A grayscale image processing method for HTPB propellant

Abstract

The invention belongs to the field of image processing technology, and discloses a grayscale image processing method for HTPB propellant, comprising the following steps: pre-processing the original image based on information entropy; segmenting the image using a threshold, and obtaining a seed point for regional growth of an AP layer based on an image morphological processing algorithm; restoring the original AP particles in the region where the seed points are located using a regional growth algorithm, and obtaining an AP layer by morphologically processing the restored AP particles; correcting the AP layer, and segmenting the remaining part in layers; re-synthesizing the images of the segmented AP layer, Al layer, matrix layer, and pore layer to obtain a complete image for identifying the component structure of HTPB propellant. The CT image of HTPB propellant obtained after the algorithm of the invention is improved has a better visual effect, and the boundaries between components are clear and easy to identify, and it is easier to observe the distribution and form of damage. It is also possible to perform statistics on each component in the image, which is convenient for propellant research and porosity calculation.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a grayscale image processing method for HTPB propellant. Background Art

[0002] Hydroxybutyl propellant is a composite propellant composed of terminal hydroxyl polybutadiene (HTPB) as the main binder, inorganic oxides, energy chemical additives, isocyanate curing agents and bonding agents. This propellant has good mechanical properties under a wide range of working temperatures, a large adjustable range of burning rate, and a low burning rate pressure index, and is low in cost. This propellant has been used in various tactical, strategic missile technologies, and solid-state engines for space models, and is one of the important types of solid-state engine propellants. In order to study the microscopic mechanical properties, combustion mechanism, and formula improvement of the propellant, it is necessary to scan and reconstruct the interior of the hydroxybutyl propellant to obtain the microscopic morphology of the bonding interface for easy observation and research.

[0003] Usually, micro-CT equipment or synchrotron radiation light source equipment is used to scan and analyze the propellant specimen. The prior art uses Skyscan1172 micro-CT equipment to scan the HTPB propellant specimen, and obtains a set of X-ray projection images, which are stored in the form of 16-bit TIFF files. The X-ray images of this group will constitute a set of digital groups. On this basis, the grayscale image of the scanned specimen can be obtained by using an appropriate reconstruction algorithm, which is convenient for intuitively displaying the internal microstructure of the specimen and distinguishing damage defects. However, the grayscale value ranges of different components in the image obtained by the system's own algorithm generally have intersections, resulting in unclear boundaries between different components and poor visual effects. Especially for the study of microscopic loss defects inside the propellant, the quality and clarity of the image are important factors affecting the accuracy of quantitative damage analysis. Therefore, how to obtain higher quality cross-sectional images is an important issue faced in studying the microscopic mechanical properties of propellants. Summary of the invention

[0004] The present invention aims to provide a grayscale image processing method for HTPB propellant, so as to solve the technical problems in the prior art that the HTPB propellant image processing algorithm has a single processing method, unclear component boundaries and poor composition effect.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A grayscale image processing method for HTPB propellant comprises the following steps:

[0007] S1. Preprocessing the image based on information entropy: denoising the original image obtained after scanning by the CT device;

[0008] S2. Finding regional growth seed points: First, according to the components of HTPB propellant, the image is directly segmented using the gray value threshold to obtain the ammonium perchlorate (AP) part, and then the AP part is processed using the image morphological processing algorithm to obtain the seed points for regional growth of the AP layer;

[0009] S3, obtaining a seed layer: using a region growing algorithm to restore the original AP particles in the region where the seed point is located, and using morphological processing to restore the AP particles to obtain an AP layer;

[0010] S3, AP layer correction: register the images of the threshold segmented basal layer and the AP layer, delete the overlapping area in the AP layer image, and obtain the corrected AP layer;

[0011] S4, layer segmentation: matching and comparing the AP layer with the original image, and performing segmentation processing on other components Al, matrix layer and pores in the remaining part of the original image;

[0012] S5. Image synthesis: The segmented AP layer, Al layer, matrix layer and pore layer images are reassigned to different colors and then matrix-added to obtain a complete image for component structure identification of HTPB propellant.

[0013] Furthermore, before denoising the original image in step S1, the original image needs to be enhanced. The denoising method is based on the neighborhood idea of ​​the mean filter, that is, points are taken one by one from the first pixel to the last pixel of the enhanced image to obtain a certain point. Centered on 8 neighboring pixels , if there are no less than 5 pixel values ​​in the 8 points in the neighborhood of the point that are equal, then the pixel value of the center point is equal to this pixel value.

[0014] Furthermore, the image morphological processing operation in step S2 is specifically to first perform an erosion operation on the AP part directly obtained by threshold segmentation, and then fill the holes of the AP part to fill all the holes, delete the small blocks in the image caused by the erosion of linear and ring noise, and perform an expansion operation on the remaining pixel blocks to make multiple blocks belonging to the same AP particle become a connected domain, and use the function bwmorph (BW, 'shrink', Inf) provided by Matlab to shrink each block to a pixel point as the seed point for the growth of the AP layer area.

[0015] Furthermore, the region growing algorithm in step S3 is specifically designed using the similarity criterion of grayscale gradients. The grayscale image of HTPB propellant is a discrete two-dimensional function, namely

[0016] ;

[0017] That , which represents the grayscale image of HTPB propellant;

[0018] The gradient of the grayscale image of HTPB propellant is the difference between adjacent pixels.

[0019] ;

[0020] Adjust the brightness of the image. Remove some overly bright and dark areas to reduce the time of region growing retrieval; The pixels in the remaining area are grayscale stretched to increase the gradient between pixels, so that the region growth can be more accurate to obtain AP particles.

[0021] Furthermore, the starting point of the seed point of the region growing algorithm is obtained by traversing the entire image from the first point in the AP layer region growth seed group image. If a point (x, y) is a seed point, this point is used as the starting point of the region growing, and the corresponding AP particles are obtained at the position of this point in the original image using the region growing algorithm. The generated particles are morphologically processed and the obtained AP particles are superimposed in the same image until all the seed points are grown.

[0022] Furthermore, the morphological processing in step S3 specifically refers to corroding and reducing a block grown from a seed point to restore it to a complete particle, and then judging whether it belongs to an AP particle based on the area size of the particle. If it meets the scale condition, it is stored in the same position of the new matrix using matrix operation until all seed points are grown and screened, and the AP layer is output.

[0023] Furthermore, the threshold segmentation in step S4 specifically refers to performing threshold segmentation by reducing the gray value range of the matrix (reducing the upper limit and setting the lower limit to 0), and deleting blocks smaller than the hole area in the obtained image of the matrix layer.

[0024] Furthermore, in step S5, the Al layer is segmented to remove the blocks in the Al layer whose area is larger than the maximum cross-sectional area of ​​the Al particles, and the removed parts are classified as the AP layer, and the remaining parts are the final Al layer.

[0025] Furthermore, the segmentation of the matrix layer and the pore layer in step S6 is to directly use threshold segmentation in the remaining part of the image to obtain the matrix layer and the pore layer.

[0026] In summary, the present invention has the following beneficial effects:

[0027] The present invention uses 8-neighborhood processing based on information entropy during image preprocessing, and fuses and combines image processing algorithms such as region growing algorithm, image morphology processing, threshold segmentation, histogram threshold segmentation, etc. to obtain an algorithm for AP layer segmentation processing, so that the image obtained after the algorithm is improved has better visual effect, the boundaries between components are clear and easy to identify, and the distribution and form of damage are easier to observe. In addition, the algorithm improvement not only increases the visual effect of the CT image of the HTPB propellant, but also can perform statistics on each component in the image, which is convenient for propellant research and porosity calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A flowchart of a grayscale image processing method for HTPB propellant according to the present invention;

[0029] Figure 2 The comparison diagram of the original image of HTPB propellant before and after preprocessing, a is the original image, b is the AP layer obtained by threshold segmentation;

[0030] Figure 3 The image morphological processing diagram of the AP layer of HTPB propellant, a is the AP layer after corrosion operation, b is the AP layer after filling, c is the AP layer after expansion; d is the AP layer seed;

[0031] Figure 4 The morphological processing diagram of the AP layer of HTPB propellant, a is the AP particle diagram obtained by regional growth, b is the AP particle after morphological processing; c is the AP layer after morphological processing;

[0032] Figure 5 The comparison diagram of the AP layer of HTPB propellant before and after calibration, a is the AP layer before calibration, b is the AP layer after calibration;

[0033] Figure 6 The AP layer segmentation diagram of HTPB propellant, a is the A1 layer diagram, b is the matrix layer and pore layer diagram;

[0034] Figure 7 This is a structural diagram of the components of HTPB propellant; DETAILED DESCRIPTION

[0035] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments:

[0036] A grayscale image processing method for HTPB propellant of the present invention comprises the following steps:

[0037] S1. Preprocessing images based on information entropy

[0038] First, the original image of HTPB propellant obtained by CT scanning (such as Figure 2The image enhancement process is performed as shown in a. The image needs to be manually calibrated before processing, and the corresponding gray value ranges are given to the four components. A, B, C, and D are used to represent the upper gray value thresholds of pores, matrix, AP, and Al, respectively.

[0039] ;

[0040] Where a, b, c, and d represent the gray values ​​assigned to pores, matrix, AP, and Al during the enhancement process, respectively. During the processing, 0, 10, 20, and 30 are selected as the gray values ​​of each component of HTPB propellant, respectively, in order to achieve the purpose of enhancing the distinction between each component.

[0041] The image is denoised based on the neighborhood idea of ​​the mean filter, that is, points are taken one by one from the first pixel to the last pixel of the enhanced image to obtain a certain point Centered on 8 neighboring pixels , if there are at least 5 equal pixel values ​​in the 8 neighboring points of the point, the pixel value of the center point is equal to this pixel value. This has a good effect on the processing of some isolated noise and linear noise. And this method also effectively solves the problem that the gray value of the Al grain boundary is equal to the AP gray value due to the influence of X-rays in the CT image of HTPB propellant.

[0042] The image of the component structure of HTPB propellant has many components, and the grayscale value ranges representing the density of different components overlap. The grayscale value of the same component is also not uniform. By first enhancing the image and then using the 8-neighborhood pixel information method to denoise the image, the boundary will not be blurred and the impact of denoising on the details of the image will be reduced.

[0043] S2. Finding regional growth seed points

[0044] First, directly use the threshold segmentation to obtain Figure 2 The AP layer shown in b is eroded, and then the holes of the AP layer are filled. After all the holes are filled, the smaller blocks in the image caused by the erosion of linear and ring noise are deleted, and the remaining pixel blocks are expanded to make multiple blocks belonging to the same AP particle become a connected domain. The built-in function bwmorph (BW, 'shrink', Inf) of Matlab is used to shrink each block to a pixel point as the seed point for the growth of the AP layer area.

[0045] Some blocks in the AP layer obtained by direct segmentation belong to the matrix layer and the Al layer. The parts belonging to the Al layer and the matrix layer are regarded as noise in the AP layer. Such noise is generally linear and circular, and the width of these noises is small. Through the corrosion operation, these noises can be eliminated or divided into small pixel sets, such as Figure 3 At the same time, the AP particles in the segmented AP layer contain holes. After the corrosion operation, the internal holes will be enlarged. By filling the holes in the AP layer, all the holes can be filled to reduce the boundary distance scanned in subsequent operations, as shown in Figure 3 As shown in b. The smaller blocks in the image due to the erosion of linear and ring noise are deleted, and the remaining pixel blocks are mostly AP particle blocks. In these processing processes, AP particles will be processed into concentrated small blocks due to uneven grayscale values. When generating seeds, each block will be processed into a seed. In order to reduce the number of seeds, the expansion operation is used to make multiple blocks belonging to the same AP particle become a connected domain, as shown in Figure 3 As shown in c. At this time, most of the blocks in the image belong to the AP layer. Using the built-in function bwmorph (BW, 'shrink', Inf) of Matlab, each block is reduced to a pixel point as the seed for the growth of the AP layer area, as shown in Figure 3 As shown in middle d.

[0046] S3. Get the seed layer

[0047] The region growing algorithm is designed based on the similarity criterion of grayscale gradient, and the grayscale image is a two-dimensional function. , then its differential is a partial differential, so:

[0048] ;

[0049] The grayscale image of HTPB propellant is a discrete two-dimensional function. The value of cannot be infinitely small, because the image is discretized according to pixels, and only adjacent pixels are considered during growth. can only be 1, so the above formula can be transformed into the following form:

[0050] ;

[0051] This is the gradient of the image in the x and y directions. From this, we can conclude that the gradient of the image is the difference between adjacent pixels.

[0052] We first choose to process the AP layer, and the grayscale value range of the AP layer is in the middle range, so we use the function

[0053] ;

[0054] Adjust the brightness of the image. Remove some overly bright and dark areas to reduce the time of region growing retrieval; The pixels in the remaining area are grayscale stretched to increase the gradient between pixels, so that the region growth is more accurate to obtain AP particles, such as Figure 4 As shown in a.

[0055] The grayscale values ​​of different components of hydroxy propellant overlap, but the grayscale values ​​between the boundaries of different components vary greatly, and the grayscale value fluctuations of the same components are small. The regional growing algorithm is designed using the similarity criterion of grayscale gradients, which can make the AP particles obtained by regional growth more accurate.

[0056] The first point in the AP layer region growth seed group image obtained based on image processing traverses the entire image. If a certain point (x, y) is a seed point, this point is used as the starting point of region growth. The corresponding AP particles are obtained at the position of this point in the original image using the region growing algorithm. The generated particles are morphologically processed and the obtained AP particles are superimposed in the same image until all seed points are grown.

[0057] Usually, the seed points of the region growing algorithm need to be obtained by clicking on the image. However, a single CT image of HTPB propellant contains hundreds of AP particles, and each scan will produce hundreds of images. The human-computer interaction method is slow and labor-intensive. By improving the method of obtaining the starting point of the region growing algorithm, the workload of obtaining the seed point can be reduced and the acquisition efficiency can be improved.

[0058] A block (AP particle, such as Figure 4 As shown in a) in the figure, corrosion reduction is performed to restore it to a complete particle, such as Figure 4 As shown in b, the area of ​​the particle is then used to determine whether it is an AP particle. If it meets the scale condition, it is stored in the same position of the new matrix using matrix operations until all seed points are grown and screened, and the AP layer is output, as shown in Figure 4 As shown in c.

[0059] S4, AP layer correction

[0060] The threshold segmentation is performed by reducing the gray value range of the matrix (reducing the upper limit and setting the lower limit to 0), and the blocks smaller than the hole area are deleted from the obtained matrix layer image. Figure 5 The AP layer image is registered as shown in a), and the overlapping area is deleted to obtain the corrected AP layer, as shown in Figure 5 As described in b.

[0061] The grayscale value ranges of the matrix and AP overlap. Directly segmenting the matrix layer will cause too many parts belonging to the AP layer to be segmented together, which is not conducive to subsequent processing. The intersection of the grayscale values ​​of the matrix and AP particles is manifested in small blocks in the AP particles with the same grayscale value as the matrix. Threshold segmentation is performed by reducing the grayscale value range of the matrix (reducing the upper limit and the lower limit to 0), and deleting blocks smaller than the hole area in the obtained matrix layer image, which can reduce the area belonging to the AP layer in the matrix layer.

[0062] S5. Hierarchical segmentation

[0063] The AP layer is matched and compared with the original image, and the remaining part of the original image is segmented into other components Al, matrix layer and pores. The blocks in the Al layer with an area larger than the maximum cross-sectional area of ​​the Al particles are removed, and the removed part is attributed to the AP layer, and the remaining part is the final Al layer, such as Figure 6 As shown in a, the threshold segmentation is directly used in the remaining part of the image to obtain the matrix layer and the pore layer, as shown in Figure 6 As shown in b.

[0064] The grayscale value range of AP is between Al and the matrix. When the area of ​​the AP layer is determined, the grayscale histogram of the remaining area of ​​the original image becomes a bimodal function. The peak on the left represents the darker matrix and pores, and the peak on the right represents the brighter Al. Set the average value of the upper and lower thresholds of the AP grayscale value as the lower grayscale value for segmenting the Al layer, perform threshold segmentation, and obtain the Al layer. The Al layer contains some highlighted parts belonging to the AP layer, so the blocks in the Al layer with an area larger than the maximum cross-sectional area of ​​the Al particles are removed, and the removed parts are attributed to the AP layer, and the remaining parts are the final Al layer.

[0065] The matrix appears as a network of connected domains in the image. The pores are generally considered to be vacuum, and there may also be a very small amount of propellant gas decomposition products, so the gray value representing the pore density in the image is relatively small. At present, the calibration of pores still relies on manual calibration. It is generally believed that the area within the gray value range of the calibrated pores belongs to the pore layer, so the threshold segmentation is directly used in the remaining part of the image to obtain the matrix layer and the pore layer.

[0066] S6, image synthesis: The segmented AP layer, Al layer, matrix layer and pore layer images are reassigned to different colors and then matrix added to obtain a complete image for the structural identification of the HTPB propellant components, such as Figure 7 shown.

[0067] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the technical solution of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A grayscale image processing method for HTPB propellant, characterized in that: The steps include: S1. Preprocessing the image based on information entropy: denoising the original image obtained after scanning by the CT device; S2. Finding regional growth seed points: First, according to the components of HTPB propellant, the image is directly segmented using the gray value threshold to obtain the ammonium perchlorate part, and then the AP part is processed using the image morphological processing algorithm to obtain the seed points for the growth of the AP layer region; The image morphological processing operation in step S2 specifically includes firstly performing an erosion operation on the AP part obtained by directly using the threshold segmentation, and then performing hole filling on the AP part to fill all the holes, deleting the small blocks in the image caused by the erosion of linear and ring noise, and performing an expansion operation on the remaining pixel blocks to make multiple blocks belonging to the same AP particle become a connected domain, and using the bwmorph function provided by Matlab to reduce each block to a pixel point as a seed point for the growth of the AP layer area; S3, obtaining a seed layer: using a region growing algorithm to restore the original AP particles in the region where the seeds are located, and using morphological processing to restore the AP particles to obtain an AP layer; The region growing algorithm is specifically designed using the similarity criterion of grayscale gradient. The grayscale image of HTPB propellant is a discrete two-dimensional function, namely ； in Grayscale image representing HTPB propellant; S4, AP layer correction: register the images of the threshold segmented basal layer and the AP layer, delete the overlapping area in the AP layer image, and obtain the corrected AP layer; S5, layer segmentation: matching and comparing the obtained AP layer with the original image, and performing segmentation processing on other components Al, matrix layer and pores in the remaining part of the original image; S6. Image synthesis: The segmented AP layer, Al layer, matrix layer and pore layer images are reassigned to different colors and then matrix-added to obtain a complete image for component structure identification of HTPB propellant.

2. A grayscale image processing method for HTPB propellant according to claim 1, characterized in that: Before denoising the original image in step S1, the original image needs to be enhanced. The denoising method is based on the neighborhood idea of ​​the mean filter, that is, points are taken one by one from the first pixel to the last pixel of the enhanced image to obtain a certain point. Centered on 8 neighboring pixels , if there are no less than 5 pixel values ​​in the 8 points in the neighborhood of the point that are equal, then the pixel value of the center point is equal to this pixel value.

3. A grayscale image processing method for HTPB propellant according to claim 2, characterized in that: The gradient of the grayscale image of the HTPB propellant in step S3 is the difference between adjacent pixels, and the function ； Adjust the brightness of the image. Remove some overly bright and dark areas to reduce the time of region growing retrieval; The pixels in the remaining area are grayscale stretched to increase the gradient between pixels, so that the region growth can be more accurate to obtain AP particles.

4. A grayscale image processing method for HTPB propellant according to claim 3, characterized in that: The starting point of the seed point of the region growing algorithm is obtained by traversing the entire image at the first point in the AP layer region growth seed group image. If a point (x, y) is a seed point, this point is used as the starting point of the region growing, and the corresponding AP particles are obtained at the position of this point in the original image using the region growing algorithm. The generated particles are morphologically processed and the obtained AP particles are superimposed in the same image until all seed points are grown.

5. A grayscale image processing method for HTPB propellant according to claim 1, characterized in that: The morphological processing in step S3 specifically refers to corroding and reducing a block grown from a seed point to restore it to a complete particle, and then judging whether it belongs to an AP particle based on the area size of the particle. If it meets the scale condition, it is stored in the same position of the new matrix using matrix operation until all seed points are grown and screened, and the AP layer is output.

6. A grayscale image processing method for HTPB propellant according to claim 1, characterized in that: The threshold segmentation in step S4 specifically refers to performing threshold segmentation by reducing the gray value range of the matrix, and deleting blocks smaller than the hole area in the obtained image of the matrix layer.

7. A grayscale image processing method for HTPB propellant according to claim 1, characterized in that: In step S5, the Al layer is segmented to remove the blocks in the Al layer whose area is larger than the maximum cross-sectional area of ​​the Al particles, and the removed parts are classified as the AP layer, and the remaining parts are the final Al layer.

8. A grayscale image processing method for HTPB propellant according to claim 1, characterized in that: The segmentation of the matrix layer and the pore layer in step S6 is to directly use threshold segmentation in the remaining part of the image to obtain the matrix layer and the pore layer.

Images