CN113077463B - Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method - Google Patents

Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method Download PDF

Info

Publication number
CN113077463B
CN113077463B CN202110477969.8A CN202110477969A CN113077463B CN 113077463 B CN113077463 B CN 113077463B CN 202110477969 A CN202110477969 A CN 202110477969A CN 113077463 B CN113077463 B CN 113077463B
Authority
CN
China
Prior art keywords
energy
ray
mirror
chirplet
lion
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110477969.8A
Other languages
Chinese (zh)
Other versions
CN113077463A (en
Inventor
吴萌
王姣
任义
贾旻
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian University of Architecture and Technology
Original Assignee
Xian University of Architecture and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian University of Architecture and Technology filed Critical Xian University of Architecture and Technology
Priority to CN202110477969.8A priority Critical patent/CN113077463B/en
Publication of CN113077463A publication Critical patent/CN113077463A/en
Application granted granted Critical
Publication of CN113077463B publication Critical patent/CN113077463B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G06T5/30—Erosion or dilatation, e.g. thinning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10116—X-ray image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20036—Morphological image processing
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E30/00—Energy generation of nuclear origin
    • Y02E30/10—Nuclear fusion reactors

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Health & Medical Sciences (AREA)
  • Analysing Materials By The Use Of Radiation (AREA)

Abstract

The invention discloses a Chirplet energy-guided lion group disturbance optimization ancient copper mirror X-ray fusion flaw detection method, which comprises the following steps: preprocessing X-ray diffraction images generated by the bronze mirror aiming at surface light sources with different energies to obtain X-ray images with different surface light sources; for X-ray images with different surface light source energies, constructing a sparse decomposition atom library based on Chirplet energy-guided lion disturbance optimization Matching Pursuit, and selecting a atom library with stronger sparsity
Figure DDA0003047759900000011
Constructing a required atom library; carrying out X-ray energy statistics on X-ray images of different surface light sources by using Chirplet coefficients to obtain energy parameters of the decomposition positions of the bronze mirror; and (3) carrying out improved guidance on lion disturbance factors by using Chirplet energy, then carrying out lion optimized Matching Pursuis sparse decomposition on X-ray images with different energies, decomposing the X-ray images of the copper mirror into upper-layer texture detail information and bottom-layer copper mirror structure information, and obtaining diseases and calibration results of the copper mirror. The method can observe disease conditions of different areas in the X-ray copper mirror image.

Description

Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method
Technical Field
The invention belongs to the field of digital image processing, and relates to a method for detecting ancient copper mirror X-ray fusion flaw detection by using Chirplet energy to guide lion group disturbance optimization.
Background
Four thousands of years ago, along with the appearance of alloy technology, our country began to use copper and tin or silver lead, etc. to make copper mirrors. The ancient copper mirror has beautiful shape, gorgeous patterns and rich inscription, gradually becomes an artwork of ancient life, is a treasure in ancient cultural heritage of China, and has high artistic and research values. Because of the long ages, most of the copper mirrors which come out of the earth have very serious disease erosion and dirt coverage. Before proper cleaning and repair are carried out, the state of the copper mirror is required to be known, and the X-ray imaging technology is a good nondestructive testing technology means. The internal structural characteristics, the texture patterns, the disease areas and the like of the copper mirror can be reflected by shooting the internal form of the copper mirror with rust through non-contact penetrability. However, in the process of photographing by the X-ray camera, the area light source generates penumbra blurring due to the X-ray signal, and the influence of compton scattering is added, so that the transmissivity of the X-ray signal in each area (the textured area and the mirror edge area) of the copper mirror is uneven, and the energy is unevenly distributed. And the thickness of the mirror edge area of the copper mirror is different from that of the textured area, and the optimal X-ray diffraction energy is also different. In order to observe all information of the ancient copper mirror, different optimal diffraction energies are required to image the mirror edge area and the texture area respectively, so that a plurality of surface light source X-ray images with different energies are generated. This makes it very difficult for the history specialist to observe the disease condition of different areas simultaneously in one X-ray copper image.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides a method for detecting the X-ray fusion flaw detection of an ancient bronze mirror by using Chirplet energy to guide lion disturbance optimization.
In order to achieve the purpose, the method for detecting the X-ray fusion flaw detection of the ancient copper mirror for optimizing the disturbance of the lion group by using the Chirplet energy comprises the following steps:
1) Preprocessing X-ray diffraction images generated by the bronze mirror aiming at surface light sources with different energies to obtain X-ray images with different surface light sources;
2) For X-ray images with different surface light source energies, constructing a sparse decomposition atom library based on Chirplet energy-guided lion disturbance optimization Matching Pursuit, and selecting L with stronger sparsity 2,12 Constructing a required atom library;
3) Carrying out X-ray energy statistics on X-ray images of different surface light sources by using Chirplet coefficients to obtain energy parameters of the decomposition positions of the bronze mirror;
4) Performing improved guidance on lion disturbance factors by using Chirplet energy, then performing lion optimized Matching Pursuis sparse decomposition on different energy X-ray images based on the constructed energy parameters of the needed atomic library and the decomposition position of the bronze mirror, and decomposing the bronze mirror X-ray images into upper-layer texture detail information and bottom-layer bronze mirror structure information;
5) And (3) obtaining diseases and calibration results of the copper mirror based on the upper layer texture detail information and the bottom layer copper mirror structure information obtained by the decomposition in the step (4), and completing the ancient copper mirror X-ray fusion flaw detection of the Chirplet energy-guided lion group disturbance optimization.
The X-ray images with different surface light source energies comprise X-ray images with high brightness and clear mirror edge textures generated by a high-energy surface light source and X-ray images with clear mirror edge textures and low brightness generated by a low-energy surface light source.
In the step 1), the preprocessing process of the X-ray diffraction image generated by the bronze mirror aiming at the surface light sources with different energies comprises the following steps: the X-ray diffraction image is standardized through geometric transformation of translation, transposition, mirroring, rotation and scaling, the center position and the rotation angle of the X-ray diffraction image are guaranteed to be the same, and meanwhile, the system error of an image acquisition system and the random error of the instrument position are corrected.
In the step 3), energy statistics is carried out on X-ray images of different surface light source energies by using Chirplet coefficients, the linear characteristic degree of the X-ray signals of the surface light source is described by adopting the Chirplet coefficients as the energy coefficients aiming at the enhancement expression and fusion of the texture characteristics of the paleo mirror, and the two-dimensional M is input 1 ×M 2 The bronze mirror image is respectively subjected to row and column decomposition on the basis of one-dimensional signal decomposition, and the specific detailed energy function is as follows:
Figure BDA0003047759880000031
wherein ,
Figure BDA0003047759880000032
is->
Figure BDA0003047759880000033
Coefficient energy of horizontal, vertical and diagonal decomposition is transformed by Chirplet respectively to obtain points with strong linear energy, and the decomposition position is determined during decomposition:
Figure BDA0003047759880000034
the perturbation factors of the female lion are:
Figure BDA0003047759880000035
Figure BDA0003047759880000041
wherein ,
Figure BDA0003047759880000042
represents the maximum step size during movement, +.>
Figure BDA0003047759880000043
and />
Figure BDA0003047759880000044
The average value of the maximum value and the minimum value of each dimension in the movable range is respectively, v is an energy weighting coefficient, T is the current iteration number, and T is the maximum iteration number.
The perturbation factors of young lion are:
Figure BDA0003047759880000045
the specific operation of the step 5) is as follows:
and carrying out multi-scale fusion and morphological enhancement on the upper layer texture detail information and the bottom layer copper mirror structure information by using a multi-size multi-structure enhanced image algorithm based on mathematical morphology, carrying out maximum inter-class segmentation by using a maximum inter-class method to obtain a disease image, registering and calibrating the disease image to obtain a final disease and a calibration result thereof, and completing the Chirplet energy-guided lion disturbance optimized ancient copper mirror X-ray fusion flaw detection.
The invention has the following beneficial effects:
when the Chirplet energy is used for guiding lion group disturbance factors in specific operation, the Chirplet energy is used for improving and guiding the lion group disturbance factors, then the lion group optimized Matching sparse decomposition is carried out on different energy X-ray images, the copper mirror X-ray images are decomposed into upper layer texture detail information and bottom layer copper mirror structure information, and then diseases and calibration results of the copper mirror are used for observing disease conditions of different areas in the X-ray copper mirror images.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a high energy X-ray image area light source image of an ancient copper mirror according to the invention;
FIG. 3 is a representation of a low energy X-ray image area source image for an ancient copper mirror in accordance with the present invention;
FIG. 4 is a graph showing the effect of the registration of the X-ray images of the bronze mirror in the present invention;
FIG. 5 is a diagram showing MP decomposition effects obtained by selecting decomposition positions of Chirplet energy coefficients according to the present invention;
FIG. 6 is a graph showing the effect of fusion enhancement of an ancient copper mirror according to the present invention after mathematical morphology treatment;
FIG. 7 is a graph showing the effect of the maximum inter-class segmentation according to the present invention;
FIG. 8 is a graph showing the effect of the split crack after the split treatment according to the present invention;
FIG. 9 is a graph showing the marking of cracking disease in an X-ray image of an ancient copper mirror according to the present invention.
Detailed Description
The invention is described in further detail below with reference to the attached drawing figures:
the invention relates to a Chirplet energy-guided lion group disturbance optimized ancient copper mirror X-ray fusion flaw detection method, which comprises the following steps:
1) The method comprises the steps of preprocessing X-ray diffraction images generated by a bronze mirror aiming at different energy area light sources to obtain X-ray images with different area light source energies, wherein the X-ray images with different area light source energies comprise X-ray images with high brightness of mirror center areas and clear mirror edge textures generated by high energy area light sources and X-ray images with clear mirror center area textures and low brightness of mirror edges generated by low energy area light sources.
2) For X-ray images with different surface light source energies, a sparse decomposition atom library is built based on Chirplet energy guiding lion disturbance optimization Matching Pulse (MP), and then L with stronger sparsity is adopted 2,12 Constructing a required atom library.
3) And carrying out X-ray energy statistics on X-ray images of different surface light sources by using Chirplet coefficients to obtain energy parameters of the decomposition position of the bronze mirror.
4) Using Chirplet energy to guide lion disturbance factors, performing lion optimization Matching (MP) sparse decomposition on different energy X-ray images, and decomposing the copper mirror X-ray images into upper layer texture detail information and bottom layer copper mirror structure information;
5) And carrying out multi-scale fusion and image enhancement on the upper layer texture detail information and the bottom layer copper mirror structure information by using a multi-scale and multi-structure enhanced image algorithm based on mathematical morphology, so that the copper mirror image highlights crack characteristics, which is convenient for crack extraction, and carrying out maximum inter-class separation by using a maximum inter-class method to obtain a disease image, and then registering and calibrating the disease image to obtain final disease flaw detection and calibration results.
The specific operation of the step 1) is as follows: the X-ray diffraction images generated by the bronze mirror aiming at the surface light sources with different energies are eliminated, penumbra blurring generated by the surface light sources is acquired by the X-ray equipment, and the X-ray images to be fused are standardized by geometric transformation of translation, transposition, mirror image, rotation and scaling aiming at acquisition distance, angle and inclination angle errors, so that the standard sizes are M 1 ×M 2 And ensures that the center position is consistent with the rotation angle, and corrects the system error of the image acquisition system and the instrument positionRandom errors, and standardizing materials to be fused to prepare for subsequent operations such as decomposition, fusion, marking and the like.
In step 2), a general atom library is composed of a group of unitary operators g γ Acting on unit L 2 Established by norms, the invention selects L 2,1/2 To create an atom library d= { g γ ,γ∈Γ}，L 2 The norm refers to the square root of the sum of squares of the elements of the vectors, compared to L 1 The norm will select more features, and the features are all close to 0, the invention is keeping L 2 On the premise of preventing overfitting and solving stability and rapidness, the norm is compared with L 2 ,L 2，1 And L is equal to 2，1/2 Is used for selecting L in the construction of an overcomplete atomic library of an X-ray copper mirror 2,1/2 The norm ensures a more stable decomposition process.
Figure BDA0003047759880000071
Figure BDA0003047759880000072
Figure BDA0003047759880000073
Let the input vector x= (X) 1 ,x 2 ,...,x n ) T ，L 2 ,L 2，1 L and L 2，1/2 The method comprises the following steps of:
Figure BDA0003047759880000074
Figure BDA0003047759880000075
Figure BDA0003047759880000076
input as a matrix
Figure BDA0003047759880000077
Will L 2 ,L 2，1 and L2，1/2 Respectively comparing;
due to
Figure BDA0003047759880000078
And->
Figure BDA0003047759880000079
Let inequality->
Figure BDA00030477598800000710
If true, only prove
Figure BDA00030477598800000711
And (3) difference between the two:
Figure BDA00030477598800000712
/>
then
Figure BDA00030477598800000713
Similarly->
Figure BDA00030477598800000714
Namely, the above-mentioned symptoms are marked by->
Figure BDA00030477598800000715
It is understood that although the invention selects L 2,1/2 Norms relative to L 2 The method is complex in calculation, has optimal sparsity, and simultaneously maintains the basic characteristics of overfitting prevention and stable and rapid calculation. For the X-ray signal fusion requirement of the ancient bronze mirror, the sparsity degree of decomposition and the final fusion effect are more focused, so that the L with more prominent sparsity is preferentially selected in the acceptable calculation bearing range 2,1/2 To create an atom library.
In the step 3), the result of energy statistics by using the Chirplet coefficient is used for designing the X-ray energy parameter of the archaizing mirror.
The main problem faced by the X-ray light source bronze mirror images generated by different energies in the decomposition process is that a surface light source has no focus and is in a scattering state under the influence of Compton scattering. To constrain the state of this scattering, an energy-stable imaging effect is chosen. The invention adopts Chirplet coefficient to count the X-ray energy and set the distribution parameter of the X-ray energy. In signal decomposition, the Chirplet coefficients represent signals with more pronounced linear characteristics. The Chirplet coefficient is used for describing the degree of the enhancement expression of the texture characteristics of the bronze mirror in the X-ray signals, and the parameter can be used for extracting and reconstructing information with larger line energy represented in X-ray images of the bronze mirrors with different energies.
In contrast to short-time fourier transforms and wavelet transforms, the kernel functions of the Chirplet transform include scaling, modulation, time shifting, and time chirping. The five kinds of linear frequency modulation transformation can obtain a kernel function with time-frequency distribution of approximately diamond, and the image information of the paleo-copper mirror can have better resolution in a time-frequency domain by adjusting parameters.
And performing Chirplet calculation on the X-ray image of the bronze mirror before decomposition, and describing the linear characteristic degree of the surface light source X-ray signal by adopting the Chirplet coefficient as an energy coefficient. The larger the Chirplet energy coefficient of the region with obvious linear characteristics such as texture, crack and the like is, the smaller the Chirplet energy coefficient is. For input two-dimensional M 1 ×M 2 And (3) performing row-column decomposition on the X-ray image of the bronze mirror again on the basis of one-dimensional signal decomposition, and defining a specific detailed energy function as follows:
Figure BDA0003047759880000091
wherein ,
Figure BDA0003047759880000092
is->
Figure BDA0003047759880000093
The coefficient energies of the horizontal, vertical and diagonal decomposition of the Chirplet transform are respectively. The total energy function of the texture details of the bronze mirror image is as follows:
Figure BDA0003047759880000094
generating atom g γ In order to effectively approximate the contour line of two-dimensional singular points, atoms need a stable low-resolution function in the aspect of outline and represent a ground image wavelet in the singular direction, the position meeting the requirement of large energy coefficient of the comprehensive Chirplet is selected for decomposition by counting the size of the energy function of all X-ray image points, the subsequent selection of generated atoms is facilitated, and after the optimal joint space and frequency localization of an energy distribution function and a Gaussian kernel are taken into consideration, the invention designs a combination of a Gaussian function and a second derivative thereof, and the combination is compared with the wavelet through the PSNR value of an image, wherein the comparison is based on the quality obtained by using the maximum N item when the wavelet or MP decomposition is carried out on the image. It can be seen that the matching pursuit represents an advantage over all wavelet expansions. Even with more than three times the term, the wavelet does not provide quality equivalent to matching pursuit decomposition.
g γ (x,y)＝(4x 2 -2)exp(-(x 2 +y 2 )) (9)
In the step 4), the process of optimizing MP sparse decomposition by using the lion group algorithm of Chirplet energy improvement disturbance factor to the bronze mirror X-ray image comprises the following steps:
setting a decomposition position selected by a Chirplet energy coefficient, decomposing according to MP to form a series of atoms, forming a whole atom library, and searching the optimal atoms in the overcomplete atom library according to a lion group algorithm of an energy improvement disturbance factor; and removing components of the image residual on the atoms from the bronze mirror X-ray image or the residual of the image to finish one-step decomposition, and finally judging the decomposition according to different standards.
In each iteration process, the Chirplet energy guiding lion disturbance optimization (MP) algorithm selects atoms which are most matched with X-ray images of different energy from the over-complete dictionary D to construct sparse approximation, and meanwhile obtains the X-ray image signals of the bronze mirror to represent residual errors. Then, the atoms which are most matched with the residual errors of the X-ray image signals of the bronze mirror are selected continuously, and after a certain number of iterations, the X-ray image signals of the bronze mirror with different energies can be represented by a plurality of atoms in a linear mode.
Suppose that the X-ray images of different energy bronze mirrors of the study are e, and the image size is M 1 ×M 2 ,M 1 and M2 The length and width of the image, respectively. If the different energy bronze mirror X-ray images are decomposed on a set of perfectly orthogonal bases, the set of satisfactory bases should be M 1 ×M 2 And each. The sparsity of the distribution of the set of satisfactory bases in the space formed by the copper images is determined by the orthogonality of the bases, so that the energy of the different energy copper-X-ray images will be distributed on the different bases after decomposition.
Let D= { g γ Gamma e Γ is an overcomplete library for sparse decomposition of paleo-copper X-ray images, g γ For the atoms defined by the parameter group γ, atoms are constructed using different methods, and the parameters and the number of parameters contained in the parameter group γ are different. Atom g γ The size of the image is the same as the size of the X-ray image of the bronze mirror, and atoms are normalized, namely the atomic number g γ I=1, Γ is the set of parameter sets γ. If the overcomplete library d= { g is denoted by a γ If the number of atoms in gamma E gamma is greater than the size M of the X-ray image of the bronze mirror 1 ×M 2 。
In the step 4), the process of searching the best atoms by using the lion group algorithm based on the Chirplet energy improved disturbance factor is as follows:
in the process of performing sparse decomposition of different-energy archaeological copper image X-ray images by using a Chirplet energy-guided lion disturbance optimization (MP) algorithm, each step of decomposition is to calculate the projection of copper images or the residues of copper image decomposition on each atom in the constructed overcomplete library. According to g γk Meeting the conditions ofIt can be seen that each decomposition step in the decomposition process is required to be performed in a very high dimension (M 1 ×M 2 ) Is spatially processed multiple times by inner product calculation<R k e,g γ >. The optimal solution of the objective function can be obtained by the algorithms such as particle swarm, artificial bee colony, universal gravitation search, lion swarm and the like under the low-dimensional space, and the lion swarm algorithm has the highest convergence speed in terms of convergence speed. Other algorithms are premature and easily trapped in local optimum under the high-dimensional space, and the lion group algorithm can be quickly close to the optimum in the high-dimensional space. Compared with other algorithms, the lion group algorithm has higher convergence speed, and is easy to jump out of local optimum to find global optimum. Based on excellent performance of the lion group algorithm on a standard function library, the lion group algorithm is selected for carrying out optimal atom optimization in a decomposed atom library of the X-ray copper mirror image for the high-dimensional optimization problem generated by the decomposition of the ancient copper mirror image.
The optimization of the high-dimensional optimization problem generated by the image decomposition of the ancient copper mirror requires the combination of the Chirplet energy parameters of the image decomposition of the copper mirror to improve the global reconnaissance capability of the mother lion and the young lion. And weighting the energy coefficient of the disturbance factor to find the approximate position of an optimal solution, and then enhancing the local refinement capability of the disturbance factor. The new disturbance factor can change the foraging range of the female lion and the young lion, the transition is from big to small, when a rough range is locked, the movable range is gradually narrowed, and finally, the value of the movable range which is infinitely close to zero is kept. When the Chirplet energy coefficient is in the corresponding range, the energy weighting coefficient of the disturbance factor adopts different values, the linear energy contained in the range is judged according to the energy coefficient, and upsilon is judged according to the energy. When the energy is less, the energy weighting is adopted to make the foraging range of the female lion shrink fast, so as to achieve the purpose of rapid convergence; when more energy is contained, the energy weighting is adopted to slow down the contraction speed of the foraging range of the female lion, so that more detailed information can be obtained. The disturbance factor of the young lion can play a role in stretching or compressing the movable range. The disturbance factor can well balance the global investigation capability and the local refinement capability of the mother lion and the young lion. The convergence is rapid, the premature can be effectively avoided, and the optimal solution is easy to obtain.
The perturbation factors of the female lion are:
Figure BDA0003047759880000121
Figure BDA0003047759880000122
wherein ,
Figure BDA0003047759880000123
represents the maximum step size during movement, +.>
Figure BDA0003047759880000124
and />
Figure BDA0003047759880000125
And the average value of the maximum value and the minimum value of each dimension in the movable range is respectively represented, v is an energy weighting coefficient, T is the current iteration number, and T is the maximum iteration number.
The perturbation factors of young lion are:
Figure BDA0003047759880000126
taking a parameter group gamma of an atom as a parameter to be optimized, and adopting a bronze mirror image or an absolute value of the inner product of the residual of the bronze mirror image signal and the atom as an objective function<R k e,g γ >|。
In the process of decomposing and optimizing the bronze mirror image by adopting an energy-improvement-based disturbance factor lion group algorithm, the occupied scale factor of the adult lion is lambda=0.2, and T=200.
In the step 5), the process of carrying out multiscale morphological enhancement on the decomposed bronze mirror images with different energies comprises the following steps:
gray level equalization processing is carried out on the fused bronze mirror image, morphological structural elements are set, opening operation and closing operation are carried out on the set structural elements, then combined algorithms such as opening operation, closing operation summation and the like are carried out, and finally reconstruction is carried out to obtain the bronze mirror image with multi-scale morphological enhancement.
Gray scale morphology is the basis of multi-scale mathematical morphology research, the basic operators of which are erosion, open operation, expansion and closed operation,
let e be the X-ray image of the bronze mirror with different energy, D be the structural element, D x and Dd The definition fields of x and d respectively, the open operation and the closed operation involved in the definition fields are as follows:
gray scale on operation
Figure BDA0003047759880000131
Gray scale closure operation
Figure BDA0003047759880000132
The final reinforced copper mirror pattern is obtained by the following steps:
511 All the entries of the difference model corresponding to the gray level on operation are added to obtain the bronze mirror image composed of the bright features under the multi-scale condition.
Figure BDA0003047759880000133
512 And similarly, adding all entries of the difference model corresponding to the gray level closing operation, and forming the bronze mirror image by the dim characteristics under the multi-scale condition.
Figure BDA0003047759880000134
513 The original gray level image and the image formed by the bright and dim characteristics under the multi-scale condition are processed in series to obtain the final multi-scale morphological enhanced bronze mirror image.
e En (x,y)＝e(x,y)+λS°(x,y)-(1-λ)S c (x,y) (17)
Wherein lambda is a copper mirror bright-dark characteristic relation parameter, lambda has strong correlation with the bright-dark characteristic of the copper mirror, and lambda epsilon (0, 1). And selecting the copper mirror brightness characteristic relation parameters according to the actual requirements of the images of the different-energy x-ray paleo-copper mirrors in the enhancement process.
In the step 5), the adaptive global threshold segmentation process based on the maximum inter-class method for the upper multi-scale morphological enhancement copper mirror image comprises the following steps:
521 Let the bronze mirror image e (x, y) have L gray levels, where the ith pixel is N i Number of pixels of the image
Figure BDA0003047759880000141
522 Calculating probability of occurrence of ith level pixel
Figure BDA0003047759880000142
523 Setting a threshold value, wherein the image is divided into two main categories of a target and a background by T, and the two categories respectively comprise the gray level of the image of 0-k and k+1-L-1;
524 Calculating the total average gray level of the image
Figure BDA0003047759880000143
525 Calculating the average gray level of the target and background areas as:
Figure BDA0003047759880000144
526 Instruction) command
Figure BDA0003047759880000145
And->
Figure BDA0003047759880000146
The resulting inter-class variance is:
Figure BDA0003047759880000147
527 Calculating k to be transformed between 0 and L-1 to obtain different inter-class variances, wherein k corresponding to the largest inter-class variance is the optimal threshold.
In step 5), the self-adaptive global threshold segmentation method using the maximum inter-class method has an over-segmentation phenomenon, and the edges of the image copper mirror are segmented together while the crack is segmented, and the over-segmentation phenomenon needs to be processed, and the specific process is as follows:
531 Reading in the segmented bronze mirror crack image.
532 Reading information of manually defined points in the image.
533 When the abscissa is calculated, the abscissa is col of the image of the bronze mirror crack, the column coordinate is row of the image of the bronze mirror crack, the abscissa is unchanged, and the ordinate row-y;
524 The information of the cutting point is obtained, the over-cutting phenomenon of the image is eliminated, and finally the crack of the copper mirror image is obtained.
In the step 5), the mask image of the crack disease mark is mapped back to the X-ray copper mirror image to carry out disease distribution expression, and the specific process is as follows:
541 Reading in an X-ray gray level image and a binary mask image obtained by disease segmentation;
542 Setting registration parameters and using mutual information as a measure for registration;
543 Performing superposition mapping on the registered images, and displaying a final X-ray flaw detection marking result;
mutual information is an important concept in information theory, and mainly describes the correlation between two systems or how much information is contained in each other. In image registration, the mutual information of two images reflects the mutual inclusion degree of the information between the two images through the entropy and the joint entropy of the two images, and the image e is an image of the bronze mirror X-ray with different energies 1 ,e 2 In terms of this mutual information is expressed as:
MI(e 1 e 2 )＝H(e 1 )+H(e 2 )-H(e 1 ,e 2 ) (20)
registration conditions are: the largest of the mutual information Mutual Information (MI) between the template and each sub-image is found to be the registered image.

Claims (6)

1. A Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method is characterized by comprising the following steps:
1) Preprocessing X-ray diffraction images generated by the bronze mirror aiming at surface light sources with different energies to obtain X-ray images with different surface light sources;
2) For X-ray images with different surface light source energies, constructing a sparse decomposition atom library based on Chirplet energy-guided lion disturbance optimization Matching Pursuit, and selecting L with stronger sparsity 2,1/2 Constructing a required atom library;
3) Carrying out X-ray energy statistics on X-ray images of different surface light sources by using Chirplet coefficients to obtain energy parameters of the decomposition positions of the bronze mirror;
4) Performing improved guidance on lion disturbance factors by using Chirplet energy, then performing lion optimized Matching Pursuis sparse decomposition on different energy X-ray images based on the constructed energy parameters of the needed atomic library and the decomposition position of the bronze mirror, and decomposing the bronze mirror X-ray images into upper-layer texture detail information and bottom-layer bronze mirror structure information;
5) Based on the upper layer texture detail information and the bottom layer copper mirror structure information obtained by the decomposition in the step 4), obtaining diseases and calibration results of the copper mirrors, and finishing the ancient copper mirror X-ray fusion flaw detection of the Chirplet energy guiding lion group disturbance optimization;
the perturbation factors of the female lion are:
Figure FDA0004180691300000011
Figure FDA0004180691300000012
wherein ,
Figure FDA0004180691300000013
represents the maximum step size during movement, +.>
Figure FDA0004180691300000014
and />
Figure FDA0004180691300000015
The average value of the maximum value and the minimum value of each dimension in the movable range is respectively, v is an energy weighting coefficient, T is the current iteration number, and T is the maximum iteration number;
the perturbation factors of young lion are:
Figure FDA0004180691300000021
2. the method for detecting the X-ray fusion flaw detection of the ancient copper mirror with optimized Chirplet energy-guided lion disturbance according to claim 1, wherein the X-ray images with different surface light source energies comprise an X-ray image with a high-energy surface light source generating a bright mirror center area and a clear mirror edge texture and an X-ray image with a low-energy surface light source generating a clear mirror center area and a low mirror edge texture.
3. The method for detecting the X-ray fusion flaw detection of the ancient copper mirror for the disturbance optimization of the Chirplet energy-guided lion group according to claim 1, wherein the process of preprocessing the X-ray diffraction images generated by the ancient copper mirror aiming at the surface light sources with different energies in the step 1) is as follows: the X-ray diffraction image is standardized through geometric transformation of translation, transposition, mirroring, rotation and scaling, the center position and the rotation angle of the X-ray diffraction image are the same, and meanwhile, the system error of the image acquisition system and the random error of the instrument position are corrected.
4. The Chi of claim 1An ancient bronze mirror X-ray fusion flaw detection method with rplet energy guiding lion disturbance optimization is characterized in that in step 3), energy statistics is carried out on X-ray images of different surface light source energy by using a Chirplet coefficient, the linear characteristic degree of a surface light source X-ray signal is described by adopting the Chirplet coefficient as an energy coefficient aiming at the enhancement expression and fusion of the texture characteristics of the ancient bronze mirror, and the linear characteristic degree of the surface light source X-ray signal is described for input two-dimensional M 1 ×M 2 And respectively carrying out row-column decomposition on the bronze mirror image on the basis of one-dimensional signal decomposition.
5. The method for detecting the X-ray fusion flaw detection of the ancient copper mirror by using the Chirplet energy to guide lion group disturbance optimization according to claim 1, wherein the energy function is as follows:
Figure FDA0004180691300000031
wherein ,
Figure FDA0004180691300000032
is->
Figure FDA0004180691300000033
Coefficient energy of the Chirplet transformation horizontal, vertical and diagonal decomposition is respectively obtained, so that points with strong linear energy are obtained, and the decomposition position is determined during decomposition:
Figure FDA0004180691300000034
6. the method for detecting the X-ray fusion flaw detection of the ancient copper mirror for optimizing the disturbance of the Chirplet energy-guided lion group according to claim 5, wherein the specific operation of the step 5) is as follows:
and carrying out multi-scale fusion and morphological enhancement on the upper layer texture detail information and the bottom layer copper mirror structure information by using a multi-size multi-structure enhanced image algorithm based on mathematical morphology, carrying out maximum inter-class segmentation by using a maximum inter-class method to obtain a disease image, and then carrying out calibration and calibration on the disease image to obtain a final disease and a calibration result thereof, thereby completing the Chirplet energy-guided lion disturbance optimized ancient copper mirror X-ray fusion flaw detection.
CN202110477969.8A 2021-04-29 2021-04-29 Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method Active CN113077463B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110477969.8A CN113077463B (en) 2021-04-29 2021-04-29 Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110477969.8A CN113077463B (en) 2021-04-29 2021-04-29 Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method

Publications (2)

Publication Number Publication Date
CN113077463A CN113077463A (en) 2021-07-06
CN113077463B true CN113077463B (en) 2023-06-06

Family

ID=76616090

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110477969.8A Active CN113077463B (en) 2021-04-29 2021-04-29 Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method

Country Status (1)

Country Link
CN (1) CN113077463B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109872330A (en) * 2019-01-25 2019-06-11 安徽理工大学 A kind of two-dimentional Otsu Fast image segmentation method for improving lion group's optimization
CN110147890A (en) * 2019-05-13 2019-08-20 湖北工业大学 A kind of method and system based on lion group's algorithm optimization extreme learning machine integrated study
CN111582128A (en) * 2020-04-30 2020-08-25 电子科技大学 Mechanical fault sparse representation method based on wolf pack parameterized joint dictionary

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2976746B1 (en) * 2013-03-21 2019-01-09 Koninklijke Philips N.V. A method and x-ray system for computer aided detection of structures in x-ray images

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109872330A (en) * 2019-01-25 2019-06-11 安徽理工大学 A kind of two-dimentional Otsu Fast image segmentation method for improving lion group's optimization
CN110147890A (en) * 2019-05-13 2019-08-20 湖北工业大学 A kind of method and system based on lion group's algorithm optimization extreme learning machine integrated study
CN111582128A (en) * 2020-04-30 2020-08-25 电子科技大学 Mechanical fault sparse representation method based on wolf pack parameterized joint dictionary

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Rolling bearing fault feature extraction using Adaptive Resonance-based Sparse Signal Decomposition;Kaibo Wang等;《Engineering Research Express》;20210115;第1-19页 *
基于狮群优化的FastSLAM算法;周宁亚等;《计算机应用与软件》;20200712(第07期);全文 *

Also Published As

Publication number Publication date
CN113077463A (en) 2021-07-06

Similar Documents

Publication Publication Date Title
Li et al. A novel non-local means image denoising method based on grey theory
Kervrann et al. Optimal spatial adaptation for patch-based image denoising
Yan et al. Fine-grained attention and feature-sharing generative adversarial networks for single image super-resolution
CN111553858B (en) Image restoration method and system based on generation countermeasure network and application thereof
Liang et al. GIFM: An image restoration method with generalized image formation model for poor visible conditions
Zhao et al. Single image super-resolution based on deep learning features and dictionary model
CN110378924B (en) Level set image segmentation method based on local entropy
CN108470338B (en) A kind of water level monitoring method
Kasem et al. Spatial transformer generative adversarial network for robust image super-resolution
CN103080979A (en) System and method for synthesizing portrait sketch from photo
Liu et al. Hyperspectral image restoration based on low-rank recovery with a local neighborhood weighted spectral–spatial total variation model
CN115147709B (en) Underwater target three-dimensional reconstruction method based on deep learning
CN116739899A (en) Image super-resolution reconstruction method based on SAUGAN network
CN106934398B (en) Image de-noising method based on super-pixel cluster and rarefaction representation
Chen et al. Scene segmentation of remotely sensed images with data augmentation using U-net++
Tan et al. High dynamic range imaging for dynamic scenes with large-scale motions and severe saturation
CN113077463B (en) Chirplet energy-guided lion disturbance optimization ancient copper mirror X-ray fusion flaw detection method
Guan et al. DiffWater: Underwater image enhancement based on conditional denoising diffusion probabilistic model
Li et al. Super‐Resolution Reconstruction of Underwater Image Based on Image Sequence Generative Adversarial Network
Mi et al. A method of plant root image restoration based on GAN
CN114119369A (en) Automatic integral constraint super-resolution image gradient reconstruction method
CN113393387A (en) Mathematical morphology self-adaptive structural element construction method
Shao et al. Fine-grained recognition based on improved EfficientNet model
Li Texture feature extraction of a landscape design image based on the contour wave transform
Shahbakhsh et al. Enhancing face super-resolution via improving the edge and identity preserving network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant