one-way check valve to prevent gas from rebounding after each pressurization. If the sample is a foaming plant extract, a microporous venting membrane that is gas-repellent but not liquid-repellent is preferably installed at the venting end to release air first and then prevent liquid leakage. Through this pressurization and correction action, the detection window outputs a micro-confined liquid film with complete boundaries, limited thickness, and continuous probe dispersion. Once this liquid film is formed, the optical path thickness described above... Mesopore diameter and completeness They together become the call boundary for the subsequent step two: step two only needs to deal with a single geometric thickness, a stable background, and a transparent window area, and no longer undertakes the task of preprocessing defect repair.

[0037] In use, the first step transforms the plant extract from a complex bulk sample—dark in color, highly viscous, and containing polyphenols—into a micro-confined liquid film with limited thickness, low quenching risk, and sustainable spreading. Among these features is the optical path thickness. Responsible for compressing the bulk material into the readable region of the terminal, mesopore diameter Responsible for separating identifiable molecules from quenching molecules, achieving full completion. This ensures a consistent geometric shape across the entire detection window. These three elements are not simply stacked together, but rather locked in a sequential relationship: without a restricted thickness, subsequent imaging lacks effective transmission; without mesoporous exclusion, the long-persistence probe will be poisoned by polyphenols in the near-field within the liquid film; without pressure compensation and shaping, geometric defects within the same detection window will directly disrupt subsequent fringe inversion and dark phase mapping. Therefore, the micro-confined liquid film output from step one serves as a chemically quenching-resistant carrier, an optically transparent window, and a unified imaging substrate relied upon for edge-to-pixel frequency band inversion in step two.

[0038] Step 2: Use the spatial stripes in the white border of the card box to invert the line-by-line reading rhythm of the mobile terminal image sensor, and establish a soft phase-locked reference for the external pulse LED controller accordingly.

[0039] Step one has already leveled the raw material of the plant extract to be tested into a micro-confined liquid film. However, if the reading cycle of the mobile terminal is inconsistent with the emission cycle of the external pulse LED controller, the dark phase position in the subsequent image will still drift on different terminals, causing step three to fail to convert the scrolling stripes into controllable spatial slices. Traditional methods usually rely on dedicated phase-locked loop hardware, while rapid screening of plant extracts faces heterogeneous mobile terminals and processor scheduling states that change with temperature rise.

[0040] Therefore, the first frame image is directly used as the inversion object: as long as the white border of the card box printing enters the field of view, the fringe spatial period will carry the information of the line-by-line reading beat, and the fringe complex phase will carry the information of the initial alignment position. Thus, the output of step two is... and The image constitutives are not attached to the equipment model list, but are directly derived from the shooting scene.

[0041] The following actions are executed collaboratively by the mobile terminal and the external pulse LED controller. The mobile terminal first secures the microfluidic colorimetric cassette to the clamp in front of the lens, ensuring the printed white edge of the cassette falls within the edge of the field of view. Then, it sends the first nominal drive frequency to the external pulse LED controller. The external pulse LED controller receives the driving frequency and generates pulsed light. The mobile terminal's image sensor acquires the first frame of the original image in fixed exposure, gain, and progressive scan mode. Instead of invoking the system's automatic anti-flicker process, the mobile terminal extracts a single-column pixel array of the printed white border from the original frame buffer. It first suppresses the DC background, then performs a one-dimensional fast Fourier transform, recovers the spatial period from the main peak position, recovers the starting phase from the main peak phase, and then... and Write-back serves as a common constraint for subsequent data acquisition and LED driving.

[0042] In the first frame of the original image, the mobile terminal selects a continuous reference narrow band along the white edge of the card box printing. The reference narrow band is preferably arranged outside the detection window and parallel to the long side of the detection window, and its width is preferably two to eight columns of pixels.

[0043] Subsequently, the mobile terminal calculates the average value column by column to obtain a single-column equivalent brightness sequence, and subtracts the sequence mean to remove the slowly varying background introduced by uneven reflection of white edge ink, forming the net stripe sequence used in subsequent transformations: ; Where: net stripe sequence : In row coordinates The DC-free brightness value extracted from the reference narrowband serves to separate spatial fringes from the white background; its value is a real number. (Original brightness) The first frame of the original image is in row coordinates. Column coordinates The pixel grayscale value at a given location serves as the base sampling value for the reference narrowband, and its value is a non-negative real number; the reference column number... The value is a positive integer, referring to the number of pixel columns contained in the narrow band. Its function is to balance bad pixel suppression and phase purity. ; Image height : The total number of rows in the first frame of the original image; its function is to limit the length of the discrete transform; its value is a positive integer; row coordinates. Discrete positions along the progressive scan direction, serving to carry spatial periodic information, taking integer values, and satisfying the following conditions: Column coordinates : Referring to the discrete column positions within the narrow band, its function is to synthesize the average of adjacent columns, with the value being an integer; After obtaining the net stripe sequence, the mobile terminal performs a one-dimensional fast Fourier transform on it and searches for the main energy peak index in the positive spectrum. Because line-by-line scanning maps temporal sampling to spatial sampling, the main peak index... There is a one-to-one correspondence between the line-by-line readout time and the single-line exposure readout time. Inversion using the following formula: ; Where: Single-line exposure readout time The actual time interval between two adjacent rows read by the current mobile terminal image sensor serves as the basis time stamp for subsequent soft phase-locked loop and out-of-phase micro-frequency offset modulation; its value is a positive real number. (Main peak index) The position of the main energy peak in the positive spectrum of the net fringe sequence after one-dimensional fast Fourier transform, which represents the spatial period of the fringe, and takes a positive integer value; image height. Consistent with the previous definition, the value remains unchanged; Nominal drive frequency The pulse frequency received and executed by the external pulse LED controller during the first frame handshake phase serves to establish a conversion bridge between the spatial main peak and the temporal beat, and its value is a positive real number. In one embodiment, the operator attaches a clamp containing a microfluidic colorimetric card cartridge to the front of the mobile terminal's lens, with the white border printed on the cartridge located on the left side of the image. The mobile terminal only captures four columns of pixels from the white border, performs averaging, background removal, and transformation, and then obtains the main peak index. It can be read out without touching the pixels in the sample area or relying on a terminal model database. It transforms the unknown device cycle time into measurable white-edged stripes and isolates the interference of color changes in the plant extract on the inversion process from the sampling location.

[0044] Only single-line exposure read time This is not enough to guarantee that subsequent stripes will start from the same position, because there is still an unknown phase shift between the starting point of the LED pulse and the starting point of the first line of reading from the image sensor when the first frame is acquired.

[0045] Therefore, the mobile terminal obtains the main peak index. Then, in addition to reading the amplitude of the main peak, the complex spectral value of the main peak was also read. The argument is used to recover the grating start phase of the first frame. : ; Where: grating initial phase The phase difference between the top reference line of the first frame and the emission cycle of the external pulsed LED controller determines which line the subsequent dark fringes begin to enter the field of view; its value is a real number in radians. Complex spectral value. Net stripe sequence in main peak index The one-dimensional Fast Fourier complex result at that point serves to provide both amplitude and phase information simultaneously. Imaginary part operator The imaginary part operator is used to extract the components of the complex spectral value along the imaginary axis; the real part operator... The function of extracting the real part of a complex number is to extract the components of the complex spectrum along the real axis; the arctangent operator. The arctangent operation with quadrant determination restores the real and imaginary parts to a unique phase angle, with a value range of [value missing]. Main Peak Index Consistent with the previous definition, the value remains unchanged; Subsequently, the mobile terminal reads the data using a single-line exposure time. and grating start phase As input, the corrected phase-locked drive frequency is written back to the external pulse LED controller. With initial delay Soft phase-locked loop (PLL) does not establish a simulated PLL in hardware. Instead, the mobile terminal resends a set of digital drive parameters based on the current image after each pre-detection handshake.

[0046] Preferably, the external pulse LED controller adopts a structure in which a microcontroller, a crystal oscillator, and a constant current drive stage are connected in series. The microcontroller receives the phase-locked drive frequency sent by the mobile terminal. With initial delay Then, wait for the initial delay. Then press the phase-locked drive frequency The output pulse sequence has a fixed duty cycle; the mobile terminal then saves this set of parameters into the frame header configuration of this screening session. Therefore, if the same mobile terminal changes the image sensor readout rate, it only needs to reacquire the first frame and recalculate the single-line exposure readout time. and grating start phase The soft phase-locked reference will then be re-established. In one embodiment, after the mobile terminal completes the first frame inversion, the phase-locked drive frequency is... and initial delay The signal is sent to the external pulse LED controller. Upon triggering the preview again, the white border stripe position remains fixed within the same group of lines. If a different mobile terminal is used, the operator does not need to modify the card holder; simply re-execute the first frame handshake, and the new single-line exposure reading time will be used. and grating start phase It will overwrite old values. This technique directly transforms the measurement results obtained from the one-dimensional fast Fourier transform into driving constraints, thereby completing the closed loop from passive observation to active alignment.

[0047] When the white border of the printed card case is obscured by stains, there is condensation on the front surface of the lens, or some mobile devices perform non-disabled light smoothing on the raw frame, the main peak index... The energy will be weakened, and it is difficult to determine the stable main peak by a single one-dimensional fast Fourier transform. At this time, the mobile terminal enters the frequency sweeping degradation path: first, it sends a continuous sequence of candidate frequencies to the external pulse LED controller, and each candidate frequency corresponds to a frame of white-edge image. Then, it calculates the main peak amplitude and phase coherence for each frame. Finally, it selects the candidate frequency with the highest phase coherence as the new phase-locked drive frequency. The determination of phase coherence does not depend on the sample area, but only compares whether the complex directions of the white edge fringes in the neighborhood of the main peak remain concentrated. Therefore, even if the white edge brightness is uneven, as long as the fringes still exist, the degradation path can still complete the alignment.

[0048] In one embodiment, if the operator finds that the white border stripes in the first frame are too faint, the mobile terminal will continuously send multiple sets of candidate frequencies. The external pulse LED controller will output pulses sequentially, and the spacing of the white border stripes will change frame by frame with the candidate frequencies. After sorting the results of each frame, the mobile terminal will select the frame with the most regular stripes as the alignment reference and write back the corresponding frequency as the phase-locked loop drive frequency. If the white border of the card box printing is located on the right, top, or bottom of the image, the extraction direction of the narrow band should be adjusted accordingly. If the external light source is not a single LED, but a ring LED array, a linear LED array, or an equivalent pulse lighting component, it should be driven uniformly by the same control channel and executed according to the same soft-locking chain.

[0049] When in use, the striped spatial structure in the white border of the card box printing is converted into a single-line exposure reading time. and grating start phase This is further converted into a phase-locked drive frequency that can be executed by an external pulse LED controller. With initial delay The subsequent step three uses the aligned stripe starting points and known line-by-line timescales as a basis. Since the reference narrow band is taken only from the white edge and not from inside the detection window, step two will not mistakenly introduce color fluctuations of the plant extract into the inversion chain.

[0050] Step 3: Under the known single-line exposure reading time and grating start phase Under the premise of injecting controlled micro-frequency deviation into an external pulse LED controller The progressive scan stripes are transformed into a spatiotemporal slice sequence with a defined drift direction, drift step size, and inter-frame coordinates.

[0051] If, after step two is completed, the external pulse LED controller still strictly follows the phase-locked drive frequency... When light is emitted, the dark stripes in each frame will remain stably in approximately the same row area, and the bright blind areas in the detection window that are not covered by the dark phase will also remain in approximately the same position. Even if extreme value mapping is performed in step four, only the same batch of pixels can be used repeatedly, and the entire micro-confined liquid film cannot be completely scanned. Conversely, if the emission frequency is changed arbitrarily, although the dark stripes will move, the amount of movement is unpredictable, and the inter-frame coordinates cannot strictly correspond to the physical positions in the micro-confined liquid film, ultimately mixing the real dark phase with random rolling.

[0052] Based on this contradiction, this step neither pursues absolute stillness nor allows free drift, but rather injects only a small frequency offset with limited amplitude on top of the soft phase-locked reference already established in step two. This allows the stripes to slide frame by frame according to a calculable step size, thereby replacing the temporal phase shift with spatial slicing.

[0053] The following actions are executed collaboratively by the mobile terminal, the external pulse LED controller, the lens clamp, and the mobile terminal's built-in gyroscope. The mobile terminal first reads the output from step two. Grating start phase Phase-locked drive frequency With initial delay Then, based on the target number of frames collected this time and current camera frame rate Calculate the frequency offset injection window. Then, the mobile terminal sends an out-of-phase drive command to the external pulse LED controller, causing it to... It emits pulsed light; simultaneously, the mobile terminal locks the exposure time, lens focal length, white balance, and gain, and continuously acquires data. The system generates the original image frame by frame, writing a sequence number, starting phase, and slice coordinates to each frame. The gyroscope continuously outputs attitude change data during continuous shooting. If the attitude change exceeds the reconfigurable range, the entire current sequence is discarded, and erroneous frames are not included in step four. With this setup, step three outputs not just the image itself, but a valid image sequence labeled with stripe drift trajectories.

[0054] Furthermore, the mobile terminal adjusts the current camera frame rate. Target frame count The phase-locked drive frequency given in step two Establish a frequency offset window and define the out-of-phase drive frequency as The frequency offset window is constrained by the following formula: ; Where: minute frequency offset : Relative to the phase-locked drive frequency The artificially injected controlled mismatch acts to drive a calculable drift in dark stripes between consecutive frames, and its value is a non-zero real number; camera frame rate. The number of image frames acquired per unit time by the mobile terminal in this round of continuous shooting; its function is to provide the frame-level slice density that can be accommodated per unit time; the value is a positive real number; target frame number. The total number of image frames to be captured in this round of continuous shooting determines the number of slices to be decomposed in one scan, and the value is a positive integer greater than 1. Based on this, the mobile terminal sets the out-of-phase drive frequency to and the initial delay amount The drive duty cycle, pulse width, and frame start signal are sent together to the external pulse LED controller. The external pulse LED controller preferably employs a series structure of a microcontroller, crystal oscillator, constant current driver stage, and external LED array. After receiving the parameters from the mobile terminal, the microcontroller first waits for the start delay. Then according to the out-of-phase drive frequency Output pulse sequence. The reason for not directly modifying the line-by-line readout rhythm of the mobile terminal's image sensor, but instead choosing to modify the external light source's rhythm, is that the internal clock of the mobile terminal's camera is more tightly encapsulated by the operating system, while the external pulse LED controller is easier to drive precisely. Therefore, the soft phase-locked loop reference established in step two is still retained, but a phase drift source with a very small amplitude and a clear direction is superimposed around this reference.

[0055] In one embodiment, the operator secures the microfluidic colorimeter cartridge in the lens clamp, and the mobile terminal reads the data saved from the previous first-frame handshake. and Then, a pulse LED controller with a slight frequency deviation is sent to it. The out-of-phase drive command. In the preview, the dark stripes that were originally stationary began to move slowly in a stable direction, and the stripe displacement between adjacent frames remained consistent. The operator did not need to remove the cartridge or refocus; the stripes were transformed from static occlusions into sliding slices with just one parameter command.

[0056] Simply moving the stripes is insufficient for step four, as step four requires knowing the physical location within the micro-confined liquid film corresponding to each row of pure dark pixels in a frame. Therefore, the inter-frame drift step size is calculated synchronously during continuous shooting, and the step size is accumulated into slice coordinates. For the first... For a given frame, its stripe drift relative to the first frame is defined as: ; Relative to the phase-locked drive frequency Artificially injected micro-mismatch frequency offset Its purpose is to generate controllable scroll drift, which is set by the mobile terminal according to the scanning direction and coverage requirements; the actual frame rate during continuous shooting by the mobile terminal. Step 3: Incorporate the time advance of each frame into the displacement model, obtained from the timestamp returned by the camera drive, or from the average of the actual inter-frame times. Step 3: The out-of-phase drive frequency actually sent to the external pulse LED controller. Convert the phase drift into line space displacement, from get.

[0057] Frame-level drift : No. The equivalent displacement of the dark stripes in the frame relative to the first frame in the line-by-line direction maps the temporal mismatch to the spatial slice position in the image, and its value is a real number; frame number. The sequential number of the current image in this round of continuous shooting describes the chronological relationship of consecutive slices on the timeline. It takes an integer value and satisfies... Slight frequency deviation Consistent with the previous definition, the value remains unchanged; single-line exposure readout time The actual readout interval between two adjacent rows of the image sensor in a mobile terminal is used to convert the frequency difference into a sampling displacement in the row-by-row direction, and its value is a positive real number. when When the image row number increases, the stripes move in the direction that the image row number increases; when When the image row number decreases, the stripes move in the direction that the image row number decreases. Otherwise, the up / down scrolling drift only contains text and has no mathematical correspondence.

[0058] The mobile terminal uses this to calculate the frame-level drift for each frame. This information is then written into the metadata area of ​​the corresponding frame. Subsequently, the mobile terminal extracts the grating start phase from each frame according to the information obtained in step two. Locate the center line of the dark stripes in this frame, and then combine it with the frame-level drift. Determine the slice coordinates for this frame. To reduce the interference of occasional bright spots in the plant extract on the determination of the fringe center, the centerline is preferably located by searching for local valley values ​​after summing along the reference narrow band. If a valley value persists within three adjacent rows, it is considered a valid dark fringe center. After this registration, each frame is no longer just an image file, but a combined data object of image content and slice coordinates.

[0059] In one embodiment, after the mobile terminal completes the acquisition of the first frame, it immediately records its slice coordinates as the reference coordinates; the second and third frames are based on the frame-level drift. New slice coordinates are written sequentially. The operator sees dark stripes gradually moving downwards on the screen, while the mobile terminal internally synchronously registers the frame number, stripe center line, and slice coordinate band as a group. For vertically mounted mobile terminals, the line-by-line direction remains parallel to the long side of the detection window; for horizontally mounted mobile terminals, the mobile terminal performs coordinate system rotation registration before acquisition to reduce frame-level drift. It always corresponds to the uniform long side direction of the detection window. This parallel path changes the coordinate mapping, but does not change the principle of slice generation.

[0060] Rapid screening of plant extracts is often performed using handheld devices. If the lens clamp experiences significant shaking during continuous shooting, the movement of dark stripes in the image will no longer be solely determined by... This can also cause overall affine displacement. To prevent misjudging the displacement caused by hand tremors in step four as a true slice displacement, an attitude stability criterion is introduced: ; Where: attitude disturbance amount : No. The overall attitude change intensity relative to the previous frame is used to determine whether the current frame is still within the refittable range; its value is a non-negative real number. Angle change amount. : No. The equivalent rotation angle changes of the mobile terminal around the lens optical axis and horizontal and vertical axes during frame acquisition are used to characterize rotational disturbances and are taken as real numbers; the translation change vector... : No. The equivalent translation component between the frame and the previous frame on the image plane represents the overall translational disturbance and is a two-dimensional vector. Weighting coefficient The proportionality coefficient that incorporates translational changes into the same criterion serves to balance the effects of rotational and translational disturbances; its value is a positive real number; 1-norm. The sum of the absolute values ​​of all components of a vector; its function is to compress two-dimensional translation changes into scalar measurements. The attitude perturbation is generated jointly by the mobile terminal's built-in gyroscope and the inter-frame matching results of the images. When the attitude perturbation amount of two consecutive frames When the preset threshold is exceeded, the mobile terminal directly discards the current continuous shooting buffer and starts a new scan from the soft phase-locked loop reference in step two; when only a few frames exceed the threshold and the preceding and following frames can still be closed and matched, the mobile terminal discards the abnormal frames and retains the remaining frames.

[0061] Preferably, the lens clamp adopts a combination structure of a rigid outer frame, an elastic clamping arm, and a cartridge limiting groove to fix the microfluidic colorimetric cartridge relative to the lens optical axis. If a desktop stand is used instead of a handheld method, the mobile terminal can disable the gyroscope gating and retain only the inter-frame matching gating. If the external pulse LED controller uses a single LED, a ring LED array, or a linear LED array, as long as it is driven by the same out-of-phase frequency... Control will still be exercised according to the same discard criteria.

[0062] The mobile terminal preferably extracts the positioning marks at the four corners of the printed white border of the card box as inter-frame registration feature points, and uses RANSAC to solve the affine transformation matrix between two adjacent frames. When the proportion of interior points is lower than a preset threshold... If the determinant of the affine transformation matrix deviates from the unit value by more than a preset threshold, it is determined that the current sequence has unacceptable handheld shaking; in this case, it is preferable to discard the entire continuous shooting buffer and restart from the soft lock phase handshake in step two. Wherein: The meaning is the RANSAC in-point ratio threshold; its purpose is to measure the reliability of inter-frame geometric correspondence, and it is calibrated by pre-experimentation. The affine matrix determinant: its purpose is to determine whether there is abnormal scaling or severe misalignment, and it is directly output by the registration algorithm.

[0063] When used, the obtained The frame sequence is no longer a regular burst of images, but a spatiotemporal slice sequence with a clear order, clear coordinates, and has been filtered by shake. The subsequent step four uses this to extract pure dark pixel rows from each frame and perform non-interpolated extreme value mapping according to the slice coordinates, thereby stitching together the photon coverage obtained from multiple real-time sweeps into a complete dark canvas.

[0064] Step 4: Extract only the pixel rows that are actually in the dark phase window from the spatiotemporal slice sequence in Step 3, map these pixel rows to the holographic canvas according to their physical absolute coordinates, and then complete the holographic lossless reconstruction and CFU equivalent concentration output in a non-interpolation manner.

[0065] Step three has already caused controlled drift of the dark stripes between consecutive frames, but the original sequence still contains three types of signals: excitation residue during LED appearance, short-lived autofluorescence of the plant extract itself, and long-persistence signal released by the ZIF-8@LPL core-shell probe during the dark phase. If the first two types of signals are not excluded first, the subsequent canvas reconstruction will superimpose the appearance residue and plant background, causing discrete microbial hotspots to be submerged. On the other hand, if thresholding is only performed within a single frame, physical blind spots will still remain in areas not covered by the bright and dark stripes.

[0066] Based on this, pure dark phase extraction, physical coordinate projection, and connected component determination are sequentially connected in series. The output of the previous loop directly limits the input of the next loop. Mathematical interpolation is not allowed to fill the gaps, nor is mean diffusion allowed to replace real swept photons.

[0067] The mobile terminal first reads the valid image sequence output in step three, and then sorts it by frame number. The original grayscale image and frame-level shift amount of each frame are loaded sequentially. attitude disturbance Subsequently, the mobile terminal uses an out-of-phase driving frequency. Single-line exposure reading time Grating start phase and background latency For each row of pixels, a dark phase determination is performed, retaining only rows of pure dark phase pixels that satisfy the conditions that the LED is off and the plant's autofluorescence has decayed below the background threshold. Then, the mobile terminal shifts the pure dark phase pixel rows of each frame according to the frame-level drift amount. The image is projected onto a blank holographic canvas, and when it falls onto the same physical coordinates, maximum projection is used to preserve the upper limit of the true photon count. Finally, the mobile terminal performs connected component extraction, boundary removal, and area conversion on the reconstructed absolute black background image, and outputs the CFU equivalent concentration.

[0068] Because mobile terminal image sensors use progressive scanning, the sampling time for each row in the image is not the same. Therefore, the entire frame cannot be simply assumed to be in a simultaneously off state. The row coordinates in each frame are re-associated with the driving timescale given in step three. First, the row-level phase margin is calculated, and then the background delay is added. The gating condition is used to extract only the pure dark phase pixel rows. Its phase margin is calculated using the following formula: ; Where: phase margin : No. Frame line coordinates The foldback phase relative to the out-of-phase driving cycle is used to determine whether the line is in the active or passive phase region, and its value range is [value range missing]. Out-of-phase drive frequency Step 3 sends the actual driving frequency to the external pulse LED controller. Its function is to establish the mapping between row coordinates and the light emission period; the value is a positive real number. (Row coordinates...) The discrete positions of a single frame image along the line-by-line direction serve to carry the timing difference caused by line-by-line reading. The values ​​are integers and satisfy the following conditions: Single-line exposure reading time The actual readout interval between two adjacent rows of the image sensor in a mobile terminal is used to convert row coordinates into sampling time, and its value is a positive real number. Grating start phase The phase zeros obtained in step two are used to determine the reference alignment between the first line and the driving cycle, and their values ​​are real numbers in radians; Frame number : The sequential number of the current image in the continuously acquired sequence. Its function is to incorporate the inter-frame phase accumulation into the same expression. It takes the value of an integer and satisfies... Slight frequency deviation The controlled mismatch injected in step three serves to create an accumulative inter-frame phase offset, and its value is a non-zero real number. Camera frame rate The sampling rate of this round of continuous shooting is used to write the frame-level time advance into the phase expression, and its value is a positive real number; modulo operator :right The periodic reversal operation is used to push the accumulated phase back into a single period; The mobile terminal obtains the phase margin Next, it is determined whether the row falls near the center of the dark phase window, and whether the sampling time of the row relative to the LED off edge has a delay greater than the background delay. Background latency The calibration table provided is based on the pre-stored matrix of blank plant extract, which is used to wait for the plant autofluorescence to decay first, and then read the long afterglow signal.

[0069] Preferably, the mobile terminal places the rows that meet the above dual conditions into the set of pure dark phase pixel rows, and sets the remaining rows to zero, in order to form the first... Dark phase slice image corresponding to the frame.

[0070] In one embodiment, after continuous shooting ends, the mobile terminal reads the sequence of images frame by frame and calculates the phase margin line by line. Although the rows near the edges of the bright stripes are less bright, their sampling time has not yet exceeded the background delay. The mobile terminal will not include it in the set of pure dark pixel rows; only the row located in the center of the dark window and meeting the delay threshold will be retained.

[0071] Since step three has already written the frame-level drift amount for each frame, Therefore, each pure dark pixel row corresponds to an absolute vertical position in the micro-confined liquid film. First, the row coordinates are mapped to physical absolute coordinates. Then, a maximum projection is performed on the holographic canvas, retaining only the actually sampled upper limit of brightness and not generating any intermediate pixels that did not originally exist. The physical absolute coordinates are written using the following formula: ; Where: physical absolute coordinates : No. Frame line coordinates Corresponding to the vertical position after unifying the holographic canvas, its function is to convert intra-frame coordinates into consistent physical coordinates across frames, taking real numbers; frame-level drift amount Step 3 gives the first The line-by-line shift relative to the first frame is used to compensate for fringe slip caused by out-of-phase driving, and its value is a real number; line coordinates Consistent with the previous definition, the value remains unchanged; After completing the coordinate projection, the mobile terminal performs maximum projection on all pure dark phase slices, and the reconstructed canvas is generated by the following formula: ; In the formula: reconstructed canvas : Unified holographic canvas on the vertical axis Horizontal coordinates The reconstructed brightness at that point, taking into account the projection results of all valid pure dark pixels, takes a non-negative real number value; dark slice image. : No. Frame in line coordinates Column coordinates The pure dark phase brightness at a given location, used as the fundamental data for cross-frame projection, takes the value of a non-negative real number; the vertical coordinate... The unified physical absolute coordinates in the holographic canvas are used to receive the corresponding dark phase slices from different frames, and their values ​​are real numbers. Horizontal coordinate : Discrete positions of the image along the column direction, used to maintain the lateral structure of microbial hotspots, with values ​​taking integer values; maxima operator The maximum value among multiple candidate brightness values ​​falling at the same canvas coordinate is selected to preserve the most complete coverage of the real photon sweep without introducing mean dilution. The above expression does not include an interpolation kernel function or a neighborhood smoothing coefficient. If a canvas coordinate has never been swept by any pure dark pixel row, the coordinate remains blank; if multiple frames cover the same coordinate, the strongest brightness value is retained.

[0072] In one embodiment, after the mobile terminal projects the first pure dark phase slice onto a blank holographic canvas, only sparse bright bands appear on the canvas. As subsequent frames are projected sequentially, the previously blank vertical coordinates are gradually filled, and the discrete bright spots maintain continuous outlines at the same horizontal position. If two pure dark phase slices fall on the same vertical coordinate, the mobile terminal only retains the record with the higher brightness and does not perform average mixing.

[0073] Furthermore, the highlighted areas in the reconstructed canvas do not automatically correspond to microbial hotspots; they may still contain reflections from card boundaries, dust particle scattering, or local background. To address this, the mobile terminal first performs threshold separation on the reconstructed canvas, then performs connected component extraction, and finally performs boundary filtering based on region location, area, and shape.

[0074] Among them, areas within the card box boundary protection zone are directly removed, areas with an area lower than the isolated noise threshold are directly removed, and thin, elongated areas aligned with the stripe direction are considered residual light stripes. After screening, the number of remaining connected components is... Converted to CFU equivalent concentration: ; Where: CFU equivalent concentration : The converted concentration of microbial contamination per unit effective reconstructed area, used as the final output of this method, and taken as a non-negative real number; conversion factor. The concentration conversion factor obtained from standard bacterial suspension calibration is used to map the number of connected components to CFU equivalent concentration, and its value is a positive real number; the number of connected components The number of independent hotspots retained after boundary screening and noise removal reflects the number of discrete microbial events in the reconstructed map and is a non-negative integer. Effective reconstructed area The effective detection area actually covered by the pure dark phase slice in the holographic canvas is used to eliminate the influence of coverage differences on the output concentration. The value is a positive real number. In one embodiment, after the mobile terminal completes threshold separation on the reconstructed canvas, multiple separate bright spot clusters appear in the central region, and a thin bright band near the edge of the card box is removed because it is within the boundary protection zone. Subsequently, the mobile terminal performs connected component numbering on the remaining region and writes the number of retained connected components to the results page. If the deployment scenario only requires outputting the pollution level and not the numerical concentration, the mobile terminal directly follows the... and The segmented threshold outputs low-pollution, medium-pollution, or high-pollution labels; if the deployment scenario requires numerical results, the aforementioned conversion formula is used to generate CFU equivalent concentrations.

[0075] Therefore, the output of each link in this chain is always the hard input of the next, thus avoiding situations where interpolation is used to fill gaps, averages are used to smooth out areas, or subjective thresholds are used to guess hotspots. This allows for rapid visual screening of dark-colored, highly viscous, and high-polyphenol plant extracts even without decolorization.

[0076] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0077] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0078] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0079] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0080] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A rapid visual screening method for microbial contamination in plant extracts, characterized in that: include, The plant extract to be tested is introduced into a microfluidic colorimetric cartridge and mixed with a pre-set core-shell probe, which includes a ZIF-8 mesoporous shell and a long-afterglow luminescent core to form a micro-confined liquid film. The mobile terminal controls the external pulse LED to flash and acquire the first frame image. A single-column pixel array of the white edge printed on the cartridge is extracted and subjected to a one-dimensional fast Fourier transform to invert the single-line exposure reading time and the grating start phase. Based on the single-line exposure readout time, an out-of-phase drive command with a slight frequency offset is sent to the external pulse LED to control the CMOS continuous shooting to obtain a sequence of frames carrying the physical space sliding slice coordinates; Extract rows of pure dark-phase pixels that are in the LED off period and have delayed background decay from the sequence frames, map them to a uniform holographic canvas according to physical absolute coordinates and perform maximum projection reconstruction, and output CFU equivalent concentration.

2. The rapid visual screening method for microbial contamination of plant extracts according to claim 1, characterized in that: The microfluidic colorimetric cartridge includes a detection port, a flow channel capillary injection channel, a mixing and release chamber, a detection window, a tail physical airbag, and an exhaust end, which are connected in sequence. The plant extract to be tested enters the detection window through the flow channel capillary injection channel and is flattened into a micro-confined liquid film. When capillary stagnation occurs, the tail physical airbag provides micro-positive pressure to push it.

3. The rapid visual screening method for microbial contamination of plant extracts according to claim 2, characterized in that: The core-shell probe is pre-placed in the transition area between the mixing and release chamber and the detection window, and is fixed in the form of a dry film layer. After the original extract of the plant to be tested enters, it first wets and releases the core-shell probe, and then forms a uniformly dispersed state in the detection window before participating in subsequent image acquisition.

4. The rapid visual screening method for microbial contamination of plant extracts according to claim 3, characterized in that: When the mobile terminal captures the first frame image, it outputs the original frame and locks the exposure parameters, focal length parameters, and white balance parameters, and turns off the automatic anti-flicker processing chain. A single-column pixel array is formed by combining and removing the column mean of the continuous narrow band region of the white border printed on the card box, and then performing a one-dimensional fast Fourier transform.

5. The rapid visual screening method for microbial contamination of plant extracts according to claim 4, characterized in that: When the main peak identification of the one-dimensional fast Fourier transform does not meet the preset signal-to-noise conditions, the mobile terminal sends a wideband sweep handshake signal to the external pulse LED, collects stripe images for different candidate driving frequencies, and determines the phase-locked drive frequency and the start delay based on the prominence of the main peak and the phase continuity of the stripes.

6. The rapid visual screening method for microbial contamination of plant extracts according to claim 5, characterized in that: The mobile terminal sends out out-of-phase drive commands with slight frequency offsets to the external pulse LED based on the phase-locked drive frequency, and keeps the exposure parameters, focal length parameters and white balance parameters unchanged during continuous acquisition; each frame in the sequence records the frame number, grating start phase and physical space sliding slice coordinates.

7. The rapid visual screening method for microbial contamination of plant extracts according to claim 6, characterized in that: The mobile terminal combines the positioning features of the gyroscope output and the white edge of the card box printing to perform inter-frame affine registration. When the inter-frame matching error or attitude disturbance exceeds the preset threshold, the current continuous shooting buffer is discarded, and the acquisition of the first frame image, the inversion of the single-line exposure reading time, and the inversion of the grating start phase are re-executed.

8. The rapid visual screening method for microbial contamination of plant extracts according to claim 7, characterized in that: The mobile terminal determines the sampling time of each row of pixels in the sequence frame based on the single-line exposure reading time, the starting delay amount, the grating starting phase and the slight frequency offset. It retains only the pure dark phase pixel rows that are in the LED off period and have been delayed by background attenuation, and sets the pixel rows that do not meet the conditions to zero.

9. The rapid visual screening method for microbial contamination of plant extracts according to claim 8, characterized in that: The mobile terminal projects the pure dark phase pixel rows onto a unified holographic canvas based on the physical space sliding slice coordinates and the row coordinates of each pure dark phase pixel row. When multiple pure dark phase pixel rows from different sequence frames are mapped to the same canvas coordinates, the pixel values ​​with higher brightness are retained, and the canvas coordinates that are not covered remain blank.

10. The rapid visual screening method for microbial contamination of plant extracts according to claim 9, characterized in that: The mobile terminal performs threshold separation and connected component labeling on the reconstructed unified holographic canvas, removes connected components located in the edge area of ​​the card box and isolated noise points that do not meet the preset area conditions, and outputs the CFU equivalent concentration based on the number of retained connected components.

Images