Identification method for monitoring growth trend of haematococcus pluvialis

Abstract

The invention discloses an identification method for monitoring the growth trend of haematococcus pluvialis, and belongs to the field of image processing. The method comprises the following steps: carrying out image edge detection and image enhancement on a cell map in a haematococcus pluvialis illumination process; removing all connected regions smaller than a given area in a binary image from the obtained cell image; carrying out pit detection on the processed image to extract a cell part with adhesion; marking connected regions in the binary image, calculating the center of mass, the area and the perimeter of each marked region to obtain position information and area information of each cell, and storing the position information and the area information in an array; all calculated cells in the array are traversed, the center-of-mass coordinate of each cell is obtained, the cell radius is calculated according to the area information, a viscircules function is used for drawing cell contours, cell numbers are marked beside each contour, and a final result is obtained. Compared with the prior art, the haematococcus pluvialis growth condition monitoring system can realize rapid and accurate monitoring of haematococcus pluvialis growth conditions under different conditions.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to an identification method for monitoring the growth trend of Haematococcus pluvialis. Background Art

[0002] At present, the growth monitoring technology of Haematococcus pluvialis mainly relies on the hemocytometer counting method and spectrophotometry to manually count cells and directly measure the content of various pigments. Some researchers also use other chemical means to break the cell wall and analyze the content of lipids, nitrogen, phosphorus, etc. to analyze the growth status of algal cells.

[0003] However, for the growth trend analysis of Haematococcus pluvialis: the traditional use of spectrophotometry and other chemical analysis methods to determine the content of various pigments cannot avoid problems such as damage to cell structure, high cost and cumbersome operation; the use of traditional blood cell counting plates to manually count and monitor cell concentration changes requires a lot of time and manpower, and the counting efficiency is low; the existing method of calculating the number and size of cells through image processing may identify adherent cells as one cell, affecting the final quantitative statistics; therefore, there is an urgent need for an image processing method that can divide cells, which not only avoids the complex and time-consuming traditional analysis methods, but also can intuitively and accurately monitor cell growth trends. Summary of the invention

[0004] In order to solve the above problems, the present invention provides an identification method for monitoring the growth trend of Haematococcus pluvialis.

[0005] To implement the above technology, the steps are as follows:

[0006] S1. Perform image edge detection and image enhancement on the cell image of Haematococcus pluvialis during illumination. The steps are as follows:

[0007] S1.1, crop the target area of ​​the cell image of Haematococcus pluvialis during illumination using the imcrop function;

[0008] S1.2, perform HSV processing on the cropped image;

[0009] HSV processing is as follows: extract the H, S, and V channels in the image, and select the saturation channel image with the clearest cell outline and the least noise based on color feature recognition, detail retention, and robustness to illumination changes;

[0010] S1.3. Binarize the saturation channel in the HSV processed image and perform the Canny edge detection algorithm to obtain an enhanced image. The steps of the Canny edge detection algorithm are as follows:

[0011] S1.3.1. Use the Sobel operator to calculate the gradient of the image and obtain the gradient magnitude and gradient direction;

[0012] First, the input image is smoothed by Gaussian filtering to remove noise. The expression for convolution using Gaussian filter is as follows:

[0013]

[0014] I(x,y)=I 0 (x,y)*G(x,y)

[0015] Where x and y are the pixel coordinates in the image; σ is the standard deviation of the Gaussian kernel, which determines the degree of smoothing; I 0 (x,y) is the original image; G(x,y) is the Gaussian filter;

[0016] The horizontal gradient is used to detect the horizontal edge in the image, and the vertical gradient is used to detect the vertical edge in the image; the template G of the horizontal gradient Sobel operator convolution kernel x And the template G of the Sobel operator convolution kernel in the vertical direction y The expression is as follows:

[0017]

[0018] Use the Sobel operator to perform convolution to calculate the horizontal direction of the gradient G' x and vertical direction G' y , the expression is as follows:

[0019] G' x =I(x,y)*G x , G' y =I(x,y)*G y ;

[0020] The gradient amplitude expression of the image at each pixel position is as follows:

[0021]

[0022] The expression of the gradient direction is as follows:

[0023]

[0024] S1.3.2, perform non-maximum suppression operation on the gradient amplitude according to the gradient direction angle;

[0025] S1.3.3, performing maximum and minimum normalization operation on the gradient amplitude after the non-maximum suppression operation, and dividing the gradient amplitude after the maximum and minimum normalization operation into edge points by setting a high threshold and a low threshold;

[0026] S1.4, performing a binarization operation on the saturation channel image;

[0027] Since the saturation channel performs better in color feature recognition, detail retention, and robustness to lighting changes, the present invention selects the saturation channel for processing;

[0028] The saturation channel image of the Prototheca cell is f(x,y), with a size of 3072×4096 pixels. The image f(x,y) is converted to a binary image. The binary operation formula is:

[0029]

[0030] Where s(x,y) is the pixel value of (x,y) in image f, and thresh is the set binarization threshold;

[0031] The result of the binary operation, 1 represents the foreground and 0 represents the background;

[0032] S1.5. Adding the results of the binarization operation and the Canny edge detection algorithm can not only improve the accuracy of image segmentation and the quality of edge recognition, but also reduce noise and simplify the amount of data, thereby achieving cell image enhancement.

[0033] S2, according to the morphological operation, remove all connected regions smaller than a given area in the binary image from the image processing result of S1, the steps are as follows:

[0034] S2.1, performing corrosion operation on the image;

[0035] A disk-shaped structure element with a radius of 2 pixels is used to perform an erosion operation on the S1 result to eliminate small areas in the image. The calculation formula for the erosion operation is:

[0036] (A!B)(x,y)=min (s,t)∈B {A(x+s,y+t)}

[0037] Where A represents the input image; B represents the structural element; A(x, y) is the pixel value of the input image, (s, t) is the offset coordinate of the structural element B, and erosion is used to eliminate small areas or thin lines in the image; A(x+s, y+t) represents the erosion of the input image under the structural element;

[0038] S2.2, dilate the image processed by S2.1 using a disk-shaped structure element with a radius of 1 pixel;

[0039] The target object in the image can be expanded through the dilation operation;

[0040] The expansion operation expression is as follows:

[0041]

[0042] Where A(xs, yt) represents the dilation of the input image under the structural element;

[0043] S2.3, performing a filling operation on the image processed by S2.2;

[0044] The filling operation is based on the connectivity principle of the image, filling the hole areas in the image so that the target area becomes connected and complete;

[0045] S2.4, removing all connected regions smaller than a given area from the image processed by S2.3 using a preset area threshold;

[0046] Set an area threshold T = 200, and use the function to identify and remove connected areas with an area smaller than T. The expression is as follows:

[0047] A out ={A cc |Area(A cc )≥T}

[0048] Among them, A cc Represents the connected area in the image A(x,y), Area(A cc ) is the area of ​​the connected region. After removing the small region, the area of ​​the remaining region is greater than or equal to the area threshold;

[0049] S3, performing concave point detection on the image processed in S2 to obtain adhesion cells, and performing segmentation operation on the adhesion cells;

[0050] S3.1, based on the result of S2.4, remove cells that are adhered to the border using a preset border adhesion threshold;

[0051] The border adhesion threshold is set to 700 pixels;

[0052] S3.2, after removing the cells adhered to the border, the adhered cells are obtained by using a preset adhesion demarcation threshold, and the adhered cells are divided;

[0053] Here are the steps:

[0054] S3.2.1, setting a preset adhesion demarcation threshold to obtain adhesion cells;

[0055] When two or more cells overlap, they will form adhesion cells. In the actual analysis process, the adhesion cells need to be separated before subsequent analysis operations can be performed. After the cell adhesion division, there will be two situations: convex hull and concave hull. Among them, the highest point of the convex hull forms a convex point, and the lowest point of the concave hull forms a concave point. The formation of the concave hull is obtained by the overlap of two or more cells, so the concave hull will also have two or more convex points.

[0056] The preset adhesion demarcation thresholds include: the concave point detection area threshold is 2000 pixels, the perimeter threshold is 350 pixels, and the binarization threshold is 0.3;

[0057] S3.2.2, the concave point distance is obtained by calculating the minimum Euclidean distance between the convex points of the adhesion cells;

[0058] Euclidean distance calculation formula:

[0059]

[0060] Where P 1 =(x 1 ,y 1 ) and P 2 =(x 2 ,y 2 ) are the coordinates of the convex point I and the convex point II respectively;

[0061] S3.2.3, obtaining the pixel points of the convex points at both ends of the adhesion cells by the concave point distance, and drawing the dividing line between the two convex points by the straight line equation;

[0062] Equation of a Line:

[0063] y=mx+b

[0064] Where m is the slope, b is the intercept, Draw connecting lines, which are the dividing lines between adhesion cells, where (x min ,y min ) represents a convex point in a small area, (x max ,y max ) represents the convex point with the largest area; x represents the concave point distance;

[0065] Get the minimum Euclidean distance by traversing each Euclidean distance;

[0066] After obtaining the segmentation line, the adherent cells can be segmented by software to obtain a segmented binary image;

[0067] S4, marking the binary images after the segmentation operation, calculating the centroid, area and perimeter of each binary image after the segmentation operation, obtaining the coordinate information and area information of each cell in the binary image after the segmentation operation, and storing them in an array;

[0068] The centroid is the geometric center of the cell region and the weighted average position of all pixels in the region; the centroid coordinates (C x ,C y ) is calculated as:

[0069]

[0070] In the formula, S' represents the pixel point of the area;

[0071] The area is obtained by the number of pixels in the region;

[0072] The perimeter is the length of the region boundary, obtained using a chain code based on boundary pixels;

[0073] S5, traverse all calculated cells in the array, obtain the area information of each cell, calculate the centroid coordinates and cell radius, use the viscircles function to draw the cell outline, and mark the cell number next to each outline, and obtain the identification result of the growth trend of Haematococcus pluvialis through the size of the cell outline;

[0074] Each cell is approximately a circle, and the relationship between the cell area S and the radius r of the circle is: S = πr 2 ; Draw the contour according to the radius, the center coordinate of the cell (C x' ,C y' ) represents the centroid of the cell;

[0075] The equation of a circle is:

[0076] (x'-C x ) 2 +(y'-C y ) 2 =r 2

[0077] Where (x', y') is the coordinate of any point on the circle;

[0078] To make it clear to the human eye, each cell is numbered and the cell number is placed on the left side of the cell circle at (C x -r,C y ) to ensure that the numbering does not overlap with the circle;

[0079] The number of cells is obtained by traversing all the calculated cells in the array, the size of each cell is obtained through the cell outline, and the growth trend of Haematococcus pluvialis is obtained according to the number and size of cells.

[0080] Beneficial effects of the present invention:

[0081] The present invention provides a rapid identification method specifically for monitoring the growth trend of Haematococcus pluvialis. Chemical or biological detection methods can be effectively avoided through image processing. By obtaining pictures of different stages in the growth process of microalgae and calculating and analyzing the number and area changes of algae cells, compared with traditional technologies, rapid and accurate monitoring of the growth status of Haematococcus pluvialis under different conditions can be achieved.

[0082] The present invention divides cell adhesion in image processing and can achieve more accurate quantitative statistics of the number of cells. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0084] Figure 1 is a flow chart of the steps of the present invention;

[0085] Figure 2 The schematic diagram of the rapid identification method for monitoring the growth trend of Haematococcus pluvialis of the present invention;

[0086] Figure 3 It is a schematic diagram of the cutting process of the present invention;

[0087] Figure 4 The HSV processing result diagram of the present invention; wherein, part (a) is the original image; part (b) represents the hue channel (H); part (c) represents the hue channel (S); and part (d) represents the hue channel (V);

[0088] Figure 5 This is a binary edge detection result diagram of the present invention;

[0089] Figure 6 The present invention is a morphological processing result diagram;

[0090] Figure 7 The extraction and concave point detection and segmentation result diagram of adhesion cells in the algae cell image of the present invention; wherein, part (a) is the result diagram after preprocessing; part (b) is the result diagram of removing adhesion edge cells; part (c) is the image of cells with adhesion; part (d) is the concave point segmentation result diagram of adhesion cell image;

[0091] Figure 8 This is the final calculation result diagram of the algae cell image of the present invention. DETAILED DESCRIPTION

[0092] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0093] like Figure 1 and Figure 2 As shown, a method for monitoring the growth trend of Haematococcus pluvialis comprises the following steps:

[0094] S1. Perform image edge detection and image enhancement on the cell image of Haematococcus pluvialis during illumination. The steps are as follows:

[0095] S1.1. Use the imcrop function to crop the target area of ​​the cell image of Haematococcus pluvialis during illumination, such as Figure 3 As shown;

[0096] The cropping method is: manually cropping the calculation area containing all algal cells and removing the background area outside the cells to improve the accuracy of image recognition;

[0097] S1.2, perform HSV processing on the cropped image, such as Figure 4 As shown;

[0098] HSV processing is to extract the H, S, and V channels in the image, and select the saturation channel image with the clearest cell outline and the least noise based on color feature recognition, detail retention, and robustness to illumination changes, which is automatically completed by computer software; Figure 4 Part (a) shows the original image; Figure 4 The middle part (b) represents the hue channel (H); Figure 4 Part (c) in the middle represents the saturation channel (S); Figure 4 The middle part (d) represents the value channel (V);

[0099] S1.3, the saturation channel in the HSV processed image is subjected to binarization operation and Canny edge detection algorithm to obtain an enhanced image, such as Figure 5 As shown; the steps of Canny edge detection algorithm are as follows:

[0100] S1.3.1. Use the Sobel operator to calculate the gradient of the image and obtain the gradient magnitude and gradient direction;

[0101] Performing gradient calculation means calculating the gradient size and direction of the image;

[0102] The Sobel operator is a first-order discrete difference operator used for edge detection. It calculates the gradient values ​​of the image in the horizontal and vertical directions respectively through two 3x3 convolution kernels;

[0103] First, the input image is smoothed by Gaussian filtering to remove noise. The expression for convolution using Gaussian filter is as follows:

[0104]

[0105] I(x,y)=I 0 (x,y)*G(x,y)

[0106] Where x and y are the pixel coordinates in the image; σ is the standard deviation of the Gaussian kernel, which determines the degree of smoothing; I 0 (x,y) is the original image; G(x,y) is the Gaussian filter;

[0107] The horizontal gradient is used to detect the horizontal edge in the image, and the vertical gradient is used to detect the vertical edge in the image; the template G of the horizontal gradient Sobel operator convolution kernel x And the template G of the Sobel operator convolution kernel in the vertical direction y The expression is as follows:

[0108]

[0109] Use the Sobel operator to perform convolution to calculate the horizontal direction of the gradient G' x and vertical direction G' y , the expression is as follows:

[0110] G' x =I(x,y)*G x , G' y =I(x,y)*G y ;

[0111] The gradient amplitude expression of the image at each pixel position is as follows:

[0112]

[0113] The expression of the gradient direction is as follows:

[0114]

[0115] S1.3.2, perform non-maximum suppression operation on the gradient amplitude according to the gradient direction angle;

[0116] Specifically, the non-maximum suppression operation is used to refine the edge, retain the pixel with the largest gradient amplitude, and suppress other edge pixels; that is, according to the gradient direction θ, it is determined whether the gradient amplitude of each pixel is a local maximum: if the gradient amplitude of the current pixel is greater than the gradient amplitudes of its two adjacent pixels, the gradient amplitude of the pixel is retained; otherwise, the gradient amplitude of the pixel is set to 0;

[0117] The non-maximum suppression operation can obtain the judgment result through MATLAB code;

[0118] S1.3.3, performing maximum and minimum normalization operation on the gradient amplitude after the non-maximum suppression operation, and dividing the gradient amplitude after the maximum and minimum normalization operation into edge points by setting a high threshold and a low threshold;

[0119] Specifically, the gradient amplitude is normalized to the maximum and minimum values ​​in the range of [0,1], and the pixel points are divided by setting the high and low thresholds, that is, the strength of the edge is determined; the high threshold is set to 0.65, and the low threshold is set to 0.4*0.65; pixels greater than the high threshold are considered edge points by the software, pixels less than the low threshold are considered not edge points by the software, and pixels between the two thresholds are considered edge candidate points by the software, which are determined by the connectivity with the strong edge;

[0120] The results are automatically generated by setting the high and low thresholds through the Canny edge detection algorithm;

[0121] S1.4, performing a binarization operation on the saturation channel image;

[0122] Since the saturation channel performs better in color feature recognition, detail retention, and robustness to lighting changes, the present invention selects the saturation channel for processing;

[0123] The saturation channel image of the Prototheca cell is f(x,y), with a size of 3072×4096 pixels. The image f(x,y) is converted to a binary image. The binary operation formula is:

[0124]

[0125] Where s(x,y) is the pixel value of (x,y) in image f, thresh is the set binarization threshold, which is 0.3;

[0126] The result of the binarization operation is that 1 represents the foreground, that is, the white cell part, and 0 represents the background, that is, the black background part;

[0127] S1.5. Adding the results of the binarization operation and the Canny edge detection algorithm can not only improve the accuracy of image segmentation and the quality of edge recognition, but also reduce noise and simplify the amount of data, thereby achieving cell image enhancement.

[0128] S2, according to the morphological operation, remove all connected regions smaller than a given area in the binary image from the image processing result of S1, the steps are as follows:

[0129] S2.1, performing corrosion operation on the image;

[0130] A disk-shaped structure element with a radius of 2 pixels is used to perform an erosion operation on the S1 result to eliminate small areas in the image. The calculation formula for the erosion operation is:

[0131] (A!B)(x,y)=min (s,t)∈B {A(x+s,y+t)}

[0132] Where A represents the input image; B represents the structural element; A(x, y) is the pixel value of the input image, (s, t) is the offset coordinate of the structural element B, and erosion is used to eliminate small areas or thin lines in the image; A(x+s, y+t) represents the erosion of the input image under the structural element;

[0133] S2.2, dilate the image processed by S2.1 using a disk-shaped structure element with a radius of 1 pixel;

[0134] The target object in the image can be expanded through the dilation operation;

[0135] The expansion operation expression is as follows:

[0136]

[0137] Where A(xs, yt) represents the dilation of the input image under the structural element;

[0138] S2.3, performing a filling operation on the image processed by S2.2;

[0139] The filling operation is based on the connectivity principle of the image, filling the hole areas in the image so that the target area becomes connected and complete;

[0140] S2.4, removing all connected regions smaller than a given area from the image processed by S2.3 using a preset area threshold;

[0141] Set an area threshold T = 200, and use the function to identify and remove connected areas with an area smaller than T. The expression is as follows:

[0142] A out ={Acc |Area(A cc )≥T}

[0143] Among them, A cc Represents the connected area in the image A(x,y), Area(A cc ) is the area of ​​the connected region. After removing the small region, the area of ​​the remaining region is greater than or equal to the area threshold. The morphological operation processing results are as follows: Figure 6 As shown;

[0144] S3, performing concave point detection on the image processed in S2 to obtain adhesion cells, and performing segmentation operation on the adhesion cells;

[0145] S3.1, based on the result of S2.4, remove cells that are adhered to the border using a preset border adhesion threshold;

[0146] The border adhesion threshold is set to 700 pixels;

[0147] The removal method is implemented through code, removing the border adhesion part and retaining the remaining cells for calculation. The result is as follows Figure 7 As shown in parts (a) and (b);

[0148] S3.2, after removing the cells adhered to the border, the adhered cells are obtained by using a preset adhesion demarcation threshold, and the adhered cells are divided;

[0149] Specifically, assuming that the cells on the image are a set of pixels {(x 1 ,y 1 ),(x 2 ,y 2 ),…,(x n ,y n )}; The goal of calculating the convex hull is to find the outermost point of the point set. If the concentrated point is located in the convex hull, there is a straight line connecting the concentrated point and any point on the convex hull. This straight line will not cross the boundary of the convex hull, and thus the convex hull of the regional contour points is calculated; the original image is subtracted from the convex hull image to obtain the concave area, that is, the area outside the convex hull, and then the obtained convex points are matched, and the distances between the convex points in the concave area with the largest area and the convex points in the remaining concave areas are calculated respectively. The two points with the shortest distance are taken as the segmentation concave points, and the coordinates of the two matched points are stored in an array. The plot function is used to connect the matched concave points to achieve the segmentation of the adhesion cells, and finally a segmented binary image is obtained;

[0150] Here are the steps:

[0151] S3.2.1, setting a preset adhesion demarcation threshold to obtain adhesion cells;

[0152] When two or more cells overlap, they will form adhesion cells. In the actual analysis process, the adhesion cells need to be separated before subsequent analysis operations can be performed. After the cell adhesion division, there will be two situations: convex hull and concave hull. Among them, the highest point of the convex hull forms a convex point, and the lowest point of the concave hull forms a concave point. The formation of the concave hull is obtained by the overlap of two or more cells, so the concave hull will also have two or more convex points.

[0153] The preset adhesion demarcation thresholds include: the concave point detection area threshold is 2000 pixels, the perimeter threshold is 350 pixels, and the binarization threshold is 0.3;

[0154] S3.2.2, the concave point distance is obtained by calculating the minimum Euclidean distance between the convex points of the adhesion cells;

[0155] Euclidean distance calculation formula:

[0156]

[0157] Where P 1 =(x 1 ,y 1 ) and P 2 =(x 2 ,y 2 ) are the coordinates of the convex point I and the convex point II respectively;

[0158] S3.2.3, obtaining the pixel points of two convex points between the adhesion cells by the concave point distance, and drawing the segmentation line between the two convex points by the straight line equation;

[0159] Equation of a Line:

[0160] y=mx+b

[0161] Where m is the slope, b is the intercept, Draw connecting lines, which are the dividing lines between adhesion cells, where (x min ,y min ) represents a convex point in a small area, (x max ,y max ) represents the convex point with the largest area; x represents the concave point distance;

[0162] Get the minimum Euclidean distance by traversing each Euclidean distance;

[0163] After obtaining the segmentation line, the adherent cells can be segmented by software to obtain a segmented binary image. The result is as follows: Figure 4 As shown in (c) and (d);

[0164] S4, marking the binary images after the segmentation operation, calculating the centroid, area and perimeter of each binary image after the segmentation operation, obtaining the coordinate information and area information of each cell in the binary image after the segmentation operation, and storing them in an array;

[0165] The centroid is the geometric center of the cell region and the weighted average position of all pixels in the region; the centroid coordinates (C x ,C y ) is calculated as:

[0166]

[0167] In the formula, S' represents the pixel point of the area;

[0168] The area is obtained by the number of pixels in the region;

[0169] The perimeter is the length of the region boundary, obtained using a chain code based on boundary pixels;

[0170] S5. Traverse all the calculated cells in the array, obtain the area information of each cell, calculate the centroid coordinates and cell radius, use the viscircles function to draw the cell outline, and mark the cell number next to each outline. The identification result of the growth trend of Haematococcus pluvialis is obtained through the size of the cell outline, such as Figure 8 As shown;

[0171] Each cell is approximately a circle, and the relationship between the cell area S and the radius r of the circle is: S = πr 2 ; Draw the contour according to the radius, the center coordinate of the cell (C x' ,C y' ) represents the centroid of the cell;

[0172] The equation of a circle is:

[0173] (x'-C x ) 2 +(y'-C y ) 2 =r 2

[0174] Where (x', y') is the coordinate of any point on the circle;

[0175] To make it clear to the human eye, each cell is numbered and the cell number is placed on the left side of the cell circle at (C x -r,C y ) to ensure that the number does not overlap the circle.

[0176] In summary, the present invention provides an identification method for monitoring the growth trend of Haematococcus pluvialis, so as to realize efficient, rapid and non-destructive monitoring of the growth status of Haematococcus pluvialis under different conditions. Figure 8 As can be seen from the result graph, the number of cells and the area information of each cell can be calculated simply by reading in the picture of the algae cells. During the experiment, only pictures of the algae cells at different growth stages need to be taken to quickly analyze their growth trends, avoiding the use of conventional complex detection methods such as chemistry and biology.

[0177] This application uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for monitoring the growth trend of Haematococcus pluvialis, characterized in that: The following steps are involved: S1. Perform image edge detection and image enhancement on the cell image of Haematococcus pluvialis during illumination; S2, removing all connected regions smaller than a given area in the binary image from the image processing result of S1 according to morphological operations; S3, performing concave point detection on the image processed in S2 to obtain adhesion cells, and performing segmentation operation on the adhesion cells; S4, marking the binary images after the segmentation operation, calculating the centroid, area and perimeter of each binary image after the segmentation operation, obtaining the coordinate information and area information of each cell in the binary image after the segmentation operation, and storing them in an array; S5. Traverse all calculated cells in the array, obtain the area information of each cell, calculate the centroid coordinates and cell radius, use the viscircles function to draw the cell outline, and mark the cell number next to each outline to obtain the identification result of the growth trend of Haematococcus pluvialis.

2. A method for monitoring the growth trend of Haematococcus pluvialis according to claim 1, characterized in that: The steps of performing image edge detection and image enhancement on the cell image during the illumination of Haematococcus pluvialis are as follows: S1.1, crop the target area of ​​the cell image of Haematococcus pluvialis during illumination using the imcrop function; S1.2, perform HSV processing on the cropped image; The HSV processing is as follows: extracting the three channels H, S, and V in the image; wherein H represents the hue channel; S represents the saturation channel; and V represents the lightness channel; S1.

3. Binarize the saturation channel in the HSV processed image and perform the Canny edge detection algorithm to obtain the enhanced image. The steps are as follows: S1.3.

1. Use the Sobel operator to calculate the gradient of the image and obtain the gradient magnitude and gradient direction; First, the input image is smoothed by Gaussian filtering to remove noise. The expression for convolution using Gaussian filter is as follows: I(x,y)=I0(x,y)*G(x,y) Where x and y are the pixel coordinates in the image; σ is the standard deviation of the Gaussian kernel, which determines the degree of smoothing; I0(x,y) is the original image; G(x,y) is the Gaussian filter; The horizontal gradient is used to detect the horizontal edge in the image, and the vertical gradient is used to detect the vertical edge in the image; the template G of the horizontal gradient Sobel operator convolution kernel x And the template G of the Sobel operator convolution kernel in the vertical direction y The expression is as follows: Use the Sobel operator to perform convolution to calculate the horizontal direction of the gradient G' x and vertical direction G' y , the expression is as follows: G' x =I(x,y)*G x ,G' y =I(x,y)*G y ; The gradient amplitude expression of the image at each pixel position is as follows: The expression of the gradient direction is as follows: S1.3.2, perform non-maximum suppression operation on the gradient amplitude according to the gradient direction angle; S1.3.3, performing maximum and minimum normalization operation on the gradient amplitude after the non-maximum suppression operation, and dividing the gradient amplitude after the maximum and minimum normalization operation into edge points by setting a high threshold and a low threshold; S1.4, performing a binarization operation on the saturation channel image; S1.

5. Adding the results of the binarization operation and the Canny edge detection algorithm can not only improve the accuracy of image segmentation and the quality of edge recognition, but also reduce noise and simplify the amount of data, thereby achieving cell image enhancement.

3. A method for monitoring the growth trend of Haematococcus pluvialis according to claim 1, characterized in that: The steps of removing all connected regions smaller than a given area in the binary image from the image processing result of S1 according to the morphological operation are as follows: S2.1, performing corrosion operation on the image; A disk-shaped structure element with a radius of 2 pixels is used to perform an erosion operation on the S1 result to eliminate small areas in the image. The calculation formula for the erosion operation is: (A!B)(x,y)=min (s,t)∈B {A(x+s,y+t)} Where A represents the input image; B represents the structural element; A(x, y) is the pixel value of the input image, (s, t) is the offset coordinate of the structural element B, and erosion is used to eliminate small areas or thin lines in the image; A(x+s, y+t) represents the erosion of the input image under the structural element; S2.2, dilate the image processed by S2.1 using a disk-shaped structure element with a radius of 1 pixel; The expansion operation expression is as follows: Where A(xs, yt) represents the dilation of the input image under the structural element; S2.3, performing a filling operation on the image processed by S2.2; S2.

4. For the image processed by S2.3, all connected regions smaller than a given area are removed using a preset area threshold.

4. The method for monitoring the growth trend of Haematococcus pluvialis according to claim 1, characterized in that: The steps of performing concave point detection on the image obtained after processing in S2 to obtain adhesion cells and performing segmentation operation on the adhesion cells are as follows: S3.1, based on the result of S2.4, remove cells that are adhered to the border using a preset border adhesion threshold; S3.

2. After removing the cells adhering to the border, the adhering cells are obtained by using a preset adhering boundary threshold, and the adhering cells are divided. The steps are as follows: S3.2.1, setting a preset adhesion demarcation threshold to obtain adhesion cells; S3.2.2, the concave point distance is obtained by calculating the minimum Euclidean distance between the convex points of the adhesion cells; Euclidean distance calculation formula: Wherein, P1 = (x1, y1) and P2 = (x2, y2) are the coordinates of convex point I and convex point II respectively; S3.2.3, obtaining the pixel points of two convex points between the adhesion cells by the concave point distance, and drawing the segmentation line between the two convex points by the straight line equation; Equation of a Line: y=mx+b Where m is the slope, b is the intercept, Draw connecting lines, which are the dividing lines between adhesion cells, where (x min ,y min ) represents a convex point in a small area, (x max ,y max ) represents the convex point with the largest area; x represents the concave point distance; After obtaining the segmentation line, the adherent cells can be segmented by software to obtain a segmented binary image.

5. The method for monitoring the growth trend of Haematococcus pluvialis according to claim 1, characterized in that: The binary image after the segmentation operation is marked, and the centroid, area and perimeter of each binary image after the segmentation operation are calculated to obtain the coordinate information and area information of each cell in the binary image after the segmentation operation, and the centroid stored in the array is the geometric center of the cell area, which is the centroid coordinate (C x ,C y ) is calculated as: Where S' represents the pixel point of the area.

Images