CN112164085A

CN112164085A - Fiber image segmentation and diameter statistical method based on image processing

Info

Publication number: CN112164085A
Application number: CN202011036903.7A
Authority: CN
Inventors: 刘旺玉; 吴霖; 谢卫规
Original assignee: South China University of Technology SCUT
Current assignee: South China University of Technology SCUT
Priority date: 2020-09-28
Filing date: 2020-09-28
Publication date: 2021-01-01
Anticipated expiration: 2040-09-28
Also published as: CN112164085B

Abstract

The invention discloses a fiber image segmentation and diameter statistical method based on image processing, which comprises the following steps: 1) reading the microstructure image of the nano cellulose fiber obtained by SEM shooting; 2) converting the fiber microstructure image of the nanocellulose from an RGB format into a gray level image, and then performing binarization processing on the gray level image to obtain a binary image BW; 3) removing the micro areas which do not meet the requirements in the binary image BW by using a morphological algorithm to obtain an image BW 1; 4) image segmentation; 5) and counting the width. The invention solves the problems of troublesome extraction of key information of the existing SEM fiber diagram and complicated manual repeated measurement and statistics of the fiber diameter, the generated data can be used for generating a corresponding fiber random model, and the problem of excessive interference areas in the existing cellulose binarization processing is solved.

Description

Fiber image segmentation and diameter statistical method based on image processing

Technical Field

The invention relates to the technical field of image segmentation and fiber diameter measurement and statistics, in particular to a fiber image segmentation and diameter statistical method based on image processing.

Background

The cellulose is a natural polymer which is visible everywhere in nature and has huge reserves, is a renewable resource with huge yield, has the advantages of complete biodegradation, no toxicity, no pollution, easy modification, good biocompatibility and the like, is considered as a main raw material of world energy and chemical industry in the future, and can be widely used in the fields of papermaking, fibers, textiles, cellulose derivatives and the like.

The paper making technology is a long-standing technology, can be applied to many fields, such as manufacturing lithium battery diaphragms, and has different requirements on manufactured paper structures in different fields or different aspects of the same field, so that research on the paper structures with different dimensions has been carried out for many years. Thus, cellulose, a material used in papermaking, has been studied for many years on a macro, micro, or molecular scale. For many years, researchers have established fiber models on the basis of single fibers by considering the interference of surrounding fibers, and generated two-dimensional woven fiber structures by using a computer technology; researchers have also used computer technology to randomly generate two-dimensional or three-dimensional fibrous structures. However, the model established by the two methods still has a large difference with the actual structure of the fiber, and cannot truly reflect the structure of the fiber, and only can be applied to some simple qualitative simulation analysis. Part of the reasons for this result are that some parameter statistics are manually measured many times during the processing of the microscopic images of the fibers obtained by shooting and the parameter statistics, so that enough and more accurate statistical data cannot be obtained (penlup research on multi-scale structures and mechanical properties of cellulose membranes of lithium ion batteries).

Disclosure of Invention

The invention aims to provide a fiber image segmentation and diameter statistical method based on image processing by selecting a diameter parameter from a plurality of parameters of cellulose aiming at the defects of the prior art, and the method realizes the statistics of the diameter of the cellulose through operations such as binarization processing, morphological processing, skeletonization, edge extraction and the like, thereby obtaining enough and more accurate diameter data, solving the problems of complexity and inaccuracy of manual measurement and statistics for many times, facilitating the subsequent establishment of a two-dimensional or three-dimensional random model of the cellulose to provide favorable help and enabling the subsequent cellulose structure model to be more fit with an actual fiber structure.

The invention is realized by at least one of the following technical schemes.

A fiber image segmentation and diameter statistical method based on image processing comprises the following steps:

1) reading, by software, a nanocellulose structure image obtained by Scanning Electron Microscope (SEM);

2) converting the image from an RGB format into a gray image, and then carrying out binarization processing on the gray image to obtain a binary image BW;

3) removing regions which do not meet the requirements in the binary image BW by using a morphological algorithm to obtain an image BW 1;

4) segmenting the image obtained by the processing in the step 3);

5) and carrying out width statistics on the patterns in the segmented images, and establishing a two-dimensional or three-dimensional random model of the cellulose.

Preferably, the binarization processing in step 2) is to divide the grayscale image into eight binary pictures by using a bitmap layer layering algorithm, and select an eighth layer bitmap layer containing the most image information as a target picture for further processing.

Preferably, the removing process in step 3) is performed before and after inverting the binarized image BW by a bwaeopen function of Matlab.

Preferably, the image segmentation in step 4) is implemented by extracting the central skeleton and the edge of each fiber one by one as a basis, and the specific process is as follows:

4.1) carrying out connected domain numbering on the BW1 images subjected to the removal processing, and extracting one by one according to the numbering sequence for carrying out the next operation;

4.2) extracting an edge line BW1_1 and a central skeleton BW1_2 of the ith connected domain of the BW1 image by using a remove module and a skin module of a bwmorphh function of Matlab;

4.3) judging the connectivity of all points in the central framework BW1_2 of the BW1 image, and judging whether each point is two-connectivity or not;

4.4) measuring the distance between the stored skeleton crossing points, and removing points which do not meet the distance requirement;

4.5) the finally saved framework cross points are used as circle centers one by one, and the edge point closest to the point on the edge line BW1_1 is obtained through a scanning algorithm;

4.6) connecting the edge points with the corresponding framework cross points to form a line and extending the line so as to achieve the cutting effect and finally obtain a cut image BW 2.

Preferably, the judging process in step 4.3) is as follows:

4.3.1) if not, removing the point, and continuing to judge the connectivity of the next point until the judgment of the connectivity of all the points is finished;

4.3.2) if yes, storing the point, and continuing to judge the connectivity of the next point until the judgment of the connectivity of all the points is finished.

Preferably, the width statistics in step 5) is implemented by extracting the central skeleton and the edge of each fiber one by one, so as to realize the statistics of the width of the pattern in the picture.

Preferably, the specific process of step 5) is as follows:

5.1) carrying out connected domain numbering on the cut image BW2, and extracting the images one by one according to the numbering sequence for carrying out subsequent column operation;

5.2) extracting an edge line BW1_1 and a central skeleton BW1_2 of the ith connected domain of the BW2 image by using a remove module and a skin module of bwmorphh;

5.3) randomly taking a certain number of points on a central framework BW2_2 of the BW2 image, and obtaining an edge point closest to the test point on an edge line BW2_1 through a scanning algorithm;

5.4) measuring the distance r between the test point and the corresponding edge point to obtain the diameter d of the fiber at the position as 2 r; when the width statistics are complete, different fiber diameter data can be obtained.

The method realizes the segmentation and diameter statistics of the cellulose image by performing binarization, morphological processing, edge and skeleton extraction, node extraction and other processing on the SEM image. The binarization is an algorithm for extracting an eighth layer which contains most image information in a bitmap layer; the morphological processing uses a bweraopen function to remove the regions that are not satisfactory. The problems that the key information of the existing SEM fiber diagram is troublesome to extract and the fiber diameter is difficult to measure and count manually for multiple times are solved, and the generated data can be used for generating a corresponding fiber random model.

Compared with the prior art, the invention has the following advantages and beneficial effects:

1. the binarization algorithm adopted by the invention is to extract the eighth layer containing the most image information according to the bit map layer, and remove small areas influencing the subsequent parameter statistics by applying morphological processing, and the binary image finally obtained after processing only retains cellulose with statistical significance, thereby solving the problem of excessive interference areas in the existing cellulose binarization processing, and simultaneously avoiding the inaccurate fiber diameter measurement caused by excessive detail information loss of the binarization picture due to corrosion and expansion because of excessive morphological processing.

2. The image segmentation in the invention relates to the steps of carrying out node identification and node connectivity judgment on the extracted skeletonized image, finding the corresponding point with the shortest distance from the node to the extracted edge line, finally carrying out the connection and extension of the node and the edge point to realize the segmentation, and finally obtaining the fiber binary image without the intersection point, thereby avoiding the problem of inaccurate measurement caused by excessive fiber branches in the subsequent width statistics and leading the finally obtained data to be more persuasive.

3. The width statistics in the invention relates to the calculation of the diameter of the fiber by measuring the shortest distance from the extracted skeletonized image to the edge line image through a scanning algorithm, so that the diameter of each measuring position is the shortest diameter of the fiber at the point, and the result is more referential.

Drawings

FIG. 1 is a flowchart illustrating a fiber image segmentation and diameter statistics method based on image processing according to an embodiment;

FIG. 2a is a schematic view of an SEM image of the present embodiment;

fig. 2b is a schematic diagram of a binary image BW in the present embodiment;

fig. 2c is a schematic diagram of an image BW1 subjected to morphological processing and inverted in the embodiment;

FIG. 2d is a schematic diagram of a random piece of cellulose in the present embodiment;

fig. 2e is a schematic diagram of the edge line image BW1_1 of the present embodiment;

fig. 2f is a schematic diagram of the central skeleton image BW1_2 of the present embodiment;

FIG. 2g is a schematic view of a cut fiber of the present embodiment;

fig. 2h is a schematic diagram of the completely cut image BW2 of the present embodiment;

FIG. 2i is a schematic view of a connecting line image of the present embodiment, which is obtained by scanning to find a random point of the central skeleton and the shortest distance from the edge;

FIG. 2j is a statistical chart of the fiber diameter parameters finally obtained in this example.

Detailed Description

The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.

As shown in fig. 1, the present embodiment provides a method for reconstructing a two-dimensional structure of a cellulose fiber of paper based on image processing, which includes the following steps:

1) reading the image of the microstructure of the nanocellulose fibers obtained by SEM (as shown in fig. 2a) by Matlab software;

2) converting the fiber microstructure image of the nanocellulose from an RGB format into a gray level image, and then performing binarization processing on the gray level image to obtain a binary image BW (as shown in FIG. 2 b);

the binarization processing is to divide the grayscale image into eight binarized pictures by using an algorithm of bit layer layering (okalays digital image processing), and select the eighth layer bit layer containing the most image information as a target picture for subsequent processing.

3) Removing the micro areas which do not meet the requirements in the binary image BW by using a morphological algorithm to obtain an image BW1 (as shown in figure 2 c); the removal processing is respectively performed before and after the inversion of the binary image BW through a bweraopen function of Matlab. The purposes are respectively as follows: and small areas which do not meet the conditions are removed through a bwaeopen function, and redundant miscellaneous points in the fibers are removed, so that the subsequent processing is facilitated.

4) Image segmentation;

4.1) numbering the BW1 images in a connected domain, and extracting the BW1 images one by one according to the numbering sequence (as shown in figure 2d) for the next operation;

4.2) extracting an edge line BW1_1 (shown in figure 2e) and a central skeleton BW1_2 (shown in figure 2f) of the ith connected domain by using a remove module and a skin module in a bwmorphh function of Maltab;

4.3) judging the connectivity of all points in the central framework BW1_2, and judging whether each point is two-connectivity:

4.3.2) if yes, storing the point, and continuing to judge the connectivity of the next point until the judgment of the connectivity of all the points is finished;

4.4) measuring the distance between the stored skeleton intersections, and removing other skeleton intersections with the distance between the three pixels;

4.6) connecting the edge points with the corresponding skeleton crossing points to form a line and extending the line to achieve the cutting effect (as shown in figure 2g), and finally obtaining a cut image BW2 (as shown in figure 2 h).

5) And the width statistics is used for establishing a two-dimensional or three-dimensional random model of the cellulose.

5.1) numbering the BW2 images in a connected domain, and extracting the images one by one according to the numbering sequence for carrying out subsequent column operation;

5.2) extracting an edge line BW1_1 and a central skeleton BW1_2 of the ith connected domain by using a remove module and a skin module of bwmorphh;

5.3) randomly taking 20 points from the central framework BW2_2, and obtaining the edge point closest to the test point on the edge line BW2_1 by a scanning algorithm;

5.4) measuring the distance r between the test point and the corresponding edge point, and obtaining the diameter d of the fiber at the position as 2 r (shown in figure 2 i).

When the width statistics algorithm is completed, a sufficient amount of fiber diameter can be obtained (see fig. 2 j).

The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can substitute or change the technical solution of the present invention and the inventive concept within the scope of the present invention disclosed by the present invention.

Claims

1. A fiber image segmentation and diameter statistical method based on image processing is characterized in that: the method comprises the following steps:

4) segmenting the image obtained by the processing in the step 3);

2. The fiber image segmentation and diameter statistic method based on image processing as claimed in claim 1, wherein: the binarization processing in the step 2) is to divide the gray level image into eight binarization pictures by using a bit layer layering algorithm, and select an eighth layer bit layer containing most image information as a target picture to perform the next processing.

3. The fiber image segmentation and diameter statistic method based on image processing as claimed in claim 2, wherein: the removing process in the step 3) is to remove the binary image BW before and after inverting the binary image BW through a bweraopen function of Matlab.

4. The fiber image segmentation and diameter statistic method based on image processing as claimed in claim 3, wherein: the image segmentation in the step 4) is realized by extracting the central skeleton and the edge of each fiber one by one as a basis, and the specific process is as follows:

5. The fiber image segmentation and diameter statistic method based on image processing as claimed in claim 4, wherein: the judgment process of the step 4.3) is as follows:

6. The fiber image segmentation and diameter statistic method based on image processing as claimed in claim 5, wherein: the width statistics in the step 5) is to realize the statistics of the pattern width in the picture by taking the central skeleton and the edge of each fiber as the basis one by one.

7. The fiber image segmentation and diameter statistic method based on image processing as claimed in claim 6, wherein: the specific process of the step 5) is as follows: