CN113610717B - Enhancement method for ultraviolet fluorescence image of skin disease - Google Patents
Enhancement method for ultraviolet fluorescence image of skin disease Download PDFInfo
- Publication number
- CN113610717B CN113610717B CN202110806197.8A CN202110806197A CN113610717B CN 113610717 B CN113610717 B CN 113610717B CN 202110806197 A CN202110806197 A CN 202110806197A CN 113610717 B CN113610717 B CN 113610717B
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- frequency
- gray
- noise
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000002073 fluorescence micrograph Methods 0.000 title claims abstract description 34
- 208000017520 skin disease Diseases 0.000 title claims description 10
- 230000009466 transformation Effects 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 18
- 230000002708 enhancing effect Effects 0.000 claims abstract description 5
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 34
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 5
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 16
- 206010008570 Chloasma Diseases 0.000 description 6
- 208000003351 Melanosis Diseases 0.000 description 6
- 206010047642 Vitiligo Diseases 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- XUMBMVFBXHLACL-UHFFFAOYSA-N Melanin Chemical compound O=C1C(=O)C(C2=CNC3=C(C(C(=O)C4=C32)=O)C)=C2C4=CNC2=C1C XUMBMVFBXHLACL-UHFFFAOYSA-N 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 239000000049 pigment Substances 0.000 description 2
- 102000008186 Collagen Human genes 0.000 description 1
- 108010035532 Collagen Proteins 0.000 description 1
- 208000022535 Infectious Skin disease Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229920001436 collagen Polymers 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000002500 effect on skin Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000799 fluorescence microscopy Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 206010040872 skin infection Diseases 0.000 description 1
- 206010040882 skin lesion Diseases 0.000 description 1
- 231100000444 skin lesion Toxicity 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
- G06T7/0012—Biomedical image inspection
-
- 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
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
-
- 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
- G06T2207/20076—Probabilistic image processing
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30088—Skin; Dermal
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Image Processing (AREA)
Abstract
A method for enhancing ultraviolet fluorescence images for dermatological disorders, comprising: dividing the original fluorescent image into a low-frequency image and a high-frequency image by wavelet decomposition; the pixel pair gray stretching algorithm is adopted to process the low-frequency image, and the number of pixel pairs needing to be counted is greatly reduced; processing the high-frequency image by using a denoising method based on weighted TV regularization; the low frequency enhancement signal and the high frequency denoising signal are reconstructed by wavelet inverse transformation. The algorithm of the invention combines the advantages of the enhancement algorithm and the denoising algorithm to the greatest extent, not only maintains the singularity of the fluorescent image and improves the definition and contrast of the fluorescent image, but also obviously weakens the noise, displays the detail information of the fluorescent image, does not have excessive enhancement phenomenon, and has good visual effect.
Description
Technical Field
The invention relates to the technical field of fluorescent image processing, in particular to a dermatological ultraviolet fluorescent image enhancement algorithm based on pixel pair gray stretching and weighted TV regularization.
Background
After the skin is irradiated by ultraviolet light with a specific wavelength, melanin and dermal collagen can absorb the ultraviolet light; then, the blue light is emitted as fluorescence of main color, and different skin lesions can generate fluorescence of different degrees and colors, so as to reflect different degrees and types of skin diseases. Can assist doctors to make accurate diagnosis on skin diseases such as pigment abnormal diseases, skin infection and the like according to the principle.
The ultraviolet fluorescence imaging technology can image fluorescence emitted by the skin under the irradiation of ultraviolet light, however, because the fluorescence signal belongs to a weak signal, the fluorescent image of the skin disease obtained by the system is darker in brightness and lower in contrast, so that details of the fluorescent image in a plurality of dark areas cannot be distinguished, and misdiagnosis of a dermatologist is caused. Because the fluorescent image with high quality and high precision is an important basis for diagnosing skin diseases by doctors, the imaging precision is improved along with the development of artificial intelligence and machine learning, so that further intelligent analysis is facilitated, and a foundation is laid for automatic identification of skin diseases in the future. Therefore, it is necessary to complete enhancement of the fluorescence image.
Disclosure of Invention
The invention aims to provide a fluorescence image enhancement algorithm based on pixel-to-gray stretching and weighted TV regularization, so as to effectively solve the problems of dark brightness, low contrast, loss of details and the like of a fluorescence image in the background art, improve the diagnosis accuracy and efficiency of skin diseases and realize early discovery and early treatment of pigment and infectious skin diseases.
In order to achieve the above purpose, the technical scheme of the invention is as follows:
a fluorescence image enhancement algorithm based on pel-to-gray stretching and weighted TV regularization, comprising:
first, dividing an original fluorescent image into a low-frequency image and a high-frequency image by wavelet decomposition; secondly, a pixel-to-gray stretching algorithm is used for processing the low-frequency image, and the number of pixel pairs needing statistics is greatly reduced; processing the high-frequency image by using a denoising method based on weighted TV regularization; and finally, reconstructing the low-frequency enhancement signal and the high-frequency denoising signal through wavelet inverse transformation to obtain the fluorescence image with high definition contrast after processing.
The method comprises the following specific implementation steps:
step 1: the low frequency sub-band and the high frequency sub-band of the original image are separated by wavelet transform. For an mxn two-dimensional image f (x, y), its two-dimensional Discrete Wavelet Transform (DWT) is defined as:
in 1, j 0 As a function of the initial scale parameter(s),is an approximate coefficient, +.>Represented by f (x, y) is j 0 An approximation of the location; />As detail coefficient, when j is greater than or equal to j 0 When (I)>With three directional details, H represents the horizontal direction, V represents the vertical direction, and D represents the diagonal direction. />And->The method comprises the steps of respectively obtaining a scale function and a wavelet function, wherein the two functions are two-dimensionally separable;
step 2: after the low-frequency component and the high-frequency component of the image are separated by wavelet transformation, the main body energy of the image is in the low-frequency component, and the low-frequency component is enhanced by using a pixel to gray stretching algorithm, so that the aim of enhancing the local contrast of the image is fulfilled, and the main body display effect of the image can be effectively improved;
firstly, defining a non-weighted gray scale stretching coefficient g between pixels p and q pq ：
In formula 2, I P Is the gray value of the pixel p, I q For the gray value of pixel q, delta g The value of (2) isI max Is the maximum value, delta, of the gray values in the original image g- The value of (1) is delta d The value of (2). d (p, q) is between pixel p and pixel qEuclidean distance.
Secondly, a definition weight theta is introduced p Its purpose is to reduce image noise interference, θ p The formula of (2) is:
in formula 3, the gray value of the pixel r is the same as the gray value of the pixel p, d (p, r) is the Euclidean distance of the two pixels, delta θ The value of (2) is equal to 1.θ p Representing the probability that the picture element p is a non-noise point.
Then, theta is set p And theta q Multiplying and then squaring, then adding g pq Multiplying to obtain the gray scale stretching coefficient c with definition weight pq ：
Finally, setting the gray level transformation function as T, wherein T is a global transformation function, and setting the gray level value I of the acquired image p Mapping intoIn order for T to enhance the local contrast of a fluorescence image, T needs to have the following characteristics: i T (I) p )-T(I q ) I and c pq Proportional to the ratio. However, the number of pixel pairs in the fluorescent image is too large, and the equation set of T is an overdetermined equation set, and an approximate solution is needed to be found.
Define pixel P and set of pixels { q|I having gray scale difference x from it q ＝I p Cumulative gray scale stretching coefficient C of +x p (x) The method comprises the following steps:
let x >0, in order to guarantee monotonicity of T, make it satisfy:
in equation 6, α is a normalized constant. H (I) p ) Is shown below, which represents the first order difference of the transformation function T.
H(I p )＝T(I p )-T(I p -1) (7)
Combining formula 5 and formula 6, push out:
H(I p +x)＝α×C p (x) (8)
the above equation shows that each pixel p in the input image can yield a system of linear equations:
H＝αC p (9)
in 9
H＝[H(1),H(2),...,H(255)] T (10)
C p ＝[C p (-I p ),...,C p (-1),C p (1),...,C p (255-I p )] T (11)
The cost function J (H) with weights is defined as:
w in 12 p Is a diagonal matrix, and the value of the ith diagonal element is 2/(1+exp (|i-I) p I) are provided. The matrix H is subjected to bias derivative by J (H), and an optimal solution H can be obtained by taking extreme points * The method comprises the following steps:
alpha makes H * Is 255. Then, in combination with equation 7, the equation for T can be derived as:
t (0) =0 of the above formula. Although T is a global transformation function, T can be deduced from the gray scale stretching coefficients, so that it is better in local adaptation.
Step 3: a denoising optimization framework consisting of data fidelity and weighted TV regularization is designed, which is based on two weights: the ratio of the mean to variance of the intensities and the directionality of the edges are then optimized for the frame using the Douglas-Rachford segmentation method.
Detail information such as edges of the image and noise are included in the high-frequency image, and characteristics of the noise level of the fluorescent image closely related to the average intensity are considered. The ratio of the mean and variance of the intensities is to control the degree of smoothness and the directionality of the edges is to preserve the details of the image.
First, taking into account poisson distribution of low-light noise, a high-frequency image R contaminated with noise P The original image S is restored P . Thus, each M P The discrete poisson probability of (a) is as follows:
the predicted image S should be most similar to the input image M. Thus, bayesian law is used herein as follows:
maximizing P (m|s) P (S). Typically, the data fidelity term is derived by maximizing the posterior probability density P (m|s). Then, the following steps are obtained:
TV regularization was chosen as a priori distribution as follows:
where λ is the regularization parameter.
To increase computational efficiency, minimize-log (P (R|S) P (S)) is used instead of maximize P (R|S) P (S)
The following function is then minimized:
second, it is assumed that the noise is signal independent and has the same variance throughout the image. While in low light conditions, the photon counting process is the primary source of sensor noise corrupting the image. In this case, the statistics of the noise image are described by a multi-valued poisson process, the amount of noise being controlled by the total number of photons: the higher the photon count, the lower the noise damage to the image. Thus, the variation of noise is different in bright and dark areas. To solve this problem, the degree of smoothness is controlled using the ratio of the intensity variance to the mean value in the region as a weight. By means of the deviation of the pixel intensities, a simple measure B can be obtained p For estimating the scale within the region, as follows:
in sigma p As average absolute deviation, x p Is the average pixel intensity. Using the above statistics, a variance-mean weighted graph is obtained as follows:
V p ＝B p ×x p (21)
in addition, edge directivity is used as another weight L p To distinguish noise from edge details as follows:
wherein g p,q Is a weight function, which is obtained by the following formula:
and depends on whether the gradients in W are uniform. If it contains only noise, L is generally smaller than it even on the edges. For complex modes, noise-to-edge mapping is used to control the smoothness of the noise and edges, since the edges of the local window have a greater impact on the gradient direction than the noise. In weighted TV optimization, V is used P And L P A more efficient regularized denoising is formed. The objective function is expressed as:
the Alternate Direction Multiplier Method (ADMM) cannot be used to solve the minimum problem because it does not use quadratic data fidelity terms. Thus, the Douglas-Rachford splitting method is used to minimize (25) as follows:
ψ(S)＝S-IlogS，updating S and d for denoising the fluorescent image.
Step 4: f (x, y) can be reconstructed using the inverse wavelet transform to obtain the final enhancement result. The formula is as follows:
compared with the prior art, the invention has the beneficial effects that:
the fluorescence image enhancement algorithm not only maintains the singularity of the fluorescence image and improves the definition and contrast of the fluorescence image, but also remarkably reduces noise. The detail information of the fluorescent image is displayed, the phenomenon of excessive enhancement is not generated, and the visual effect is good. The method has balanced enhancement effect on the fluorescence image.
Drawings
FIG. 1 is a general block diagram of an image enhancement algorithm
FIG. 2 (a) is a diagram of vitiligo fluorescence image;
FIG. 2 (b) shows the effect of histogram equalization on vitiligo fluorescence images;
FIG. 2 (c) shows the effect of the method of the present invention on the treatment of vitiligo fluorescence images;
FIG. 3 (a) is an original representation of hypopigmented fluorescence images;
FIG. 3 (b) shows the effect of histogram equalization on hypopigmented fluorescent images;
FIG. 3 (c) shows the effect of the method of the present invention on the processing of hypopigmented fluorescence images;
FIG. 4 (a) is an original image of chloasma fluorescence;
FIG. 4 (b) shows the effect of histogram equalization on chloasma fluorescent images;
FIG. 4 (c) shows the effect of the method of the present invention on the treatment of chloasma fluorescent images.
Detailed Description
The technical scheme of the present invention is further elaborated in the following with reference to the drawings and the specific embodiments, which are only illustrative of the present invention and are not intended to limit the present invention.
The invention provides a fluorescence image enhancement algorithm based on pixel to gray stretching and weighting TV regularization, the general block diagram of the algorithm is shown in figure 1, firstly, the original fluorescence image is divided into a low-frequency image and a high-frequency image by wavelet decomposition; secondly, a pixel-to-gray stretching algorithm is used for processing the low-frequency image, and the number of pixel pairs needing statistics is greatly reduced; processing the high-frequency image by using a denoising method based on weighted TV regularization; and finally, reconstructing the low-frequency enhancement signal and the high-frequency denoising signal through wavelet inverse transformation to obtain the fluorescence image with high definition contrast after processing. The following steps are specifically implemented by using fig. 2 (a), fig. 3 (a) and fig. 4 (a) as original (weak light) fluorescence images for enhancement:
step 1: the low frequency sub-band and the high frequency sub-band of the original image are separated by wavelet transform. For an mxn two-dimensional image f (x, y), its two-dimensional Discrete Wavelet Transform (DWT) is defined as:
in 1, j 0 As a function of the initial scale parameter(s),is an approximate coefficient, +.>Represented by f (x, y) is j 0 An approximation of the location; />As detail coefficient, when j is greater than or equal to j 0 When (I)>With three directional details, H represents the horizontal direction, V represents the vertical direction, and D represents the diagonal direction. />And->The scale function and the wavelet functionBoth functions are two-dimensionally separable.
Step 2: after the low-frequency component and the high-frequency component of the image are separated by wavelet transformation, the main body energy of the image is in the low-frequency component, and the low-frequency component is enhanced by using a pixel to gray stretching algorithm, so that the aim of enhancing the local contrast of the image is fulfilled, and the main body display effect of the image can be effectively improved.
Firstly, defining a non-weighted gray scale stretching coefficient g between pixels p and q pq ：
In formula 2, I P Is the gray value of the pixel p, I q For the gray value of pixel q, delta g The value of (2) isI max Is the maximum value, delta, of the gray values in the original image g- The value of (1) is delta d The value of (2). d (p, q) is the Euclidean distance between pixel p and pixel q.
Secondly, a definition weight theta is introduced p Its purpose is to reduce image noise interference, θ p The formula of (2) is:
in formula 3, the gray value of the pixel r is the same as the gray value of the pixel p, d (p, r) is the Euclidean distance of the two pixels, delta θ The value of (2) is equal to 1.θ p Representing the probability that the picture element p is a non-noise point.
Then, theta is set p And theta q Multiplying and then squaring, then adding g pq Multiplying to obtain the gray scale stretching coefficient c with definition weight pq ：
Finally, setting the gray level transformation function as T, wherein T is a global transformation function, and setting the gray level value I of the acquired image p Mapping intoIn order for T to enhance the local contrast of a fluorescence image, T needs to have the following characteristics: i T (I) p )-T(I q ) I and c pq Proportional to the ratio. However, the number of pixel pairs in the fluorescent image is too large, and the equation set of T is an overdetermined equation set, and an approximate solution is needed to be found.
Define pixel P and set of pixels { q|I having gray scale difference x from it q ＝I p Cumulative gray scale stretching coefficient C of +x p (x) The method comprises the following steps:
let x >0, in order to guarantee monotonicity of T, make it satisfy:
in equation 6, α is a normalized constant. H (I) p ) Is shown below, which represents the first order difference of the transformation function T.
H(I p )＝T(I p )-T(I p -1) (7)
Combining formula 5 and formula 6, push out:
H(I p +x)＝α×C p (x) (8)
the above equation shows that each pixel p in the input image can yield a system of linear equations:
H＝αC p (9)
in 9
H＝[H(1),H(2),...,H(255)] T (10)
C p ＝[C p (-I p ),...,C p (-1),C p (1),...,C p (255-I p )] T (11)
The cost function J (H) with weights is defined as:
w in 12 p Is a diagonal matrix, and the value of the ith diagonal element is 2/(1+exp (|i-I) p I) are provided. The matrix H is subjected to bias derivative by J (H), and an optimal solution H can be obtained by taking extreme points * The method comprises the following steps:
alpha makes H * Is 255. Then, in combination with equation 7, the equation for T can be derived as:
t (0) =0 of the above formula. Although T is a global transformation function, T can be deduced from the gray scale stretching coefficients, so that it is better in local adaptation.
Step 3: detail information such as edges of the image and noise are included in the high-frequency image, and characteristics of the noise level of the fluorescent image closely related to the average intensity are considered. Thus, a denoising optimization framework consisting of data fidelity and weighted TV regularization is designed here. This framework is based on two weights: the ratio of the mean to the variance of the intensity and the directionality of the edge. The ratio of the mean and variance of the intensities is to control the degree of smoothness and the directionality of the edges is to preserve the details of the image. The framework was then optimized using the Douglas-Rachford segmentation method.
First, considering poisson distribution of low-light noise, one wants to obtain a high-frequency image R contaminated with noise P The original image S is restored P . Thus, each M P The discrete poisson probability of (a) is as follows:
the predicted image S should be most similar to the input image M. Thus, bayesian law is used herein as follows:
maximizing P (m|s) P (S). Typically, the data fidelity term is derived by maximizing the posterior probability density P (m|s). Then, the following steps are obtained:
TV regularization was chosen as a priori distribution as follows:
where λ is the regularization parameter.
To increase computational efficiency, minimize-log (P (R|S) P (S)) is used instead of maximize P (R|S) P (S)
The following function is then minimized:
second, it is assumed that the noise is signal independent and has the same variance throughout the image. While in low light conditions, the photon counting process is the primary source of sensor noise corrupting the image. In this case, the statistics of the noise image are described by a multi-valued poisson process, the amount of noise being controlled by the total number of photons: the higher the photon count, the lower the noise damage to the image. Thus, the variation of noise is different in bright and dark areas. To solve this problem, the degree of smoothness is controlled using the ratio of the intensity variance to the mean value in the region as a weight. By variation of pixel intensityObtain a simple metric B p For estimating the scale within the region, as follows:
in sigma p As average absolute deviation, x p Is the average pixel intensity. Using the above statistics, a variance-mean weighted graph is obtained as follows:
V p ＝B p ×x p (21)
in addition, edge directivity is used as another weight L p To distinguish noise from edge details as follows:
wherein g p,q Is a weight function, which is obtained by the following formula:
and depends on whether the gradients in W are uniform. If it contains only noise, L is generally smaller than it even on the edges. For complex modes, noise-to-edge mapping is used to control the smoothness of the noise and edges, since the edges of the local window have a greater impact on the gradient direction than the noise. In weighted TV optimization, V is used P And L P A more efficient regularized denoising is formed. The objective function is expressed as:
the Alternate Direction Multiplier Method (ADMM) cannot be used to solve the minimum problem because it does not use quadratic data fidelity terms. Thus, the Douglas-Rachford splitting method is used to minimize (25) as follows:
ψ(S)＝S-IlogS，updating S and d for denoising the fluorescent image.
Step 4: f (x, y) can be reconstructed using the inverse wavelet transform to obtain the final enhancement result.
The formula is as follows:
in order to verify the effectiveness of the fluorescence image enhancement algorithm provided by the invention, an enhancement experiment is carried out on the fluorescence image of the skin disease, and the fluorescence image is compared with the traditional image enhancement algorithm. Fig. 2 (a) is an original diagram of a vitiligo fluorescent image, fig. 2 (b) is a processing effect of a histogram equalization method on the vitiligo fluorescent image, and fig. 2 (c) is a processing effect of the method of the invention on the vitiligo fluorescent image; fig. 3 (a) is an original image of a hypopigmented fluorescent image, fig. 3 (b) is a processing effect of a histogram equalization method on the hypopigmented fluorescent image, and fig. 3 (c) is a processing effect of the method of the present invention on the hypopigmented fluorescent image; fig. 4 (a) is an original image of a chloasma fluorescent image, fig. 4 (b) is a processing effect of a histogram equalization method on the chloasma fluorescent image, and fig. 4 (c) is a processing effect of the method of the invention on the chloasma fluorescent image. From the figure, it can be seen that the original image is a weak fluorescence image, and the overall brightness of the fluorescence image is improved after histogram equalization, but there is a case where excessive enhancement occurs. The fluorescence image enhancement algorithm not only improves the definition and contrast of the fluorescence image, but also remarkably reduces noise and displays detail information of the fluorescence image.
For objective evaluation of the method of the present invention, the average gradient, information entropy and standard deviation of the fluorescence images before and after enhancement were calculated as shown in tables 1 to 3.
Table 1 average gradient contrast of the enhanced results of the two algorithms
Table 2 comparison of information entropy of enhanced results of two algorithms
Table 3 standard deviation comparison of enhanced results of two algorithms
Experimental results show that the traditional enhancement algorithm can also improve the average gradient, the information entropy and the standard deviation, but compared with the algorithm in the invention, the effect is not obvious enough. The enhancement algorithm provided by the invention increases the average gradient by approximately 2.29 times, so that the image is clearer, the standard deviation is increased by 0.43 times, and the contrast of the fluorescent image is obviously improved. The information entropy is increased by 0.1 times, and details in the image can be displayed.
Although the invention has been described above with reference to the accompanying drawings, the invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many modifications may be made by those of ordinary skill in the art without departing from the spirit of the invention, which fall within the protection of the invention.
Claims (5)
1. A method for enhancing ultraviolet fluorescence images of skin disorders, comprising the steps of:
s1: dividing the original fluorescent image into a low-frequency image and a high-frequency image by wavelet decomposition;
s2: the pixel pair gray stretching algorithm is used for processing the low-frequency image, so that the number of pixel pairs needing to be counted is greatly reduced; processing the high-frequency image by using a denoising method based on weighted TV regularization;
s3: reconstructing the low-frequency enhanced signal and the high-frequency denoising signal through wavelet inverse transformation to obtain a fluorescence image with high definition contrast after processing;
the step S1 includes:
for an mxn two-dimensional image f (x, y), its two-dimensional discrete wavelet transform is defined as:
in 1, j 0 As a function of the initial scale parameter(s),is an approximate coefficient, +.>Represented by f (x, y) is j 0 An approximation of the location; />As detail coefficient, when j is greater than or equal to j 0 When (I)>With three-directional details, H represents the horizontal direction, V represents the vertical direction, D represents the diagonal direction, < >>And->The scale function and the wavelet function are respectively, and the two functions are two-dimensionally separable.
2. The method according to claim 1, wherein the step S2 comprises:
s21: after the low-frequency component and the high-frequency component of the image are separated by wavelet transformation, the main energy of the image is in the low-frequency component, and the pixel is used for enhancing the low-frequency component by using a gray stretching algorithm;
s22: constructing a denoising optimization framework consisting of data fidelity and weighted TV regularization, the framework being based on two weights: the ratio of the mean to variance of the intensities and the directionality of the edges are then optimized for the frame using the Douglas-Rachford segmentation method.
3. The method according to claim 2, wherein the step S21 includes:
firstly, defining a non-weighted gray scale stretching coefficient g between pixels p and q pq ：
In formula 2, I P Is the gray value of the pixel p, I q For the gray value of pixel q, delta g The value of (2) isI max Is the maximum value, delta, of the gray values in the original image g- The value of (1) is delta d The value of (2); d (p, q) is the Euclidean distance between pixel p and pixel q;
secondly, a definition weight theta is introduced p Reduce image noise interference, θ p The formula of (2) is:
in formula 3, the gray value of the pixel r is the same as the gray value of the pixel p, d (p, r) is the Euclidean distance of the two pixels, delta θ The value of (2) is equal to 1; θ p Representing the probability of non-noise points of the pixel p;
then, theta is set p And theta q Multiplying and then squaring, then adding g pq Multiplying to obtain the gray scale stretching coefficient c with definition weight pq ：
Finally, setting the gray level transformation function as T, wherein T is a global transformation function, and setting the gray level value I of the acquired image p Mapping into
Define pixel P and set of pixels { q|I having gray scale difference x from it q ＝I p Cumulative gray scale stretching coefficient C of +x p (x) The method comprises the following steps:
let x >0, in order to guarantee monotonicity of T, make it satisfy:
in formula 6, α is a normalized constant; h (I) p ) The formula of (1) is as follows, which represents the first order difference of the transformation function T;
H(I p )＝T(I p )-T(I p -1) (7)
combining formula 5 and formula 6, push out:
H(I p +x)＝α×C p (x) (8)
the above equation shows that each pixel p in the input image can yield a system of linear equations:
H＝αC p (9)
in 9
H＝[H(1),H(2),...,H(255)] T (10)
C p ＝[C p (-I p ),...,C p (-1),C p (1),...,C p (255-I p )] T (11)
The cost function J (H) with weights is defined as:
w in 12 p Is a diagonal matrix, and the value of the ith diagonal element is 2/(1+exp (|i-I) p I)), is set at the right angle; the matrix H is subjected to bias derivative by J (H), and an optimal solution H can be obtained by taking extreme points * The method comprises the following steps:
alpha makes H * Is 255; then, in combination with equation 7, the equation for T can be derived as:
where T (0) =0.
4. The method according to claim 2, wherein the step S22 includes:
first, taking into account poisson distribution of low-light noise, a high-frequency image R contaminated with noise P The original image S is restored P The method comprises the steps of carrying out a first treatment on the surface of the Thus, each M P Is a discrete of (a)The poisson probability is as follows:
the Bayesian law is used as follows:
maximizing P (m|s) P (S) gives:
TV regularization was chosen as a priori distribution as follows:
where λ is a regularization parameter;
the minimization-log (P (r|s) P (S)) is used instead of maximizing P (r|s) P (S), and then the following function is minimized:
controlling the smoothness by using the ratio of the intensity variance to the mean value in the region as a weight; by the deviation of the pixel intensities, a simple measure B is obtained p For estimating the scale within the region, as follows:
in sigma p As average absolute deviation, x p Is the average pixel intensity; obtaining the square by using the statistical dataThe difference-mean weighting map is as follows:
V p ＝B p ×x p (21)
in addition, edge directivity is used as another weight L p To distinguish noise from edge details as follows:
wherein g p,q Is a weight function, which is obtained by the following formula:
using a noise-to-edge map to control the degree of smoothness of the noise and edges; in weighted TV optimization, V is used P And L P Forming more efficient regularized denoising; the objective function is expressed as:
the Douglas-Rachford splitting method was used to minimize (25) as follows:
ψ(S)＝S-IlogS，updating S and d for denoising the fluorescent image.
5. The method according to claim 4, wherein the step S3 includes:
reconstructing f (x, y) using wavelet inverse transformation to obtain a final enhancement result, as follows:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110806197.8A CN113610717B (en) | 2021-07-16 | 2021-07-16 | Enhancement method for ultraviolet fluorescence image of skin disease |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110806197.8A CN113610717B (en) | 2021-07-16 | 2021-07-16 | Enhancement method for ultraviolet fluorescence image of skin disease |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113610717A CN113610717A (en) | 2021-11-05 |
| CN113610717B true CN113610717B (en) | 2024-01-19 |
Family
ID=78337708
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110806197.8A Active CN113610717B (en) | 2021-07-16 | 2021-07-16 | Enhancement method for ultraviolet fluorescence image of skin disease |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113610717B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116894951B (en) * | 2023-09-11 | 2023-12-08 | 济宁市质量计量检验检测研究院(济宁半导体及显示产品质量监督检验中心、济宁市纤维质量监测中心) | Jewelry online monitoring method based on image processing |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296600A (en) * | 2016-08-01 | 2017-01-04 | 大连理工大学 | A kind of contrast enhancement process decomposed based on wavelet image |
| CN107895356A (en) * | 2017-12-04 | 2018-04-10 | 山东大学 | A kind of near-infrared image Enhancement Method based on steerable pyramid |
| CN108961172A (en) * | 2018-05-17 | 2018-12-07 | 贵州莜桔西科技有限公司 | A kind of method for enhancing picture contrast based on Gamma correction |
| CN109064413A (en) * | 2018-07-03 | 2018-12-21 | 赛诺微医疗科技（浙江）有限公司 | Method for enhancing picture contrast and the Image Acquisition Medical Devices for using it |
| CN111553863A (en) * | 2020-04-30 | 2020-08-18 | 河南大学 | Image enhancement method based on non-convex full-variation typing regularization |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6463181B2 (en) * | 2000-12-22 | 2002-10-08 | The United States Of America As Represented By The Secretary Of The Navy | Method for optimizing visual display of enhanced digital images |
| US9916655B2 (en) * | 2013-06-07 | 2018-03-13 | Paul Scherrer Institut | Image fusion scheme for differential phase contrast imaging |
-
2021
- 2021-07-16 CN CN202110806197.8A patent/CN113610717B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296600A (en) * | 2016-08-01 | 2017-01-04 | 大连理工大学 | A kind of contrast enhancement process decomposed based on wavelet image |
| CN107895356A (en) * | 2017-12-04 | 2018-04-10 | 山东大学 | A kind of near-infrared image Enhancement Method based on steerable pyramid |
| CN108961172A (en) * | 2018-05-17 | 2018-12-07 | 贵州莜桔西科技有限公司 | A kind of method for enhancing picture contrast based on Gamma correction |
| CN109064413A (en) * | 2018-07-03 | 2018-12-21 | 赛诺微医疗科技（浙江）有限公司 | Method for enhancing picture contrast and the Image Acquisition Medical Devices for using it |
| CN111553863A (en) * | 2020-04-30 | 2020-08-18 | 河南大学 | Image enhancement method based on non-convex full-variation typing regularization |
Non-Patent Citations (2)
| Title |
|---|
| 基于小波变换的低对比度图像增强方法;杨静;;计算机时代(第01期);16-18 * |
| 小波变换及其在医学图像处理中的应用;杨艳妮;严碧歌;;现代生物医学进展(第11期);103-104 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113610717A (en) | 2021-11-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rao | Dynamic histogram equalization for contrast enhancement for digital images | |
| Singh et al. | Principal component analysis-based low-light image enhancement using reflection model | |
| WO2019091270A1 (en) | Image enhancement method and system | |
| CN108765336B (en) | Image defogging method based on dark and bright primary color prior and adaptive parameter optimization | |
| CN110246106B (en) | NSST domain flotation froth image enhancement and denoising method based on quantum harmony search fuzzy set | |
| CN114529475B (en) | Image enhancement method and system based on two-dimensional gamma correction and tone mapping | |
| Wang et al. | Low-light image joint enhancement optimization algorithm based on frame accumulation and multi-scale Retinex | |
| CN113610717B (en) | Enhancement method for ultraviolet fluorescence image of skin disease | |
| CN115496695A (en) | High-dynamic infrared image self-adaptive enhancement and compression method | |
| CN117252773A (en) | Image enhancement method and system based on self-adaptive color correction and guided filtering | |
| Jalab et al. | A new medical image enhancement algorithm based on fractional calculus | |
| CN109003238A (en) | A kind of image haze minimizing technology based on model and histogram and grey level enhancement | |
| CN115578660A (en) | Land block segmentation method based on remote sensing image | |
| Kumar et al. | Spatial mutual information based detail preserving magnetic resonance image enhancement | |
| Singh et al. | Image enhancement by adaptive power-law transformations | |
| CN117611501A (en) | Low-illumination image enhancement method, device, equipment and readable storage medium | |
| Weichao et al. | Research on color image defogging algorithm based on MSR and CLAHE | |
| CN115456912A (en) | Tone mapping method based on multi-scale WLS filtering fusion | |
| CN110852977B (en) | Image enhancement method for fusing edge gray level histogram and human eye visual perception characteristics | |
| Hou et al. | NLHD: a pixel-level non-local retinex model for low-light image enhancement | |
| Dolly et al. | Various methods of enhancement in colored images: a review | |
| Chen et al. | Method for Correcting Low-illumination Images Based on Adaptive Two-dimensional Gamma Function | |
| CN115937016B (en) | Contrast enhancement method for guaranteeing image details | |
| CN110796609A (en) | Low-light image enhancement method based on scale perception and detail enhancement model | |
| Kumar et al. | A Combinational Histogram Equalization Related to Contrast Image Enhancement |
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 |