CN-114972054-B - Line Gaussian filtering-based solar dark bar fine structure enhancement method

Abstract

The invention relates to a line Gaussian filtering-based sun dark bar fine structure enhancement method, and belongs to the field of image enhancement. In order to solve the problems that manual marking is difficult and fiber attribute information is not extracted accurately enough due to low contrast of solar dark stripe fibers in the current H-alpha image, the invention provides an image enhancement method taking linear Gaussian convolution as a core. The H-alpha image is normalized first to initially stretch the image contrast, edges are enhanced with a Laplace-Gaussian operator while noise is suppressed, and then the contrast of the solar dark stripe fibers is enhanced with a line Gaussian filter. The self-adaptive histogram equalization and top cap and bottom cap transformation methods which limit the contrast ratio are used in the post-processing stage to solve the problem of uneven enhancement effect of the linear Gaussian filter. And finally, further improving the contrast of the solar dark stripe fiber by using the linear Gaussian filter. The method is used for enhancing the fiber filament-like structure of the dark solar strip, so that the fiber filament-like structure of the dark solar strip can be clearly and visually seen, and objective and accurate measurement of attribute information of the fiber filament-like structure of the dark solar strip is possible.

Inventors

• SHANG ZHENHONG
• YANG KANG

Assignees

• 昆明理工大学

Dates

Publication Date

20260508

Application Date

20220105

Claims (3)

1. 1. A method for enhancing a fine structure of a solar dark bar based on linear Gaussian filtering is characterized by comprising the following steps: step1, preprocessing, namely inputting an H-alpha image, normalizing the H-alpha image, sharpening a Laplace-Gaussian image, and primarily highlighting a filiform structure and suppressing noise; Step2, enhancing the filiform structure image based on the linear Gaussian filter, namely rotating and convoluting each pixel in the image and the neighborhood of each pixel with the linear Gaussian filter, wherein the maximum value of the rotating and convoluting is the corresponding pixel value of the output image in the Step, and enhancing the contrast ratio of the filiform fiber structure; Step3, post-processing, namely adopting a self-adaptive histogram equalization method for limiting contrast and a method for combining top cap and low cap conversion to inhibit the intensity value of a background pixel from rising; Step4, enhancing the image by using the linear Gaussian filter again, and further improving the contrast of the dark stripe filiform fiber structure; The specific steps of Step1 are as follows: Step1.1, normalizing the input H-alpha image, normalizing and improving the contrast of dark areas and light areas in the image through a3 sigma principle: (1) Wherein, the Representing pixel intensities at (x, y) positions in the H-alpha image; , ; 、 、 And Respectively turning the normalized image to obtain an image I, namely turning light pixels into dark pixels; step1.2, sharpening the image I by using the laplace-gaussian method: (2) Wherein, in the right side of the equation { }, the image is sharpened by adopting a Laplace differential operator, the intensity constant area in the image is set to be 0, the intensity value of the constant area can be restored by overlapping with the image I, c is the sharpening intensity coefficient, Representing convolution operation, G 2 (sigma) represents a 0-mean and standard deviation sigma two-dimensional Gaussian filter; The specific steps of Step2 are as follows: Pixels in the sharpened image I LoG , which are distributed in different directions, are convolved with a line gaussian filter to improve contrast, that is, (3) Wherein, the Gaussian filter representing length and line The same width is single pixel, the center is positioned at (x, y) position in the image I LoG , the line segment with the included angle theta with the bottom edge of the image, Is the length of The width is single pixel, the standard deviation is For each pixel in the image I LoG , the line segment L is sampled by rotating the pixel as the center, and the sampling result of each time is combined with the linear Gaussian filter And (3) convolving, wherein the maximum value after the convolution of the rotation samples is the intensity value of the corresponding pixel of the enhanced image.
2. 2. The method for enhancing the fine structure of the solar dark bars based on the linear Gaussian filtering according to claim 1 is characterized in that: the specific steps of Step3 are as follows: step3.1, adopting a self-adaptive histogram equalization method for limiting contrast, improving the local contrast of each region of the image by calculating the accumulated distribution function of each pixel neighborhood, and avoiding the excessive amplification of noise of a local flat region; Step3.2, image output to Step3.1 using gray image morphological top cap and bottom cap transformations And (3) further processing: (4) Wherein, the And Respectively representing the pair of images And performing top cap and low cap conversion.
3. 3. The method for enhancing the fine structure of the solar dark stripe based on the linear Gaussian filtering of claim 1, wherein Step4 carries out linear Gaussian convolution on the image processed by Step3 by adopting a linear Gaussian filter which is the same as Step2, so that the image enhancement of the filiform fiber structure of the solar dark stripe is realized.

Description

Line Gaussian filtering-based solar dark bar fine structure enhancement method Technical Field The invention relates to a line Gaussian filtering-based sun dark bar fine structure enhancement method, and belongs to the field of image enhancement. Background Image enhancement is an important research topic of digital image processing, is widely applied to the fields of remote sensing, astronomical images and the like, and has the main functions of purposefully emphasizing the whole or partial characteristics of the image, enabling the original unclear image to become clear or emphasizing some interesting characteristics, expanding the difference between different object characteristics in the image, inhibiting the uninteresting characteristics, improving the image quality, enriching the information quantity, enhancing the image interpretation and recognition effect and meeting the needs of some special analysis. Dark stripe fibers are the fundamental elements that make up the dark stripe of the sun, and these fibers depict the fundamental structure of the dark stripe magnetic field. Dark stripe fibers have low contrast from the background and are often densely presented in bundles, making it difficult to see and track the evolution of the dark stripe fine filamentous fiber structure directly through the H-alpha high resolution image. The enhancement of the dark stripe filiform structure image is beneficial to improving the scientific research value of the observation data of the H-alpha spectral line sun dark stripe fine structure, and has important significance for further excavating the observation potential of the existing foundation sun telescope, helping sunphysicists to deeply study the currently pending problems such as the formation, evolution, explosion model and mechanism of the dark stripe fine magnetic field structure, and the like, and promoting the observation and research of solar activity. The invention provides an image enhancement algorithm based on linear Gaussian filtering, which can obviously enhance the fine structure of a sun dark stripe in an H-alpha image, so that dark stripe fibers are obviously distinguished from a background, and a learner is helped to observe the dark stripe fibers and extract dark stripe fiber information more objectively. Disclosure of Invention The invention aims to use an image enhancement method taking linear Gaussian filtering as a core, aiming at the characteristic of low contrast of the sun dark stripe fiber on an H-alpha image, the intensity value on the sun dark stripe fiber pixel is obviously enhanced, fiber structure information in the image is emphasized, the distinguishing degree of a sun dark stripe filiform structure and a background is enlarged, and the problems that in the H-alpha image at present, the manual marking of the dark stripe fiber is difficult and the acquisition of attribute information such as the position, the length and the direction of the dark stripe fiber is inaccurate are solved. The technical scheme includes that the method for enhancing the fine structure of the solar dark strip based on the linear Gaussian comprises the following steps: step1, preprocessing, namely inputting an H-alpha image, normalizing the H-alpha image, sharpening a Laplace-Gaussian image, and primarily highlighting a filiform structure and suppressing noise; Step1.1, normalize the input H-alpha image, the distribution range of pixel intensity values in the FITS file of H-alpha is larger (the maximum and minimum values differ by more than 30000), resulting in lower contrast between dark and background areas in the image. In general, the intensity histogram of an H-alpha image is approximately normally distributed, and the contrast ratio of dark areas such as dark bars, color sphere fibers and the like to light areas in the image can be normalized and improved by a 3 sigma principle: Wherein I hα (x, y) represents the minimum, maximum, average and standard deviation of the intensity of the H-alpha image, respectively, of the pixel intensities ;tlow＝max(Imin,Im-3σhα),thigh＝min(Imax,Im+3σhα);Imin、Imax、Im and sigma hα at the (x, y) position in the H-alpha image. Furthermore, for ease of observation and analysis, we flip the normalized image to image I, i.e., light pixels to dark pixels. Equation (1) substantially increases the dynamic range of dark and light regions in the H-alpha image and normalizes the intensity values to [0,1]. Step1.2, sharpening the normalized image I by a laplace-gaussian method in order to further highlight the filiform structure and suppress noise: The right side of the equation is formed by sharpening the image by using a Laplace differential operator, enabling an intensity constant region in the image to be 0, and recovering the intensity value of the constant region by superposition with I. c is the coefficient of the sharpening intensity and, The convolution operation is represented by G 2 (σ) which represents a two-dimensional