Method, system and medium for feature extraction of immunochromatographic images

Abstract

The present invention discloses a feature extraction method, system and medium for immunochromatographic images, belonging to the field of immunochromatographic image processing. The method comprises: converting the immunochromatographic image into a grayscale image, compressing the grayscale image in sequence perpendicular to the flow direction and performing moving mean filtering on the grayscale image to obtain a one-dimensional smooth curve; determining the signal area in the grayscale image based on the derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determining the initial level set based on the contour of the signal area; iterating the initial level set using a level set update formula to obtain a final level set, wherein the level set update formula is derived by minimizing the contour model energy function as the goal, and the contour model energy function is provided with a shape term; and calculating the T-line eigenvalue and C-line eigenvalue of the grayscale image based on the final level set. This method improves the recognition accuracy and precision of immunochromatographic images in complex scenes.

Description

Technical Field

[0001] The present invention belongs to the field of immunochromatographic image processing, and more specifically, relates to a method, system and medium for extracting features from immunochromatographic images. Background Art

[0002] Immunochromatography technology achieves detection through antigen-antibody binding reactions during the chromatography process. Due to its advantages such as simple and rapid operation, no training required, and minimal or no instrumentation required, immunochromatography has been widely used in a variety of fields, including food testing, drug testing, environmental testing, and clinical diagnostic assistance.

[0003] By capturing the test card image with optical equipment and performing image processing, automated qualitative and quantitative analysis of immunochromatographic test card results can be achieved. Specifically, a light source illuminates the test card, and a photosensitive element captures the reflected or transmitted light signal from the test card, generating a color image of the test card. This image is then analyzed to determine the result.

[0004] Existing immunochromatographic image processing methods often produce suboptimal results when processing immunochromatographic images in complex scenarios. For example, after a test strip is run, the C and T lines may exhibit discontinuities, uneven intensity, and severe interference. Therefore, an image processing method that can accurately process immunochromatographic images in complex scenarios is urgently needed. Summary of the Invention

[0005] In response to the defects of the existing technology and the need for improvement, the present invention provides a feature extraction method, system and medium for immunochromatographic images, aiming to effectively identify immunochromatographic images in complex scenarios such as uneven intensity, faults, and strong interference.

[0006] To achieve the above-mentioned objectives, according to one aspect of the present invention, a feature extraction method for an immunochromatographic image is provided, comprising: converting the immunochromatographic image into a grayscale image, compressing the grayscale image in sequence along a direction perpendicular to the liquid flow and performing moving mean filtering on the grayscale image to obtain a one-dimensional smooth curve; determining the signal area in the grayscale image based on a derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determining an initial level set based on the contour of the signal area; iterating the initial level set using a level set update formula to obtain a final level set, wherein the level set update formula is derived by taking the minimization of a contour model energy function as the goal, and a shape term is provided in the contour model energy function; and calculating the T-line eigenvalue and the C-line eigenvalue of the grayscale image based on the final level set.

[0007] Furthermore, the level set update formula is:

[0008]

[0009] Among them, Φ n , Φ n+1 are the level sets before and after the n+1th iteration update, Φ 0 is the initial level set, Δt is the step parameter, δ ε () is the Dirac function, λ1 is the first area term parameter, λ2 is the second area term parameter, e1 is the first area term function, e2 is the second area term function, ν is the length term parameter, div() is the divergence function, μ is the penalty term parameter, γ is the shape term parameter, H ε () is the Heaviside function, H aver is the average Heaviside function.

[0010] Furthermore, the level set update formula is:

[0011]

[0012] Among them, Φ n , Φ n+1 are the level sets before and after the n+1th iteration update, Φ 0 is the initial level set, Δt is the step parameter, δ ε () is the Dirac function, λ1 is the first region item parameter, λ2 is the second region item parameter, c1 is the grayscale mean of the first region item, c2 is the grayscale mean of the second region item, G(x) is the grayscale image data at x, v is the length item parameter, div() is the divergence function, μ is the penalty item parameter, γ is the shape item parameter, H ε () is the Heaviside function, H aver is the average Heaviside function.

[0013] Furthermore, the level set update formula is:

[0014]

[0015] Among them, Φ n , Φ n+1 are the level sets before and after the n+1th iteration update, Φ 0 is the initial level set, Δt is the step parameter, λ is the region parameter, G(x) is the grayscale image data at x, G LFI (x) is the local fitting grayscale image data at x, m1 is the grayscale image data of the first region, m2 is the grayscale image data of the second region, δ ε () is the Dirac function, ν is the length term parameter, div() is the divergence function, μ is the penalty term parameter, γ is the shape term parameter, H ε () is the Heaviside function, Haver is the average Heaviside function.

[0016] Furthermore, the determining of the signal area in the grayscale image includes: searching for the peak point of the one-dimensional smooth curve; taking the point on the derivative curve with the same horizontal coordinate as the peak point as the starting point, searching for effective maximum points and effective minimum points on the derivative curve; and determining the signal area based on the effective maximum points and the effective minimum points.

[0017] Furthermore, the value of the peak point satisfies:

[0018] S peak >0.3*max(S)+0.7*min(S)

[0019] Among them, S peak is the value of the peak point, and S is a one-dimensional smooth curve.

[0020] Furthermore, determining the initial level set based on the contour of the signal area includes: assigning a first set value to points inside the contour, assigning a second set value to points on the contour line, and assigning a third set value to points outside the contour in the grayscale image to form the initial level set.

[0021] Furthermore, the method further includes: calculating the concentration of the analyte in the immunochromatographic detection process corresponding to the immunochromatographic image according to the T-line characteristic value and the C-line characteristic value.

[0022] According to another aspect of the present invention, a feature extraction system for immunochromatographic images is provided, comprising: a conversion and processing module for converting the immunochromatographic image into a grayscale image, and performing compression and moving mean filtering on the grayscale image in sequence along a direction perpendicular to the liquid flow to obtain a one-dimensional smooth curve; a determination module for determining the signal area in the grayscale image based on a derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determining an initial level set based on the contour of the signal area; an iteration module for iterating the initial level set using a level set update formula to obtain a final level set, wherein the level set update formula is derived by taking the minimization of a contour model energy function as the goal, and a shape term is provided in the contour model energy function; and a calculation module for calculating the T-line eigenvalue and the C-line eigenvalue of the grayscale image based on the final level set.

[0023] According to one aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method for extracting features from immunochromatographic images as described above is implemented.

[0024] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0025] (1) By adding shape prior knowledge by setting the shape term in the contour model energy function, the active contour model can better distinguish between signals and interference, and can accurately extract the T and C lines of the immunochromatography images in complex scenes. The method has stronger robustness.

[0026] (2) Compressing the image into a one-dimensional curve and then calculating the initial level set not only greatly reduces the number of iterations of the level set update, but also avoids the problem of the active contour model falling into the local optimal solution. The method has faster processing speed and higher reliability, thereby improving the speed, precision, accuracy and reliability of the concentration detection of the analyte. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flow chart of a feature extraction method for immunochromatography images provided in an embodiment of the present invention;

[0028] Figure 2A 、 Figure 2B 、 Figure 2C They are respectively the grayscale image, the initial contour of the signal region, and the final contour of the signal region after iteration provided in the first embodiment of the present invention;

[0029] Figure 3A 、 Figure 3B 、 Figure 3C They are respectively the one-dimensional curve before filtering, the one-dimensional smooth curve after filtering, the derivative curve and the corresponding initial contour grayscale image provided in the first embodiment of the present invention;

[0030] Figure 4A 、 Figure 4B 、 Figure 4C They are respectively the grayscale image, the initial contour of the signal area, and the final contour of the signal area after iteration provided in the second embodiment of the present invention;

[0031] Figure 5A 、 Figure 5B 、 Figure 5C They are respectively the one-dimensional curve before filtering, the one-dimensional smooth curve after filtering, the derivative curve and the corresponding initial contour grayscale image provided in the second embodiment of the present invention;

[0032] Figure 6A 、 Figure 6B 、 Figure 6C They are respectively the grayscale image, the initial contour of the signal region, and the final contour of the signal region after iteration provided in the third embodiment of the present invention;

[0033] Figure 7A 、 Figure 7B 、 Figure 7CThey are respectively the one-dimensional curve before filtering, the one-dimensional smooth curve after filtering, the derivative curve and the corresponding initial contour grayscale image provided in the third embodiment of the present invention;

[0034] Figure 8A 、 Figure 8B 、 Figure 8C They are respectively the grayscale image, the initial contour of the signal region, and the final contour of the signal region after iteration provided in the fourth embodiment of the present invention;

[0035] Figure 9A 、 Figure 9B 、 Figure 9C They are respectively the one-dimensional curve before filtering, the one-dimensional smooth curve after filtering, the derivative curve and the corresponding initial contour grayscale image provided by the fourth embodiment of the present invention;

[0036] Figure 10 This is a block diagram of a feature extraction system for immunochromatography images provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0038] In the present invention, the terms "first", "second", etc. (if any) in the present invention and the drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0039] Figure 1 Flowchart of the feature extraction method of immunochromatography image provided by the embodiment of the present invention. Figure 1 , combined with Figures 2A-5C , the feature extraction method of the immunochromatography image in this embodiment is described in detail, and the method includes operations S1 to S4.

[0040] Operation S1 : converting the immunochromatographic image into a grayscale image, and sequentially performing compression and moving mean filtering on the grayscale image in a direction perpendicular to the liquid flow to obtain a one-dimensional smooth curve.

[0041] In this embodiment, operation S1 includes sub-operation S11 to sub-operation S13.

[0042] In sub-operation S11 , the immunochromatographic image is converted into a grayscale image.

[0043] Capture an image within the detection window of an immunochromatographic test strip using an image acquisition device. The pixel size of the image is, for example, m×l, indicating that the image data has m rows and l columns. The image acquisition device is, for example, a common imaging capture device such as a CCD camera, a CMOS camera, or a CIS camera. The immunochromatographic test strip is, for example, an immunochromatographic test strip that can develop color under natural light, such as a colloidal gold, colloidal carbon, or colloidal selenium test strip, or it can also be a fluorescence immunochromatographic test strip that requires excitation by light of a specific wavelength.

[0044] Further, convert the original image (i.e., the immunochromatographic image) into a grayscale image G using the following formula:

[0045] G = A*r + B*g + C*b

[0046] Where r, g, and b represent the three channels of the RGB color space, and A, B, and C represent the proportions occupied by the three channels of the RGB color space.

[0047] In the above formula, A + B + C = 1, and the values of A, B, and C are different in different situations. For example, in the image of a colloidal gold immunochromatographic test strip, C < A < B is often taken; in the image of a fluorescence immunochromatographic test strip, the values of A, B, and C are determined by the color represented by the signal. The higher the proportion of the channel corresponding to the color represented by the signal.

[0048] In sub-operation S12, compress the grayscale image along a direction perpendicular to the liquid flow direction to obtain a one-dimensional curve. The liquid flow direction is the direction of liquid flow during the immunochromatographic detection process.

[0049] The one-dimensional curve corresponding to the image of an immunochromatographic test strip that can develop color under natural light is:

[0050]

[0051] The one-dimensional curve corresponding to the image of a fluorescence immunochromatographic test strip is:

[0052]

[0053] Where C i is the value of the i-th point on the one-dimensional curve, i = 1, 2,..., m, and G i,j is the pixel value of the i-th row and j-th column in the grayscale image.

[0054] In sub-operation S13, perform a moving average filter on the one-dimensional curve to obtain a one-dimensional smoothed curve.

[0055] The one-dimensional smoothed curve obtained by the moving average filter is:

[0056] S i =(C i-k +C i-k+1 +…+C i+…+C i+k-1 +C i+k ) / (2k+1)

[0057] Among them, S i is the value of the i-th point on the one-dimensional smooth curve. The value of k can be selected according to the scenario requirements, for example, k is 2.

[0058] The derivative curve of a one-dimensional smooth curve is:

[0059] D i =S i+1 -S i

[0060] Among them, D i is the value of the i-th point on the derivative curve of the one-dimensional smooth curve.

[0061] Operation S2: determining a signal region in the grayscale image according to the derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determining an initial level set according to a contour of the signal region.

[0062] According to an embodiment of the present invention, the above-mentioned determination of the signal area in the grayscale image specifically includes sub-operations S21 to S23.

[0063] In sub-operation S21 , a peak point of the one-dimensional smooth curve is searched.

[0064] The value of the peak point should meet the peak condition:

[0065] S i-p <… i-1 i >S i+1 >…>S i+p

[0066] Among them, S i is the value of the peak point, which is defined as S peak The value of p can be selected according to the scenario requirements. For example, the value of p is 3.

[0067] In addition, the peak value should also meet the following constraints:

[0068] S peak >0.3*max(S)+0.7*min(S)

[0069] Among them, S peak is the value of the peak point, and S is a one-dimensional smooth curve.

[0070] In sub-operation S22 , a point on the derivative curve having the same horizontal coordinate as the peak point is used as a starting point, and a search is performed on the derivative curve for an effective maximum point and an effective minimum point.

[0071] ​​On the derivative curve, find the effective maximum point and effective minimum point on both sides from the point with the same horizontal coordinate as the peak point. The effective maximum point refers to the point that satisfies S i ≤S peak and D i-1 <D i >D i+1 The effective minimum point is the point that satisfies S i ≤S peak and D i-1 >D i <D i+1 point.

[0072] In sub-operation S23, a signal region is determined according to the effective maximum point and the effective minimum point.

[0073] On the grayscale image, draw a line perpendicular to the liquid flow direction through the extreme points (effective maximum point, effective minimum point). This line is the initial contour line. The area between the contour lines corresponding to each pair of maximum and minimum values ​​is the ROI area of ​​C and T, that is, the signal area.

[0074] Furthermore, operation S2 includes a sub-operation S24: determining an initial level set based on the contour of the signal region. Specifically, a first set value is assigned to points inside the contour, a second set value is assigned to points on the contour line, and a third set value is assigned to points outside the contour in the grayscale image, to form an initial level set. The first set value, the second set value, and the third set value are, for example, 1, 0, and -1, respectively. The resulting initial level set can be expressed as:

[0075]

[0076] Among them, Φ i,j is the value of the i-th row and j-th column of the level set Φ.

[0077] Operation S3, iterating the initial level set using the level set update formula to obtain the final level set, wherein the level set update formula is derived by minimizing the contour model energy function as the goal, and the contour model energy function is provided with a shape term.

[0078] Specifically, the level set is updated according to the level set update formula, and after several iterations, the final level set is output. The number of iterations required varies depending on the location of the initial contour, typically ranging from dozens to hundreds. In the embodiment of the present invention, only 10-50 iterations are sufficient.

[0079] In this embodiment, the goal is to minimize the energy function of the contour model, and the level set update formula is derived through the variational method. The energy function of the contour model can be expressed as:

[0080] E total =Eregion +μ*E pena +v*E length +γ*E shape

[0081] Among them, E total is the total energy function, E region is the area term, E pena is the penalty term, E length is the length term, E shape is the shape term, μ is the penalty term parameter, v is the length term parameter, and γ is the shape term parameter.

[0082]

[0083]

[0084] E shape =∫ Ω (H ε (Φ)-H aver ) 2 dx

[0085] Among them, Ω is the entire image area, and x is a point on the image area.

[0086] In one embodiment of the present invention, the region item E region for:

[0087]

[0088]

[0089]

[0090] Among them, K σ () is the Gaussian convolution kernel, and the standard deviation σ is usually 2 to 5, K σ The size of () is [4σ+1]×[4σ+1], [] represents the smallest positive integer greater than or equal to; y is a point in the neighborhood near point x, C is the contour in the image, G(x) is the grayscale image data at x, f1(y) and f2(y) are the first neighborhood mean and second neighborhood mean of point y, respectively.

[0091] The corresponding derived level set update formula is:

[0092]

[0093] The relevant intermediate parameters in the above level set update formula are specifically:

[0094]

[0095]

[0096] e1=∫ Ω K σ (yx)|G(x)-f1(y)| 2 dy

[0097] e2=∫ Ω K σ (yx)|G(x)-f2(y)| 2 dy

[0098]

[0099] Among them, Φ n , ε n+1 are the level sets before and after the n+1th iteration update, ε 0 is the initial level set, Δt is the step parameter, δ ε () is the Dirac function, λ1 is the first area term parameter, λ2 is the second area term parameter, e1 is the first area term function, e2 is the second area term function, ν is the length term parameter, div() is the divergence function, μ is the penalty term parameter, γ is the shape term parameter, H ε () is the Heaviside function, H aver is the average Heaviside function, ε is the smoothing parameter, H averi,j is the value of the average Heaviside function at position (i, j), H εi,j is the value of the Heaviside function at position (i, j). Typically, λ1=λ2=1, ν is 65 to 130, μ is 1, γ is 200 to 2000, Δt is 0.0 to 0.1, and ε is 1.

[0100] In another embodiment of the present invention, the area item E region for:

[0101] E region =λ1*∫ in(C) |G(x)-c1| 2 dx+λ2*∫ out(C) |G(x)-c2| 2 dx

[0102]

[0103]

[0104] The corresponding derived level set update formula is:

[0105]

[0106] Wherein, c1 is the grayscale mean value of the first region item, c2 is the grayscale mean value of the second region item, and G(x) is the grayscale image data at position x.

[0107] In another embodiment of the present invention, the area item E region for:

[0108]

[0109] G LFI =m1*H ε (Φ)+m2*(1-H ε (Φ))

[0110] m1=mean(G∈({x∈Ω|Φ(x)>0}∩W k (x)))

[0111] m2=mean(G∈({x∈Ω|ε(x)<0}∩W k (x)))

[0112] The corresponding derived level set update formula is:

[0113]

[0114] Among them, λ is the regional parameter, G LFI (x) is the local fitting grayscale image data at x, m1 is the grayscale image data of the first region, and m2 is the grayscale image data of the second region. mean() is the mean calculation function, W k () is a rectangular window function, usually a Gaussian window function or a constant window function.

[0115] In operation S4 , the T-line feature value and the C-line feature value of the grayscale image are calculated according to the final level set.

[0116] Specifically, the grayscale mean of the T-line signal in the grayscale image is calculated, recorded as GAT; the grayscale mean of the C-line signal in the grayscale image is calculated, recorded as GAC; then, the grayscale mean of the other areas in the grayscale image except the T-line signal and the C-line signal is calculated, recorded as GAB; finally, the T-line eigenvalue TV = |GAT-GAB| and the C-line eigenvalue CV = |GAC-GAB| are calculated.

[0117] According to an embodiment of the present invention, the method further includes: calculating the concentration of the analyte in the immunochromatographic detection process corresponding to the immunochromatographic image based on the T-line characteristic value and the C-line characteristic value.

[0118] Example 1:

[0119] The colloidal gold immunochromatographic test paper card image is captured by a CCD camera and converted into a grayscale image using the formula (G = 0.299*r + 0.587*g + 0.114*b), as shown in FIG. Figure 2A As shown. Further, by peak detection, the initial contour of the signal area is calculated, as shown Figure 2B The one-dimensional curve formed during the peak detection process is shown in Figure 3A As shown, the obtained one-dimensional smooth curve is Figure 3B As shown, the derivative curve and initial contour grayscale image are as follows Figure 3C shown.

[0120] According to the level set update formula, the active contour model is evolved with the following parameters: λ1 = λ2 = 1, v = 65, μ = 1, γ = 400, Δt = 0.05, ε = 1, σ = 3, and the number of iterations is 20. The evolution results are shown in Figure 2. Figure 2C According to the contour evolution results, TV=48.92 and CV=73.61 are calculated.

[0121] Example 2:

[0122] The colloidal gold immunochromatographic test paper card image is captured by a CMOS camera and converted into a grayscale image using the formula (G = 0.299*r + 0.587*g + 0.114*b), as shown in Figure 4A As shown. Further, by peak detection, the initial contour of the signal area is calculated, as shown Figure 4B The one-dimensional curve formed during the peak detection process is shown in Figure 5A As shown, the obtained one-dimensional smooth curve is Figure 5B As shown, the derivative curve and initial contour grayscale image are as follows Figure 5C shown.

[0123] According to the level set update formula, the active contour model is evolved with the following parameters: λ1 = λ2 = 1, v = 80, μ = 1, γ = 5500, Δt = 0.01, ε = 1, σ = 2, and the number of iterations is 50. The evolution results are shown in Figure 2. Figure 4C According to the contour evolution results, TV=48.92 and CV=73.61 are calculated.

[0124] Example 3:

[0125] The colloidal gold immunochromatographic test paper card image is captured by a CCD camera and converted into a grayscale image using the formula (G = 0.299*r + 0.587*g + 0.114*b), as shown in Figure 6A As shown. Further, by peak detection, the initial contour of the signal area is calculated, as shown Figure 6B The one-dimensional curve formed during the peak detection process is shown in Figure 7A As shown, the obtained one-dimensional smooth curve is Figure 7B As shown, the derivative curve and initial contour grayscale image are as follows Figure 7C shown.

[0126] According to the level set update formula, the active contour model is evolved with the following parameters: λ1 = λ2 = 1, ν = 125, μ = 1, γ = 2000, Δt = 0.03, ε = 1, σ = 5, and the number of iterations is 10. The evolution results are shown in Figure 2. Figure 6C According to the contour evolution results, TV=40.21 and CV=12.71 are calculated.

[0127] Example 4:

[0128] The colloidal gold immunochromatographic test paper card image is captured by a CMOS camera and converted into a grayscale image using the formula (G = 0.299*r + 0.587*g + 0.114*b), as shown in Figure 8A As shown. Further, by peak detection, the initial contour of the signal area is calculated, as shown Figure 8B The one-dimensional curve formed during the peak detection process is shown in Figure 9A As shown, the obtained one-dimensional smooth curve is Figure 9B As shown, the derivative curve and initial contour grayscale image are as follows Figure 9C shown.

[0129] According to the level set update formula, the active contour model is evolved with the following parameters: λ1 = λ2 = 1, ν = 100, μ = 1, γ = 200, Δt = 0.1, ε = 1, σ = 2, and the number of iterations is 30. The evolution results are shown in Figure 2. Figure 8C According to the contour evolution results, TV=27.5 and CV=34.05 are calculated.

[0130] In combination with the above embodiments, it can be seen that by adding a shape prior active contour model to process immunochromatographic images, T and C line signals in immunochromatographic images of complex scenes can be effectively extracted, the accuracy of the results when processing abnormal images can be improved, and the application scenarios of the immunochromatographic image processing algorithm can be expanded.

[0131] Figure 10 This is a block diagram of a feature extraction system for immunochromatographic images provided by an embodiment of the present invention. Figure 10 The immune chromatography image feature extraction system 1000 includes a conversion and processing module 1010 , a determination module 1020 , an iteration module 1030 and a calculation module 1040 .

[0132] The conversion and processing module 1010, for example, executes operation S1 to convert the immunochromatographic image into a grayscale image, and sequentially performs compression and moving mean filtering on the grayscale image perpendicular to the liquid flow direction to obtain a one-dimensional smooth curve.

[0133] The determination module 1020, for example, performs operation S2 to determine a signal region in the grayscale image according to the derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determine an initial level set according to a contour of the signal region.

[0134] The iteration module 1030, for example, performs operation S3, which is used to iterate the initial level set using the level set update formula to obtain the final level set, wherein the level set update formula is derived by minimizing the contour model energy function, and the contour model energy function is provided with a shape term.

[0135] The calculation module 1040 performs, for example, operation S4 to calculate the T-line feature value and the C-line feature value of the grayscale image according to the final level set.

[0136] The feature extraction system 1000 for immunochromatographic images is used to perform the above Figure 1-9C The feature extraction method of the immunochromatographic image in the embodiment shown. Figure 1-9C The feature extraction method of the immunochromatographic image in the illustrated embodiment will not be described in detail here.

[0137] The embodiment of the present invention further provides a computer readable storage medium on which a computer program is stored. When the program is executed by a processor, the following is achieved: Figure 1-9C The feature extraction method of the immunochromatographic image in the illustrated embodiment will not be described in detail here.

[0138] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A feature extraction method for an immunochromatographic image, characterized in that: include: Converting the immunochromatographic image into a grayscale image, and sequentially performing compression and moving mean filtering on the grayscale image in a direction perpendicular to the liquid flow to obtain a one-dimensional smooth curve; determining a signal region in the grayscale image according to a derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determining an initial level set according to a contour of the signal region; Iterating the initial level set using a level set update formula to obtain a final level set, wherein the level set update formula is derived by minimizing a contour model energy function, wherein the contour model energy function includes a shape term; The T-line feature value and the C-line feature value of the grayscale image are calculated according to the final level set.

2. The feature extraction method of the immunochromatographic image according to claim 1, wherein: The level set update formula is: Among them, Φ n , Φ n+1 are the level sets before and after the n+1th iteration update, Φ 0 is the initial level set, Δt is the step parameter, δ ε () is the Dirac function, λ1 is the first area term parameter, λ2 is the second area term parameter, e1 is the first area term function, e2 is the second area term function, v is the length term parameter, div() is the divergence function, μ is the penalty term parameter, γ is the shape term parameter, H ε () is the Heaviside function, H aver is the average Heaviside function.

3. The feature extraction method of the immunochromatographic image according to claim 1, wherein: The level set update formula is: Among them, Φ n , Φ n+1 are the level sets before and after the n+1th iteration update, Φ 0 is the initial level set, Δt is the step parameter, δ ε () is the Dirac function, λ1 is the first region item parameter, λ2 is the second region item parameter, c1 is the grayscale mean of the first region item, c2 is the grayscale mean of the second region item, G(x) is the grayscale image data at x, v is the length item parameter, div() is the divergence function, μ is the penalty item parameter, γ is the shape item parameter, H ε () is the Heaviside function, H aver is the average Heaviside function.

4. The feature extraction method of the immunochromatographic image according to claim 1, wherein: The level set update formula is: Among them, Φ n , Φ n+1 are the level sets before and after the n+1th iteration update, Φ 0 is the initial level set, Δt is the step parameter, λ is the region parameter, G(x) is the grayscale image data at x, G LFI (x) is the local fitting grayscale image data at x, m1 is the grayscale image data of the first region, m2 is the grayscale image data of the second region, δ ε () is the Dirac function, v is the length term parameter, div() is the divergence function, μ is the penalty term parameter, γ is the shape term parameter, H ε () is the Heaviside function, H aver is the average Heaviside function.

5. The feature extraction method of the immunochromatographic image according to claim 1, wherein: Determining the signal area in the grayscale image includes: Searching for a peak point of the one-dimensional smooth curve; Taking the point on the derivative curve having the same horizontal coordinate as the peak point as a starting point, searching for an effective maximum point and an effective minimum point on the derivative curve; The signal area is determined according to the effective maximum point and the effective minimum point.

6. The feature extraction method of the immunochromatographic image according to claim 5, wherein: The value of the peak point satisfies: S peak >0.3*max(S)+0.7*min(S) Among them, S peak is the value of the peak point, and S is a one-dimensional smooth curve.

7. The feature extraction method of the immunochromatographic image according to claim 1, wherein: The determining of the initial level set according to the contour of the signal area comprises: A first set value is assigned to points inside the contour, a second set value is assigned to points on the contour line, and a third set value is assigned to points outside the contour in the grayscale image to form the initial level set.

8. The method for extracting features from immunochromatographic images according to any one of claims 1 to 7, wherein: Also includes: The concentration of the analyte in the immunochromatographic detection process corresponding to the immunochromatographic image is calculated according to the T-line characteristic value and the C-line characteristic value.

9. A feature extraction system for immunochromatographic images, characterized in that: include: a conversion and processing module, configured to convert the immunochromatographic image into a grayscale image, and sequentially perform compression and moving mean filtering on the grayscale image in a direction perpendicular to the liquid flow to obtain a one-dimensional smooth curve; a determination module, configured to determine a signal region in the grayscale image according to a derivative curve of the one-dimensional smooth curve and the one-dimensional smooth curve, and determine an initial level set according to a contour of the signal region; an iterative module, configured to iterate the initial level set using a level set update formula to obtain a final level set, wherein the level set update formula is derived by minimizing a contour model energy function, wherein the contour model energy function includes a shape term; A calculation module is used to calculate the T-line eigenvalue and the C-line eigenvalue of the grayscale image according to the final level set.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the feature extraction method for immunochromatographic images according to any one of claims 1 to 8 is implemented.

Images