对超声图像噪声进行抑制的滤波算法的研究起始于上世纪七十年代初。1971

The research on the filtering algorithm to suppress the noise of the ultrasonic image started in the early 1970s. In 1971, Turkey proposed a median filter algorithm, which uses the median pixel value of the neighborhood of the filter window to replace the central pixel value. The advantage of this algorithm is that the algorithm is simple and easy to implement, and the disadvantage is that the fixed filter window is easy to cause weak denoising ability or Loss of image details. Therefore, people have proposed enhanced Lee filtering and Frost filtering. In 1990, Perona and Malik proposed the famous PM diffusion equation, which for the first time introduced solving partial differential equations and thermal diffusion theory into image filtering. In addition, the wavelet filtering algorithm based on multi-resolution analysis and wavelet transform is also a more effective noise suppression algorithm. After years of development, a variety of filtering algorithms have been applied to ultrasound image processing. <br>The principle of the spatial domain filtering algorithm is mainly to use various image filtering window templates to smooth the image or adjust the pixel value according to the local statistical characteristics of the image to achieve the purpose of suppressing noise. There are many such filtering algorithms, and the typical ones are: Median filtering. <br>Transform domain filtering algorithms can be further divided into two categories: filtering algorithms based on frequency domain transform and filtering algorithms based on wavelet domain transform. Frequency domain transform filtering algorithms are mainly determined by Fourier transform of the image of the transform window and are determined interactively Different frequency ranges of noise and uncontaminated images, then select the appropriate frequency-domain band-pass filter for filtering processing, filter out the frequency-domain noise, and then obtain the denoised image after inverse Fourier transform. <br>The principle, derivation and characteristics of the above filtering algorithm will be introduced in detail in the second chapter of this article

The study of the filtering algorithm for the suppression of ultrasonic image noise began in the early 1970s. In 1971, Turkey proposed a median filtering algorithm, which uses the pixel median of the filter window neighborhood instead of the central pixel value, the advantage of which is that the algorithm is simple and easy to implement, the disadvantage is that the filter window is fixed and easy to cause weak noise loss or loss of image detail. As a result, enhanced Lee filtering and Frost filtering have been proposed. In 1990, Perona and Malik proposed the famous P-M diffusion equation, introducing the solution of the partial differential equation and the thermal diffusion theory into image filtering for the first time. In addition, the small wave filter algorithm based on multi-resolution analysis and wavelet transformation is also a more effective noise suppression algorithm. After years of development, there are many filtering algorithms applied to ultrasonic image processing.<br>The principle of spatial area filtering algorithm is to use various image filter window templates to smooth the image, or adjust the pixel value according to the statistical characteristics of the image bureau, in order to suppress the noise. There are many such filtering algorithms, of which typical are Median filtering.<br>The transformation domain filtering algorithm can be further divided into two categories: the filter algorithm based on frequency domain transformation and the filtering algorithm based on the wavelet domain transformation, the frequency domain transformation filter algorithm mainly determines the different frequency range of noise and unpolluted image by interactive means after the image of the transform window, and then selects the appropriate frequency domain band pass filter for filtering, filtering and removing the frequency domain noise, and then changing the image after the reverse transformation of the image.<br>The principles, derivations and characteristics of the above filtering algorithm sit in detail in the second chapter of this paper.

The research of filtering algorithm for ultrasonic image noise suppression started in the early 1970s. In 1971, Turkey proposed a median filtering algorithm, which uses the median value of the pixels in the neighborhood of the filtering window to replace the central pixel value. The advantage of this algorithm is that the algorithm is simple and easy to implement, and the disadvantage is that after the filtering window is fixed, it is easy to cause weak denoising ability or loss of image details. Therefore, enhanced Lee filter and frost filter are proposed. In 1990, Perona and Malik put forward the famous P-M diffusion equation. For the first time, partial differential equation and thermal diffusion theory were introduced into image filtering. In addition, the wavelet filtering algorithm based on multi-resolution analysis and wavelet transform is also an effective noise suppression algorithm. After years of development, many filtering algorithms have been applied to ultrasonic image processing.<br>The principle of spatial filtering algorithm is to smooth the image with various image filtering window templates, or adjust the pixel value according to the local statistical characteristics of the image, so as to achieve the purpose of noise suppression. There are many filtering algorithms, among which the typical ones are: median filtering.<br>Transform domain filtering algorithms can be further divided into two categories: filtering algorithm based on frequency domain transform and filtering algorithm based on wavelet domain transform. Frequency domain transform filtering algorithm mainly determines the different frequency range of noise and unpolluted image by means of interaction after Fourier transform on the image of transform window, and then selects appropriate frequency domain band-pass filter for filtering After filtering out the noise in frequency domain, the denoised image is obtained by inverse Fourier transform.<br>The principle, derivation and characteristics of the above filtering algorithm will be introduced in detail in the second chapter of this paper