CN109801241A - A kind of solar flare image based on modified dark priority algorithm removes cloud method - Google Patents

Abstract

A kind of solar flare image based on modified dark priority algorithm removes cloud method, comprising: describes the concept in channel using mathematic(al) representation；Assuming that cloud atlas piece expression formula: I (x)=J (x) t (x)+A (1-t (x)) seeks global atmosphere light value A.Size from dark channel diagram according to brightness takes preceding 0.1% pixel.In these positions, the value of the corresponding point with maximum brightness is found in original foggy image, as A value.Bilateral filtering is used when solving t (x) to improve.Calculate the local mean value and Local standard deviation of dark image D (x, y): according to known airlight vector A, calculation optimization projects graph expression formula.In view of when the value very little of transmission plot t, the value that will lead to J is bigger than normal, so that it is whole excessive to white field to make image, therefore a state value t generally can be set₀, when t value is less than t₀When, enable t=t₀.Therefore the image expression formula finally restored.A kind of solar flare image based on modified dark priority algorithm of the present invention removes cloud method, by modified dark concept to cloud and mist image restoring, goes cloud effect with good.

Description

A method for removing clouds from solar flare images based on an improved dark channel priority algorithm

technical field

The invention belongs to the technical field of image processing, and in particular relates to a method for removing clouds from a solar flare image based on an improved dark channel priority algorithm.

Background technique

The sun is the star most closely related to human beings, and has an inseparable relationship with human life and production activities. The sun's atmosphere is filled with magnetic fields that store enormous amounts of magnetic energy. When too much magnetic energy is stored in the magnetic field, it is released through solar eruptions, one of the most violent forms of solar activity. The eruption of flares will affect the Earth's magnetic field and the ionosphere above it, which in turn affects human activities such as satellite navigation and radio communications. Therefore, the understanding and observation of solar flare phenomena are also the practical needs of people to realize space exploration and take preventive measures.

Because of the influence of the earth's atmospheric clouds when observing the solar flare phenomenon, it is particularly important to de-cloud the flare image. The dark channel prior theory proposed by Dr. He Mingkai in 2009 can effectively de-cloud the image, but due to the minimum filtering used in the dark channel, the obtained transmittance contains halo effect and block effect. In order to solve this problem , using the soft-matting and guided filtering optimization algorithm (here, the soft-matting algorithm is selected for comparison) to optimize the transmittance, in which the soft-matting can eliminate the halo phenomenon and the block phenomenon well, but its time complexity is greatly increased; The time complexity of the guided filtering algorithm is small, but the restored image still has a certain degree of cloud in the edge area. Therefore, the bilateral filtering algorithm can be used to optimize the soft-matting algorithm and the guided filtering to optimize the calculation of transmittance.

SUMMARY OF THE INVENTION

In order to solve the above technical problem of removing clouds from solar flare images, the present invention provides a method for removing clouds from solar flare images based on an improved dark channel priority algorithm. The improved dark channel concept is used to restore cloud and fog images and has a good cloud removal effect. .

The technical scheme adopted in the present invention is:

A method for removing clouds from solar flare images based on an improved dark channel priority algorithm, comprising the following steps:

Step 1: The concept of channel is described by mathematical expression. For image J, the dark channel J _dark can be expressed as the formula:

In the above formula, J _c represents the color channel image of image J, Ω(x) represents a field range, and its pixel center is x, y is any point in the field range of Ω(x), and c represents r, g, b three channel, then J _c (y) represents the image of channel c at y in the domain of Ω(x). Its meaning is to find the minimum value of the three components of RGB, and then perform minimum value filtering on the single-channel image. The dark channel prior points out: J _dark → 0, which is equivalent to J _dark ≈ 0, so the real J _c can be obtained according to this condition.

Step 2: Suppose the cloud picture expresses the formula (2):

I(x)=J(x)t(x)+A(1-t(x)) (2)

Among them, I(x) represents the image to be processed, J(x) represents the real image, t(x) represents the transmittance, which represents the part of the light that can reach the computer system without being scattered, and A represents the global atmospheric light value.

Step 3: Obtain the global atmospheric light value A. Take the first 0.1% of the pixels from the dark channel map according to the size of the brightness. In these positions, the value of the corresponding point with the highest brightness in the original foggy image is found as the A value.

A bilateral filtering improvement is used in solving t(x).

Step 4: Calculate the local mean and local standard deviation of the dark image D(x,y), and then estimate the atmospheric light curtain through the difference between the two:

in, represents the atmospheric light curtain, D(x, y) represents the dark image, B ₁ (x, y) represents the local mean of the dark image D(x, y), and B ₂ (x, y) represents the dark image D(x, y) ), and F _B (x,y) represents the algorithm function using bilateral filtering.

because is the difference between the local mean and the local standard deviation of D(x,y), but:

Step 5: According to the known global atmospheric light value A, calculate the optimized projection map expression (6)

In the above formula, t(x,y) represents the transmittance matrix, Represents an atmospheric light curtain.

Step 6: Considering that when the value of the transmission map t is very small, the value of J will be too large, so that the overall image is excessive to the white point, so a threshold value t ₀ can generally be set. When the value of t is less than t ₀ , let t=t ₀ . So the final restored image expression is:

In the formula, J(x, y) represents the processed image, I(x) represents the original image, t(x, y) represents the transmittance matrix, A represents the global atmospheric light value, and t ₀ represents the threshold value of the selected transmittance map t .

A method for removing clouds from solar flare images based on an improved dark channel priority algorithm of the present invention has the following technical effects:

1: Use this method to de-cloud the solar flare image and simulate it. The cloud removal effect is evaluated according to the simulation results. The results show that the improved dark channel priority algorithm is effective in removing clouds and fog from solar flares.

2: The improved algorithm proposed by the present invention greatly shortens the calculation time and reduces the computational complexity. After comparing the effects of image processing, the improved algorithm obtains a clearer image, and improves the blockiness in the estimated transmittance map; the algorithm in this paper refines the transmittance and also plays a role in smoothing the edge of the image. Effect.

3: Compared with the original algorithm, the improved algorithm proposed by the present invention performs better in the average grayscale of the image, the display of detail information, the amount of picture information and the relative clarity, and it is easier to observe the position and image of the flare.

Description of drawings

Below in conjunction with accompanying drawing and embodiment, the present invention will be further described:

FIG. 1 is a flowchart of a method for removing cloud from a solar flare image based on an improved dark channel priority algorithm proposed in an embodiment of the present invention.

Figure 2 is an image of a solar flare before cloud removal.

Figure 3(1) is a solar flare image processed using the original dark channel priority algorithm.

Figure 3(2) is the solar flare image processed by the improved dark channel priority algorithm.

Detailed ways

The dark channel prior algorithm is first applied to the dehazing process of the image. Since the imaging model of the cloud is similar to the imaging model of the fog, it is the sum of the energy of the attenuation of the radiation information of the target and the attenuation of the atmospheric light, so it can be used. Dark channel prior knowledge, the algorithm is used to de-cloud the image.

Based on an improved dark channel prior algorithm. Since the original algorithm uses the soft-matting algorithm when solving the transmittance t(x, y), the computational complexity and computation time are greatly increased. The purpose is to ensure the quality of image processing at the same time.

A method for removing clouds from solar flare images based on an improved dark channel priority algorithm, comprising the following steps:

Step 1: The concept of channel is described by mathematical expression. For image J, the dark channel J _dark can be expressed as the formula:

In the above formula, J _c represents the color channel image of image J, Ω(x) represents a field range, and its pixel center is x, y is any point in the field range of Ω(x), and c represents r, g, b three channel, then J _c (y) represents the image of channel c at y in the domain of Ω(x). Its meaning is to find the minimum value of the three components of RGB, and then perform minimum value filtering on the single-channel image. The dark channel prior points out: J _dark → 0, which is equivalent to J _dark ≈ 0, so the real J _c can be obtained according to this condition.

Step 2: Suppose the cloud picture expresses the formula (2):

I(x)=J(x)t(x)+A(1-t(x)) (2)

Among them, I(x) represents the image to be processed, J(x) represents the real image, t(x) represents the transmittance, which represents the part of the light that can reach the computer system without being scattered, and A represents the global atmospheric light value.

Step 3: Obtain the global atmospheric light value A. Take the first 0.1% of the pixels from the dark channel map according to the size of the brightness. In these positions, the value of the corresponding point with the highest brightness in the original foggy image is found as the A value.

A bilateral filtering improvement is used in solving t(x).

Step 4: Calculate the local mean and local standard deviation of the dark image D(x,y), and then estimate the atmospheric light curtain through the difference between the two:

in, represents the atmospheric light curtain, D(x, y) represents the dark image, B ₁ (x, y) represents the local mean of the dark image D(x, y), and B ₂ (x, y) represents the dark image D(x, y) ), and F _B (x,y) represents the algorithm function using bilateral filtering.

because is the difference between the local mean and the local standard deviation of D(x,y), but:

Step 5: According to the known global atmospheric light value A, calculate the optimized projection map expression (6)

In the above formula, t(x,y) represents the transmittance matrix, Represents an atmospheric light curtain.

Step 6: Considering that when the value of the transmission map t is very small, the value of J will be too large, so that the overall image is excessive to the white point, so a threshold value t ₀ can generally be set. When the value of t is less than t ₀ , let t=t ₀ . So the final restored image expression is:

In the formula, J(x, y) represents the processed image, I(x) represents the original image, t(x, y) represents the transmittance matrix, A represents the global atmospheric light value, and t ₀ represents the threshold value of the selected transmittance map t .

Figure 1 is the flow chart of the method for removing cloud from solar flare images based on the improved dark channel priority algorithm; this method processes the original solar flare images, and uses bilateral filtering algorithm to compare the soft-matting algorithm and guided filtering when calculating the transmittance. Substitutions are made to optimize the calculation of transmittance.

Now randomly select a solar flare image, as shown in Figure 2, and perform cloud removal processing on it. Using MATLAB programming, the algorithm implementation process is shown in Figure 1. First, input the original image and solve its dark channel J _dark (x); then press The algorithm described in the paper obtains the value of the global atmospheric light and then estimates the atmospheric light curtain by calculating the local mean and local standard deviation of the dark image D(x,y). Then, according to the known air light vector, the projection map is calculated to obtain the optimized transmittance t(x). Finally, the distribution function of the cloud-free image is obtained according to formula (7), and the final cloud-free image is obtained by restoring it, as shown in Figure 3(2). At the same time, in order to compare with the algorithm before the improvement, the selected solar flare image is processed by the original dark channel priority algorithm, and the cloud-free image is obtained, as shown in Figure 3(1).

Combined with the characteristics of gray value, the gray value of the flare is higher, so the display is brighter, and the gray value of cloud and fog is also high, showing irregular bright colors. Therefore, from the comparison of Figure 2 with Figure 3(1) and Figure 3(2), it can be seen that no matter before or after the improvement, the cloud removal effect of the dark channel priority algorithm is very significant, but from Figure 3(1) and Figure 3(2) It can be seen from the comparison of the improved algorithm that the picture clarity is higher and the solar flare is more obvious, which reflects the superiority of the improved algorithm.

Claims (1)

1. A solar flare image cloud removing method based on an improved dark channel priority algorithm is characterized by comprising the following steps:

step 1: describing the concept of the channel by mathematical expressions, for image J, dark channel J_darkCan be expressed as the formula:

in the above formula, J_cA color channel image representing the image J,omega (x) represents a domain range, the pixel center is x, y is any point in the omega (x) domain range, c represents three channels of r, g and b, and J represents_c(y) represents the image of the c channel at y in the range of Ω (x) domain; the significance of the method is to calculate the minimum value of three components of RGB, and then carry out minimum value filtering on the single-channel image; the dark channel prior indicates: j. the design is a square_dark→ 0, corresponding to J_darkApproximately equals 0, so that the real J can be obtained according to the condition_c；

Step 2: assuming that the cloud picture expresses formula (2):

I(x)＝J(x)t(x)+A(1-t(x)) (2)

wherein, i (x) represents a graph to be processed, j (x) represents a real graph, t (x) represents a transmittance, represents a part of light which can reach a computer system and is not scattered, and a represents a global atmospheric light value;

and step 3: obtaining a global atmospheric light value A; taking the first 0.1% of pixels according to the brightness from the dark channel map; in these positions, the value of the corresponding point with the highest brightness is found in the original foggy image as the a value;

bilateral filtering improvement is adopted when solving t (x);

and 4, step 4: calculating the local mean and the local standard deviation of the dark image D (x, y), and estimating the atmosphere light curtain according to the difference between the two:

wherein,representing an atmospheric light curtain, D (x, y) representing a dark image, B₁(x, y) denotes the local mean of the dark image D (x, y), B₂(x, y) denotes the local standard deviation, F, of the dark image D (x, y)_B(x, y) represents an algorithm function that applies bilateral filtering;

due to the fact thatIs the difference between the local mean and the local standard deviation of D (x, y),then:

and 5: calculating an optimized projection diagram expression (6) according to the known global atmospheric light value A

In the above formula, t (x, y) represents a transmittance matrix,representing an atmospheric light curtain;

step 6: considering that when the value of the transmission map t is small, it will result in a large value of J, thus making the image overall transition to white, a GUO value t can be generally set₀When the value of t is less than t₀When t is equal to t₀(ii) a The final restored image expression is therefore:

wherein J (x, y) represents the processed image, I (x) represents the original image, t (x, y) represents the transmittance matrix, A represents the global atmospheric light value, t (x, y) represents the transmittance matrix, and₀the threshold value of the selected transmission map t is expressed.