CN113313817B - A three-dimensional reconstruction method and application of leather fiber bundles based on MCT slice images - Google Patents

Abstract

A three-dimensional reconstruction method of leather fiber bundles based on MCT slice images selects a fiber bundle area as a seed area in one frame image of the MCT slice images of leather fibers, and adopts a forward tracking method and a reverse tracking method to track the extension of the fiber bundles in different longitudinal cross section areas of the same fiber bundle with the selected seed area, so as to finally obtain the cross section areas belonging to the same fiber bundle. The method adopts a forward tracking method and a reverse tracking method to track the cross section areas belonging to the same fiber bundle, and finally obtains the fiber areas belonging to the same fiber bundle for reconstructing the three-dimensional tissue form of the leather fiber bundle.

Description

A three-dimensional reconstruction method and application of leather fiber bundles based on MCT slice images

technical field

The invention discloses a three-dimensional reconstruction method and application of leather fiber bundles based on MCT slice images, and belongs to the technical field of leather image reconstruction.

Background technique

The weave structure of leather fibers has always been a concern of the industry. Until now, the understanding of the weave structure of leather fibers has been superficial. Research on the weaving structure of leather fibers has important theoretical value and application value.

Some researchers use CT technology and metallographic method to obtain serial cross-sectional images of leather fibers, and perform image processing and analysis such as denoising, registration, and segmentation on the obtained images to obtain three-dimensional reconstruction of leather fiber bundles and information about leather fiber bundles. Results of some geometric morphology studies.

The author Zhang Huayong published "Research on 3D Reconstruction of Leather Fiber Weave Network" on December 16, 2015. Micro-CT and metallographic methods were used to obtain serial slice images of leather fibers. The three-dimensional digital model of the woven structure of collagen fibers in chrome-tanned bovine leather was established by using the structural technology. With the help of 3D-Doctor software, the obtained serial slice images of leather fibers were reconstructed in three dimensions, laying a foundation for the research on the correlation between leather structure and properties. key foundation. And further through the three-dimensional digital model, the morphology, weaving method, bending angle and other laws of collagen fiber bundles were studied, and it was found that it has a fractal law. By comparing the fractal dimension calculation methods, the island method is optimized to calculate the fractal dimension of the leather fiber bundle section, using the perimeter and area data of the fiber bundle section in the sequence image, and also using the surface area and the fiber bundle in the three-dimensional digital model. From the volume data, the fractal dimension of the surface of the fiber bundle in the reconstructed body was obtained by calculating the relationship between the fractal surface and the space area enclosed by it.

On May 20, 2017, the author Miao Changkun published "Research on Three-dimensional Reconstruction of Chrome-tanned Sheepskin Fiber Network", which records that based on the special structural characteristics of chrome-tanned goat leather blanks (referred to as chrome-tanned sheepskin, the same below), through the metallographic method The serial cross-sectional images of the chrome-tanned sheepskin were obtained to construct a three-dimensional digital model of the chrome-tanned sheepskin fiber network, which laid a foundation for the preliminary exploration of the relationship between its structure and properties. It is clearly recorded in the literature that appropriate image preprocessing software is used to denoise and enhance the acquired sequence images; the registration software of the self-developed registration software is used to realize the registration of the sequence two-dimensional images; 3D reconstruction. By marking each individual fiber bundle with a different color and numbering, a separate display of the fiber bundles individually or as a whole is achieved.

However, an obvious defect of the three-dimensional reconstruction of leather fiber bundles based on metallographic method is that the leather samples used to obtain images need to undergo a resin solution soaking and curing process, and this process will lead to deformation of leather fibers, thus affecting the three-dimensional reconstruction. Effect. Another defect of the metallographic method is that the distance (ie layer thickness) between the serial cross-sectional images obtained by the metallographic method is large, the number of images is small, and the continuity is poor. The image registration of the sequence images obtained in this way is also a difficult task, which is sometimes labor-intensive, and there is no satisfactory solution yet.

3D-Doctor software is an image processing software developed by Able Software in the United States that can be used for 3D reconstruction of serial cross-sectional images of MRI, CT, PET and other types, and has been widely recognized and used. But the 3D reconstruction process of 3D-Doctor is based on pre-segmentation and annotation. For some organizational structures that are easy to segment and label, 3D-Doctor's 3D reconstruction effect is very good, but for some organizational structures that are difficult to segment and label, 3D-Doctor is powerless. Due to the difficulty in distinguishing different fiber bundles of leather, before using 3D-Doctor for three-dimensional reconstruction, the fiber bundles must be marked first, and the cross-sectional areas belonging to the same fiber bundle are marked with the same mark. This work usually requires manual labor. Do it manually. Therefore, the 3D reconstruction of single fiber bundles of leather fibers with 3D-Doctor is less efficient. For the 3D reconstruction based on the cross-sectional image of leather fiber sequence, fast and effective algorithm design is very important.

The images obtained by metallography can display finer parts, but the number of images is small, and the difficulty and workload of image preprocessing are large. Micro X-ray tomography (MCT) technology can obtain slice images of the internal structure of the sample without destroying the sample. The three-dimensional reconstruction of leather fibers can be performed by using the MCT slice images of leather fibers, thereby showing the weaving structure and regularity of leather fibers. The three-dimensional reconstruction algorithm and technology based on the MCT sequence images of leather fibers have important application value for studying the structure of leather fibers.

SUMMARY OF THE INVENTION

In view of the deficiencies of the prior art, the present invention discloses a three-dimensional reconstruction method and application of leather fiber bundles based on MCT slice images. The fiber regions belonging to the same fiber bundle were obtained, which were used to reconstruct the three-dimensional structure of the leather fiber bundle.

Detailed technical scheme of the present invention

A three-dimensional reconstruction method of leather fiber bundles based on MCT slice images, characterized in that: a fiber bundle area is selected as a seed area in a frame of images in the MCT slice images of leather fibers; The tracking method tracks the extension of the fiber bundle in the longitudinal different cross-sectional areas of the same fiber bundle with the selected seed area, and finally obtains the cross-sectional area belonging to the same fiber bundle. The fiber bundle reconstructed from the same root area is called the same root. fiber bundle;

The forward tracking method refers to:

1) Select a seed area in the first frame image, which is the root area, and search for the area in the second frame image that has a common part with the seed area of the previous frame image in the forward order:

If there is an area in the second frame image that has a common part with the seed area of the previous frame image, this area is segmented, and this area is used as the seed area of the second frame image, and this area is also the root of the previous frame image. The area with the same root of the area; otherwise, the forward tracking process ends;

2) with the same root area in the second frame image as the new seed area, search for the same root area in the next frame image;

3) Continue like this until there is no new area with the same root;

The reverse tracking method refers to:

1) Take the fiber bundle area of the last frame image obtained by the forward tracking method as the seed area, search for the area that has a common part with the seed area of the previous frame image in the adjacent second frame image in reverse order, and use the The region is divided from the original image and merged with the region of the same frame image obtained by the forward tracking method as the seed region of the second frame image, which must be the same as the first frame image selected by the forward tracking method. The root zone is the same root zone;

2) the seed area of the second frame image obtained is used as the new seed area, searches the same root area in the next frame image in reverse order;

3) Go on like this until the first frame image specified by the forward tracking method ends.

The forward tracking algorithm will lose some backward branches of the fiber bundle, while the reverse tracking algorithm can search for some backward branches with the same root. The fiber bundle represented by the dotted line in Figure 1 contains such fibers that have not been tracked yet. Bundles of inverted branches. In order to track such fiber bundles, such backward branches can be tracked by reverse tracking after the forward tracking process, as shown by the more solid lines in FIG. 2 than in FIG. 1 .

According to a preferred embodiment of the present invention, the method for three-dimensional reconstruction of leather fiber bundles based on MCT slice images further includes performing forward tracking and reverse tracking on the MCT slice images of leather fibers in turn multiple times until no new tracking can be found. up to the cross-sectional area of the fiber bundle. The algorithm refers to performing forward tracking with the fiber bundle region of the first frame cross-sectional image segmented by the above process as a seed region, and then performing reverse tracking. In this way, the forward tracking and reverse tracking of the MCT slice images of leather fibers are performed alternately for many times until the new fiber bundle cross-sectional area cannot be tracked. As shown in Figures 3 and 4.

Preferably according to the present invention, the MCT slice image of the leather fiber is the MCT slice image of the leather fiber after denoising.

The method disclosed in the present invention is suitable for three-dimensional reconstruction of in-situ MCT slice images of leather fibers and MCT slice images embedded in leather fibers.

Technical advantages of the present invention:

1. A three-dimensional reconstruction method of leather fiber bundles based on MCT slice images according to the present invention, by tracking a certain fiber bundle in the MCT slice images of leather fibers to the same root area by frame image, the leather fiber can be rapidly reconstructed. The organizational structure of the three-dimensional bundle, showing the three-dimensional shape of the leather fiber bundle.

2. Since the forward tracking algorithm will lose some backward branches of the fiber bundle, the present invention introduces a backward tracking method on the basis of forward tracking, and reversely tracks the same root area for frame-by-frame images.

3. The combination of forward tracking method and reverse tracking method is called bidirectional tracking method. The bidirectional tracking algorithm is repeatedly performed on the MCT slice images of the leather fibers to be reconstructed until no new regions with the same root appear, so that the three-dimensional shape of the leather fibers is more complete, clear and accurate.

Description of drawings

1-4 are schematic diagrams of forward tracking and reverse tracking of the same root region;

Figure 1 is a schematic diagram of the same root region obtained by the first forward tracking from the root region, wherein the solid line part is the same root fiber bundle obtained by the first forward tracking, and the dotted line part is the first forward tracking failed to track In the same fiber bundle area, the arrow to the right is a forward tracking arrow;

Figure 2 is a schematic diagram of the same root region obtained by reverse tracking after the first forward tracking, wherein the solid line part is the same root fiber bundle tracked after the first forward tracking and the first reverse tracking, and the dotted line part is the first forward tracking and the first reverse tracking. For the area of the same fiber bundle that has not been traced after the first reverse tracking, the arrow to the right is an indicator arrow for forward tracking, and the arrow to the left is an indicator arrow for reverse tracking;

Figure 3 is a schematic diagram of the same root region obtained by performing forward tracking again, wherein the solid line part is the fiber bundle in the same root region, the dotted line part is the same root fiber bundle region that has not been tracked, and the rightward arrow is a forward tracking indicator arrow, The arrow pointing to the left is the reverse tracking arrow;

Figure 4 is a schematic diagram of the same root region obtained by reverse tracking again, wherein the solid line part is the same root fiber bundle, the dotted line part is the same root fiber bundle region that has not been tracked, the right arrow is the forward tracking indicator arrow, the left The arrow is the reverse tracking arrow;

Figures 5 and 6 are several bundles of leather fiber bundles reconstructed by using the in-situ MCT slice images of leather fibers obtained after forward tracking and reverse tracking according to the present invention.

Detailed ways

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

Example 1

A three-dimensional reconstruction method of leather fiber bundles based on MCT slice images, selecting a fiber bundle area as a seed area in a frame of images in the MCT slice images of leather fibers;

The forward tracking method is used to track the extension of fiber bundles in different longitudinal cross-sectional areas belonging to the same fiber bundle, and finally the cross-sectional areas belonging to the same fiber bundle are obtained.

The forward tracking method includes:

1) Select a seed area in the first frame image, which is the root area, and search for the area in the second frame image that has a common part with the seed area of the previous frame image in the forward order:

If there is an area in the second frame image that has a common part with the seed area of the previous frame image, this area is segmented, and this area is used as the seed area of the second frame image, and this area is also the root of the previous frame image. The same root region of the region; otherwise, the forward tracking process ends;

2) with the same root area in the second frame image as the new seed area, search for the same root area in the next frame image;

3) Go on like this until there is no new same-root area, as shown in Figure 1.

Then, the inverse tracking method is used to track the extension of fiber bundles in different longitudinal cross-sectional areas belonging to the same fiber bundle, and finally the three-dimensional regions of fibers belonging to the same fiber bundle are obtained, specifically:

1) Take the fiber bundle area of the last frame image obtained by the forward tracking method as the seed area, search for the area that has a common part with the seed area of the previous frame image in the adjacent second frame image in reverse order, and use the The region is divided from the original image and merged with the region of the same frame image obtained by the forward tracking method as the seed region of the second frame image, which must be the same as the first frame image selected by the forward tracking method. The root zone is the same root zone;

2) the seed area of the second frame image obtained is used as the new seed area, searches the same root area in the next frame image in reverse order;

3) Go on like this until the first frame image specified by the forward tracking method ends; as shown in Figure 2.

Example 2

A method for three-dimensional reconstruction of leather fiber bundles based on MCT slice images as described in Embodiment 1, further comprising performing forward tracking and reverse tracking on the MCT slice images of leather fibers alternately multiple times until new fibers cannot be tracked the beam section area. As shown in Figures 3 and 4. The combination of forward tracking and reverse tracking is called bidirectional tracking. In this process, the new co-rooted region of each frame image must contain the same-rooted region obtained by the forward tracking search process. The two-way tracking algorithm is repeatedly implemented until no new regions with the same root appear.

Application example 1

Utilize the method described in embodiment 1 to carry out image acquisition and processing according to the following steps:

1) Skin-like material: a small amount of fibrous tissue peeled off from the dried blue skin of American bovine buttocks;

2) MCT imaging equipment:

MCT tomography scanner: SkyScan2211; Camera lens: MX11002;

3) MCT tomography image parameters:

Resolution: 150nm; Pixels: 4032×4032; Layer Thickness: 0.31μm; Frames: 2357;

The acquired 2357 frames of MCT slice images are denoised (the MCT slice images of the leather fibers are the MCT slice images of the denoised leather fibers), and part of the fiber bundles are reconstructed three-dimensionally by the bidirectional tracking algorithm, As shown in Figure 5.

Application example 2

Utilize the method described in embodiment 2 to carry out image acquisition and processing according to the following steps:

1) Skin-like material: dried cow head blue skin;

2) MCT imaging equipment

MCT tomography scanner: SkyScan2211; Camera lens: MX11002;

3) MCT tomography image features

Pixels: 4032×4032; Layer thickness: 0.31μm; Frames: 1578;

The acquired 1578 frames of MCT slice images were denoised (the MCT slice images of the leather fibers are the denoised MCT slice images of the leather fibers), and a bidirectional tracking algorithm was used to reconstruct some of the fiber bundles in three dimensions. As shown in Figure 6.

Example 3

The methods described in Examples 1 and 2 are also applicable to the three-dimensional reconstruction of in-situ MCT slice images of leather fibers and MCT slice images embedded in leather fibers.

Claims (3)

1. a three-dimensional reconstruction method based on the leather fiber bundle of MCT slice image, is characterized in that: in the MCT slice image of leather fiber, in the slice image of a frame of leather fiber, select a fiber bundle area as seed area, The forward tracking method and the reverse tracking method are used to track the extension of the fiber bundle in the longitudinal different cross-sectional areas of the same fiber bundle with the selected seed area, and finally the fiber fiber area belonging to the same fiber bundle is obtained; The forward tracking method refers to: 1-1) Select a seed area in the first frame image, which is the root area, and search for the area in the second frame image that has a common part with the seed area of the previous frame image in the forward order: If there is an area in the second frame image that has a common part with the seed area of the previous frame image, this area is segmented, and this area is used as the seed area of the second frame image, and this area is also the root of the previous frame image. The same root region of the region; otherwise, the forward tracking process ends; 1-2) Take the same root area in the second frame image as a new seed area, and search for the same root area in the next frame image; 1-3) Repeat the above 1-1)-1-2) tracking process until there is no new same root area; The reverse tracking method refers to: 2-1) Take the fiber bundle area of the last frame image obtained by the forward tracking method as the seed area, and search for the area in the adjacent second frame image that has a common part with the seed area of the previous frame image in reverse order, This area is divided from the original image and merged with the area of the second frame image obtained by the forward tracking method as the seed area of the second frame image, which must be the same as the first frame selected by the forward tracking method. The root region in the image is the same root region; 2-2) The obtained seed area of the second frame image is used as a new seed area, and the same root area in the next frame image is searched in reverse order; 2-3) Repeat the above 2-1)-2-2) tracking process until the first frame image specified by the forward tracking method is completed; The three-dimensional reconstruction method also includes performing forward tracking and reverse tracking on the MCT slice image of the leather fiber alternately for many times until the new fiber bundle cross-sectional area cannot be tracked; the combination of the forward tracking method and the reverse tracking method is called. In the two-way tracking method, the new same-root area of each frame image must contain the same-root area obtained by the forward tracking search process, and the two-way tracking method is repeatedly performed until no new same-root area appears. 2 . The method for three-dimensional reconstruction of leather fiber bundles based on MCT slice images according to claim 1 , wherein the MCT slice images of the leather fibers are denoised MCT slice images of leather fibers. 3 . 3. The application of the method according to any one of claims 1-2: suitable for three-dimensional reconstruction of in-situ MCT slice images of leather fibers and MCT slice images embedded in leather fibers.

Images