Blood vessel image enhancement method based on rolling guide filtering and adaptive gamma correction

Abstract

The invention discloses a blood vessel image enhancement method based on rolling guide filtering and adaptive gamma correction, and belongs to the field of medical image processing, and the method comprises the following steps: decomposing an image into a plurality of scale detail layers and a base layer by using multi-scale rolling guide filtering; calculating a weight by using a correlation coefficient, and carrying out adaptive weighted fusion on the plurality of detail layers; a local median is calculated by using a sliding window, a correction parameter is obtained by applying a maximization strategy, adaptive gamma correction is performed on the fused detail layer, and a blood vessel region in the image is enhanced; performing weighted fusion on the enhanced fusion detail layer and the base layer to obtain a final enhanced blood vessel image; according to the method, a multi-scale detail layer adaptive fusion method is adopted, so that the contrast ratio of tiny blood vessels in the image is effectively improved; through an adaptive gamma correction strategy, noise amplification is reduced, and the method has better blood vessel enhancement capability and can be widely applied to the field of medical blood vessel image enhancement.

Description

Technical Field

[0001] The invention relates to the field of medical image processing, and in particular to a blood vessel image enhancement method based on rolling guided filtering and adaptive gamma correction. Background Art

[0002] Vascular images have important clinical and research value in the field of medical imaging. Accurate and clear display of vascular structure is essential for diagnosing diseases, evaluating conditions, and guiding treatment. However, the visualization quality of vascular images is often limited due to a variety of factors, such as noise, low contrast, and blur. Therefore, vascular image enhancement has become a key technology aimed at improving the visualization of vascular structures.

[0003] In the past few decades, many vascular image enhancement methods have been proposed and studied. Traditional enhancement methods are usually based on image processing techniques, such as filtering, histogram equalization, and edge enhancement. Filtering methods can smooth images, but will cause blurred vascular edges. Histogram equalization methods can increase the contrast of images, but the effect is limited for vascular images in noisy environments. Edge enhancement methods can highlight the position and shape of blood vessels, but the effect is not ideal in low-contrast environments. At present, some studies focus on the combination of multiple modules to improve image quality, such as filtering combined with histogram equalization, mathematical morphology combined with multi-scale operations, etc. Although these methods can improve the quality of images to a certain extent, they often cannot effectively highlight vascular details, especially in low-contrast and noisy environments. Therefore, there is a need for an efficient vascular image enhancement method that can highlight vascular details in low-contrast and noisy environments and improve the quality and visualization of vascular images. Summary of the invention

[0004] The present invention aims to provide a vascular image enhancement method based on rolling guided filtering and adaptive gamma correction, which uses a series of medical image processing methods to achieve vascular image enhancement, so as to effectively improve the quality and visualization effect of vascular images in low contrast and noise environments, and help doctors make more accurate diagnosis and treatment.

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

[0006] A blood vessel image enhancement method based on rolling guided filtering and adaptive gamma correction, characterized in that it comprises the following steps:

[0007] The original vascular image is decomposed into detail layers and base layers of different scales using a multi-scale rolling guided filter;

[0008] The corresponding weights are calculated using the correlation coefficients to fuse detail layers of different scales to obtain fused multi-scale detail layers;

[0009] The local median is calculated using a sliding window, and the correction parameters are calculated using a maximization strategy. The vascular area in the fused multi-scale detail layer is subjected to an improved adaptive gamma correction to obtain an enhanced fused image.

[0010] The enhanced fused image is weighted fused with the base layer to obtain an enhanced vascular image.

[0011] Preferably, the original blood vessel image is decomposed into detail layers and base layers of different scales by using a multi-scale rolling guided filter, comprising the following steps:

[0012] First, the rolling guided filter can be expressed as follows:

[0013]

[0014] Among them, J t+1 is the result of the tth iteration, K p represents normalization, p and q represent the pixel coordinates of the image, σ s and σ r They control the spatial weight and range weight respectively, N(p) is the set of neighboring pixels of p;

[0015] The rolling guided filtering of T iterations can be expressed as:

[0016] u=RGF(I,σ s ,σ r ,T)

[0017] When applied to vascular images, considering the presence of multi-scale vascular information in the image, the original image is repeatedly smoothed at different scales:

[0018]

[0019] In the formula, u k is the image after the kth layer filtering, k = 1, 2, ... L-1, L is the number of decomposition layers; in order to capture the characteristics of blood vessels from thin to thick, the initial image is set to The initial image u1 is set as the input image I;

[0020] The k-th level detail layer containing specific size information is calculated by the difference between the k-th level smoothed filtered image and the k+1-th level smoothed filtered image; the base layer contains the coarse-scale structure existing in the original image. When k=L, the k-th level filtered image is Gaussian filtered to obtain the base layer.

[0021] Preferably, the corresponding weights are calculated using the correlation coefficients to fuse the detail layers of different scales to obtain a fused multi-scale detail layer, which includes the following steps:

[0022] First, calculate the correlation coefficients between k detail layers and the original image:

[0023]

[0024] in, and are the means of the original image and the kth detail layer respectively;

[0025] Secondly, the obtained correlation coefficient is used to calculate the weight:

[0026] w k =1-corr(I,d k )

[0027] Among them, corr(I,d k ) represents the correlation coefficient between the input image and the kth detail layer calculation;

[0028] Then, multiple detail layer images are weighted fused to obtain the fused detail layer:

[0029]

[0030] Preferably, a sliding window is used to calculate the local median, a maximization strategy is applied to calculate the correction parameters, and an improved adaptive gamma correction is performed on the blood vessel area in the fused multi-scale detail layer to obtain an enhanced fused image, comprising the following steps:

[0031] First, a nonlinear function is designed to transform the fused detail image d f Convert to image b for calculating correction parameters:

[0032]

[0033] Among them, parameter a is image d f The mean of

[0034] Secondly, the local median is introduced to optimize the correction parameters. The local median can be obtained by calculating the median in the 7x7 sliding window:

[0035]

[0036] Among them, med(i,j) represents the median value in the neighborhood window centered on pixel b(i,j);

[0037] In order to reduce the amplification of noise, the maximum value strategy is adopted:

[0038] m(i,j)=max(c(i,j),b(i,j))

[0039]

[0040] In order to obtain a more natural appearance of the enhanced detail layer, a Gaussian filter is used to smooth m, and the correction parameter map is obtained as follows:

[0041] γ=Gaussian(m,σ,ksize)

[0042] Finally, adaptive gamma correction is applied to obtain the enhanced fused detail layer:

[0043]

[0044] Preferably, the enhanced fused image is weightedly fused with the base layer, wherein the weighted fusion formula can be expressed as:

[0045] F=ωΒ+(1-ω)D enhanced

[0046] Among them, ω is the weight coefficient.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention uses multi-scale theory and weighted fusion to adaptively adjust vascular features of different scales, wherein the weights in weighted fusion are obtained by the correlation coefficients of detail layers of each size, and can adaptively enhance features of each size, especially microvascular features. In view of the noise amplification problem existing in the enhancement process, the present invention proposes an improved adaptive gamma correction, which uses local information to maintain the balance between contrast enhancement and noise suppression in the enhanced image. Compared with the classical algorithm, the algorithm can effectively maintain image brightness, avoid brightness distortion, and enhance image details. In addition, it successfully suppresses noise and significantly improves the overall image quality. These advances make the proposed method highly valuable for clinical diagnosis, provide rich and valuable information, and have great potential in medical image processing and diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The flowchart of the vascular image enhancement method based on rolling guided filtering and adaptive gamma correction is shown;

[0050] Figure 2 It is a schematic diagram of the experimental effect provided in an embodiment. DETAILED DESCRIPTION

[0051] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments:

[0052] The specific implementation process is as follows:

[0053] S1. Based on the multi-scale rolling guided filter, the original vascular image is decomposed into detail layers and base layers of different scales.

[0054] In this step, since there are vascular structures of various scales in the original vascular image and the microvessels have low brightness, it is necessary to enhance their contrast in the image. Therefore, the rolling guided filter is used to extract the vascular structures of various scales in the image, that is, the detail layers of various scales. The rolling guided filter can be expressed by the following formula:

[0055]

[0056] Among them J t+1 is the result of the tth iteration, K p represents normalization, p and q represent the pixel coordinates of the image, σ s and σ r Control the spatial weight and range weight respectively, N(p) is the set of neighboring pixels of p.

[0057] The rolling guided filtering of T iterations can be expressed as:

[0058] u=RGF(I,σ s ,σ r ,T)

[0059] S11. When applied to vascular images, considering the presence of multi-scale vascular information in the image, the original image is repeatedly smoothed at different scales:

[0060]

[0061] In the formula, u k is the image after the kth layer filtering, k = 1, 2, ... L-1, L is the number of decomposition layers; in order to capture the characteristics of blood vessels from thin to thick, the initial image is set to The initial image u1 is set as the input image I;

[0062] S12, the k-th level detail layer containing specific size information is calculated by the difference between the k-th level smoothed filtered image and the k+1-th level smoothed filtered image;

[0063] S13, the base layer includes the coarse-scale structure existing in the original image. When k=L, the base layer can be obtained by performing Gaussian filtering on the k-th level filtered image.

[0064] S2. Calculate the corresponding weights using the correlation coefficients, fuse the detail layers of different scales, and obtain a fused multi-scale detail layer, including the following steps:

[0065] S21. Calculate the correlation coefficients between k detail layers and the original image:

[0066]

[0067] in, and are the means of the original image and the kth detail layer respectively;

[0068] S22. Use the obtained correlation coefficient to calculate the weight:

[0069] w k =1-corr(I,d k )

[0070] Among them, corr(I,d k ) represents the correlation coefficient between the input image and the kth detail layer calculation;

[0071] S23, weighted fusion of multiple detail layer images to obtain a fused detail layer:

[0072]

[0073] S3, using a sliding window to calculate the local median, applying a maximization strategy to calculate the correction parameters, and performing an improved adaptive gamma correction on the blood vessel area in the fused multi-scale detail layer to obtain an enhanced fused image, including the following steps:

[0074] S31. Design a nonlinear function to transform the fused detail image d f Convert to image b for calculating correction parameters:

[0075]

[0076] Among them, parameter a is image d f The mean of

[0077] S32, introducing a local median to optimize the correction parameters;

[0078] If only the intensity of each pixel is considered for gamma correction, the correction coefficients of vascular pixels and noise pixels are the same, which will inevitably lead to the enhancement of noise as the blood vessels are enhanced. In order to solve this problem, the local median is introduced to optimize the correction parameters. The local median can be obtained by calculating the median value within the 7x7 sliding window:

[0079]

[0080] Among them, med(i,j) represents the median value in the neighborhood window centered on pixel b(i,j);

[0081] S33. In order to reduce the amplification of noise, the maximum value strategy is adopted:

[0082] m(i,j)=max(c(i,j),b(i,j))

[0083]

[0084] S34. In order to obtain a more natural appearance of the enhanced detail layer, a Gaussian filter is used to smooth m, and the correction parameter map is obtained as follows:

[0085] γ=Gaussian(m,σ,ksize)

[0086] S35. Apply adaptive gamma correction to obtain an enhanced fused detail layer:

[0087]

[0088] S4. The enhanced fused image is weighted fused with the base layer, where the weighted fusion formula can be expressed as:

[0089] F=ωΒ+(1-ω)D enhanced

[0090] Among them, ω is the weight coefficient.

[0091] Figure 2 (a) is the original image, Figure 2 (h) is the image processed by the present invention. It can be seen that the present invention effectively improves the visibility and contrast of the image, and has significant advantages in maintaining the structural integrity and brightness consistency of the image. The present invention has broad application prospects in the medical field.

[0092] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions and / or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several modifications and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A vascular image enhancement method based on rolling guided filtering and adaptive gamma correction, characterized in that: The following steps are involved: S1, using a multi-scale rolling guided filter, decompose the original vascular image into detail layers and base layers of different scales; S2, using the correlation coefficient to calculate the corresponding weights, fuse the detail layers of different scales to obtain a fused multi-scale detail layer; S3, using a sliding window to calculate the local median, applying a maximization strategy to calculate the correction parameters, and performing an improved adaptive gamma correction on the vascular area in the fused multi-scale detail layer to obtain an enhanced fused image; S4. The enhanced fused image is weighted-fused with the base layer to obtain an enhanced vascular image.

2. The method for vascular image enhancement based on rolling guided filtering and adaptive gamma correction according to claim 1, characterized in that: The rolling guided filtering in step S1 can be expressed by the following formula: Among them, J t+1 is the result of the tth iteration, K p represents normalization, p and q represent the pixel coordinates of the image, σ s and σ r They control the spatial weight and range weight respectively, N(p) is the set of neighboring pixels of p; The rolling guided filtering of T iterations can be expressed as: u=RGF(I,σ s ,s r ,T) The specific steps are as follows: S11. When applied to vascular images, considering the presence of multi-scale vascular information in the image, the original image is repeatedly smoothed at different scales: In the formula, u k is the image after the kth layer filtering, k = 1, 2, ... L-1, L is the number of decomposition layers; in order to capture the characteristics of blood vessels from thin to thick, the initial image is set to The initial image u1 is set as the input image I; S12, the k-th level detail layer containing specific size information is calculated by the difference between the k-th level smoothed filtered image and the k+1-th level smoothed filtered image; S13, the base layer includes the coarse-scale structure existing in the original image. When k=L, the base layer can be obtained by performing Gaussian filtering on the k-th level filtered image.

3. The method for vascular image enhancement based on rolling guided filtering and adaptive gamma correction according to claim 1, characterized in that: The step S2 uses the correlation coefficient to calculate the corresponding weights, fuses the detail layers of different scales, and obtains a fused multi-scale detail layer, including the following steps: S21. Calculate the correlation coefficients between k detail layers and the original image: in, and are the means of the original image and the kth detail layer respectively; S22. Use the obtained correlation coefficient to calculate the weight: w k =1-corr(I,d k ) Among them, corr(I,d k ) represents the correlation coefficient between the input image and the kth detail layer calculation; S23, weighted fusion of multiple detail layer images to obtain a fused detail layer:

4. The method for vascular image enhancement based on rolling guided filtering and adaptive gamma correction according to claim 1, characterized in that: The step S3 uses a sliding window to calculate the local median, applies a maximization strategy to calculate the correction parameter, and performs an improved adaptive gamma correction on the blood vessel area in the fused multi-scale detail layer to obtain an enhanced fused image, including the following steps: S31. Design a nonlinear function to transform the fused detail image d f Convert to image b for calculating correction parameters: Among them, parameter a is image d f The mean of S32, introduce the local median to optimize the correction parameters. The local median can be obtained by calculating the median in the 7x7 sliding window: Among them, med(i,j) represents the median value in the neighborhood window centered on pixel b(i,j); S33. In order to reduce the amplification of noise, the maximum value strategy is adopted: m(i,j)=max(c(i,j),b(i,j)) S34. In order to obtain a more natural appearance of the enhanced detail layer, a Gaussian filter is used to smooth m, and the correction parameter map is obtained as follows: γ=Gaussian(m,σ,ksize) S35. Apply adaptive gamma correction to obtain an enhanced fused detail layer:

5. The method for vascular image enhancement based on rolling guided filtering and adaptive gamma correction according to claim 1, characterized in that: The enhanced fused image in step S4 is weighted fused with the base layer, where the weighted fusion formula can be expressed as: F=ωB+(1-ω)D enhanced Among them, ω is the weight coefficient.

Images