WO2018099303A1 - 一种图像去噪的方法、装置及计算机存储介质 - Google Patents

一种图像去噪的方法、装置及计算机存储介质 Download PDF

Info

Publication number
WO2018099303A1
WO2018099303A1 PCT/CN2017/112308 CN2017112308W WO2018099303A1 WO 2018099303 A1 WO2018099303 A1 WO 2018099303A1 CN 2017112308 W CN2017112308 W CN 2017112308W WO 2018099303 A1 WO2018099303 A1 WO 2018099303A1
Authority
WO
WIPO (PCT)
Prior art keywords
template
dwf
filtering
pixel point
pixel
Prior art date
Application number
PCT/CN2017/112308
Other languages
English (en)
French (fr)
Inventor
滕涛
Original Assignee
深圳市中兴微电子技术有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳市中兴微电子技术有限公司 filed Critical 深圳市中兴微电子技术有限公司
Publication of WO2018099303A1 publication Critical patent/WO2018099303A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/70Denoising; Smoothing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/20Image enhancement or restoration using local operators
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20024Filtering details
    • G06T2207/20028Bilateral filtering

Definitions

  • the present invention relates to image processing technologies, and in particular, to a method and apparatus for image denoising, and a computer storage medium.
  • Gaussian noise In the process of video image acquisition and transmission, a variety of noises are often introduced. The most common one is Gaussian noise. The presence of Gaussian noise makes the video image flicker and affects the viewer's visual experience. In order to solve this problem, it is necessary to denoise the noise introduced during the acquisition or transmission process, so that the video after the final denoising is cleaned during viewing, thereby improving the viewer's visual experience and video quality.
  • the common low-pass filter is to filter the entire picture without distinguishing the edges, details and noise of the object. Secondly, even if part of the low-pass filter divides the image, it can distinguish the edge and noise of the object, but such The distinguishing method usually obtains the result of the difference is either noise or edge or detail, that is to say, the result of the distinction is not 0 or 1, which is too arbitrary and is easy to be misdetected by noise.
  • an embodiment of the present invention is directed to a method, an apparatus, and a computer storage medium for image denoising, which can preserve edge details of an image while removing noise, and also remove direction of direction interpolation and amplification. Noise makes the edges of the video image smoother.
  • an embodiment of the present invention provides a method for image denoising, the method comprising:
  • the pixel points of the image to be processed are traversed according to a preset direction template set, and the current direction and the current direction reliability corresponding to each pixel point are obtained;
  • DWF filtering is performed according to the current direction corresponding to each pixel point and a preset Direction-based Weighted Filter (DWF) template, and the DWF filtering result corresponding to each pixel point is obtained;
  • DWF Direction-based Weighted Filter
  • the denoised image is obtained according to the DWF filtering result of each pixel and the BF filtering result being mixed and output according to the reliability of the respective current directions.
  • the direction template set includes more than one direction template; each direction template has a corresponding direction angle, and each direction template has a corresponding index number (index); an index of each direction template The number corresponds to the direction angle of each direction template itself.
  • the pixel points of the image to be processed are traversed according to the preset direction template set, and the current direction and the current direction reliability of each pixel point are obtained, which specifically includes:
  • the direction reliability corresponding to the direction template with the smallest angle energy is set as the reliability of the current direction corresponding to the pixel point.
  • the DWF filtering is performed according to the current direction corresponding to each pixel point and the preset DWF template, and the DWF filtering result corresponding to each pixel point is obtained, which specifically includes:
  • Filtering is performed according to a preset filtering strategy and a DWF filtering template corresponding to each pixel, and a DWF filtering result corresponding to each pixel is obtained.
  • the DWF filtering result and the BF filtering result according to each pixel point are mixed and output according to the credibility of the respective current directions, and the denoised image is obtained, which specifically includes:
  • Alpha is the reliability of the current direction of each pixel
  • DWF is the DWF filtering result corresponding to each pixel
  • BF is the BF filtering result corresponding to each pixel.
  • an embodiment of the present invention provides an image denoising device, which includes: a TDCD (Texture Direction and Confidence Detection) module, a DWF module, a BF module, and a hybrid (Mixing). Module; among them,
  • the TDCD module is configured to traverse the pixel points of the image to be processed according to a preset direction template set, and obtain the current direction and the credibility of the current direction corresponding to each pixel point;
  • the DWF module is configured to perform DWF filtering according to a current direction corresponding to each pixel point and a preset DWF template, to obtain a DWF filtering result corresponding to each pixel point;
  • the BF module is configured to filter the pixel points of the image to be processed according to a preset BF template to obtain a BF filtering result corresponding to each pixel point;
  • the Mixing module is configured to perform mixed output according to the DWF filtering result of each pixel point and the BF filtering result according to the reliability of each current direction, to obtain a denoised image.
  • the direction template set includes more than one direction template; each direction template has a corresponding direction angle, and each direction template has a corresponding index number index; each direction template index number and each The direction angles of the direction templates themselves correspond.
  • the TDCD module is specifically configured as:
  • the direction reliability corresponding to the direction template with the smallest angle energy is set as the reliability of the current direction corresponding to the pixel point.
  • the DWF module is configured to acquire a corresponding DWF filtering template from a preset filtering template set according to a current direction corresponding to each pixel point;
  • the filtering process is performed to obtain a DWF filtering result corresponding to each pixel.
  • the Mixing module is configured to:
  • Alpha is the reliability of the current direction of each pixel
  • DWF is the DWF filtering result corresponding to each pixel
  • BF is the BF filtering result corresponding to each pixel.
  • an embodiment of the present invention further provides a computer storage medium, where the computer storage medium stores computer executable instructions, where the computer executable instructions are used to perform image denoising according to an embodiment of the present invention. method.
  • Embodiments of the present invention provide a method, a device, and a computer storage medium for image denoising, performing DWF on an image edge, and performing BF on other regions, and finally outputting DWF and BF results according to direction reliability, thereby not only capable of Effectively removes the Gaussian noise existing in the video image, preserves the edge details of the video image, and removes the direction noise caused by the direction interpolation, making the edge of the video image smoother and cleaner.
  • FIG. 1 is a schematic flowchart of a method for image denoising according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of a direction template set according to an embodiment of the present invention.
  • FIG. 3 is a schematic flowchart of obtaining credibility of a current direction and a current direction according to an embodiment of the present disclosure
  • FIG. 4 is a schematic diagram of an index table corresponding to a direction template set according to an embodiment of the present disclosure
  • FIG. 5 is a schematic flowchart of obtaining a DWF filtering result according to an embodiment of the present invention
  • FIG. 6 is a schematic diagram of a DWF filtering template set according to an embodiment of the present disclosure.
  • FIG. 7 is a schematic diagram of an index table corresponding to a DWF filtering template set according to an embodiment of the present disclosure
  • FIG. 8 is a schematic diagram of a BF template according to an embodiment of the present disclosure.
  • FIG. 9 is a schematic structural diagram of an apparatus for image denoising according to an embodiment of the present disclosure.
  • FIG. 10 is a schematic diagram of a specific implementation of a hybrid module according to an embodiment of the present invention.
  • the basic idea of the embodiment of the present invention is to propose a denoising algorithm based on direction detection and direction credibility estimation, which can detect the edge of an image and perform Direction-based Weighted Filter (DWF) on the edge of the image.
  • DWF Direction-based Weighted Filter
  • Bilateral filtering BF, Bilateral Filter
  • BF Bilateral Filter
  • the result output of DWF and BF is mixed according to the direction reliability, so that not only the Gaussian noise existing in the video image can be effectively removed, but also the edge details of the video image are preserved. It also removes the directional noise caused by the direction interpolation and makes the edges of the video image smoother and cleaner.
  • the method may be applied to a device and a device that need to perform denoising processing on an image.
  • the method may include:
  • S101 traverse the pixel points of the image to be processed according to a preset set of direction templates, and obtain a credibility of a current direction and a current direction corresponding to each pixel point;
  • S102 Perform DWF filtering according to a current direction corresponding to each pixel point and a preset direction-based weight filtering DWF template, to obtain a DWF filtering result corresponding to each pixel point;
  • S103 Filter the pixel points of the image to be processed according to a preset bilateral filtering BF template, and obtain a BF filtering result corresponding to each pixel point;
  • the direction template is The set includes more than one direction template; each direction template has a corresponding direction angle, and each direction template has a corresponding index number index; therefore, the index number of each direction template and the direction angle of each direction template itself Corresponding; as shown in Figure 2, the direction template set includes 16 direction templates.
  • the index number of the direction template is from 1 to 16; corresponding to 16 direction angles.
  • the black square indicates the current processing.
  • Pixels, blank squares indicate the neighboring pixel points corresponding to the currently processed pixel points; the direction angle corresponding to each direction template is set to the leftmost pixel point square of the row of the current black pixel point square and the leftmost pixel point of the previous row The angle between the line between the squares and the horizontal direction.
  • the direction template with the index of 2 in FIG. 2 has a corresponding direction angle of 11.25 degrees.
  • step S101 referring to FIG. 3, the pixel points of the image to be processed are traversed according to a preset direction template set, and the current direction and the current direction corresponding to each pixel point are obtained.
  • the credibility can specifically include:
  • S1012 Determine, according to respective angular energy of each direction template, a corresponding direction reliability of each direction template;
  • S1015 Set a direction credibility corresponding to the direction template with the minimum angle energy to a credibility of a current direction corresponding to the pixel point.
  • the index table corresponding to the direction template set may be generated according to the template in the direction template set shown in FIG. LUT_dir_pattern[16][2][16], the index table is specifically shown in FIG. 4, the first dimension of the index table represents the number of direction templates, and the second dimension of the index table represents the coordinate index of the template. Divided into two arrays of abscissa and ordinate, the third dimension of the index table indicates the number of pixels in each template, the maximum is 16.
  • the specific implementation process of the example shown in FIG. 3 may include:
  • direction reliability epsilon[n] corresponding to each direction template can also be obtained by other arithmetic formulas, and is only used to enumerate a specific obtained scheme.
  • the angular energy energy[n] of each direction template is corrected according to the angular energy energy[n] of each direction template and the direction reliability epsilon[n] corresponding to each direction template, and the specific implementation process can be implemented.
  • One of the formula (3) or formula (4) is preferred:
  • the direction template corresponding to the minimum value is selected as the current direction from the angle energy energy[[] of all direction templates, and the index number of the direction template is obtained as Dir, and the reliability corresponding to the direction template is taken as The credibility of the current direction, Alpha.
  • formula (5) and formula (6) are shown in formula (5) and formula (6):
  • Index() represents the value of n corresponding to the minimum energy.
  • Energy_thr is the default configuration parameter.
  • the preferred default value is 150.
  • step S101 is the process of performing TDCD on the image to be processed.
  • step S102 DWF filtering is performed according to the current direction corresponding to each pixel point and the preset direction-based weight filtering DWF template, and the DWF filtering result corresponding to each pixel point is obtained, which specifically includes:
  • S1021 Acquire a corresponding DWF filtering template from a preset filtering template set according to a current direction corresponding to each pixel point.
  • S1022 Perform filtering processing according to a preset filtering strategy and a DWF filtering template corresponding to each pixel point, and obtain a DWF filtering result corresponding to each pixel point.
  • the filtering template in the filtering template set may correspond to the direction template in the direction template set shown in FIG. 2, and the filtering template in the filtering template set is as shown in FIG. 6.
  • the angle value represented by each filter template in the filter template set and the corresponding relationship of each direction template in the direction template set are the direction 0° corresponding direction template 1, and so on, 168.75° corresponds to the direction template 16.
  • the number 0 represents the current processing point
  • the rest represents the neighborhood pixel
  • the digital size represents the distance weight at the time of subsequent filtering. The larger the number, the smaller the weight.
  • the coordinate offset index direct_mask[16][2][9] of all pixels in the filtered template is shown in FIG. Therefore, after obtaining the current direction corresponding to each pixel point in step S101, the corresponding filtering template may be acquired from the filtering template set according to the current direction; after obtaining the filtering template, each pixel point may be performed according to formula (7).
  • DWF filtering processing :
  • dis_LUT and dif_weight are calculated as follows:
  • the default configuration parameter of noise_thr the preferred default value is 20.
  • the pixel points of the image to be processed are filtered according to the preset bilateral filtering BF template, and the BF filtering result corresponding to each pixel point is obtained, which may include:
  • the preset bilateral filtering BF template is determined.
  • the BF template is as shown in FIG. 8 , where (0, 0) represents the current pixel point, and the rest represents the offset coordinate with respect to the current pixel point, where One represents the row offset coordinates and the second represents the column offset coordinates.
  • each pixel point is filtered according to equation (8):
  • dis_LUT_BF and dis_weight_bf are calculated as follows:
  • dis_LUT_BF[3][3] ⁇ 39,56,39 ⁇ , ⁇ 56,64,56 ⁇ , ⁇ 39,56,39 ⁇
  • the DWF filtering result and the BF filtering result of each pixel point are mixed and output according to the reliability of the respective current directions to obtain the denoised image, which may specifically include:
  • Alpha is the reliability of the current direction of each pixel
  • DWF is the DWF filtering result corresponding to each pixel
  • BF is the BF filtering result corresponding to each pixel.
  • the embodiment provides a method for image denoising, which can detect the edge of the image, and performs direction-based weighted filtering (DWF) on the edge of the image, and performs bilateral filtering on other regions (BF, Bilateral). Filter), finally mixing the result output of DWF and BF according to the direction credibility, so as not only can effectively remove the Gaussian noise existing in the video image, preserve the edge details of the video image, and also remove the direction noise caused by the direction interpolation amplification. Make the edges of the video image smoother and cleaner.
  • DWF direction-based weighted filtering
  • an apparatus 90 for image denoising may be provided, which may include: a TDCD module 901, a DWF module 902, a BF module 903, and a hybrid (Mixing). ) module 904; wherein
  • the TDCD module 901 is configured to: preset pixel points of an image to be processed according to a preset direction
  • the template set is traversed to obtain a current direction Dir corresponding to each pixel point and a reliability Alpha of the current direction;
  • the DWF module 902 is configured to perform DWF filtering according to a current direction corresponding to each pixel point and a preset DWF template, to obtain a DWF filtering result corresponding to each pixel point;
  • the BF module 903 is configured to filter the pixel points of the image to be processed according to a preset bilateral filtering BF template to obtain a BF filtering result corresponding to each pixel point;
  • the Mixing module 904 is configured to perform mixed output according to the DWF filtering result of each pixel point and the BF filtering result according to the reliability of each current direction, to obtain a denoised image.
  • the direction template set includes more than one direction template; each direction template has a corresponding direction angle, and each direction template has a corresponding index number index; each direction template has an index number and each The direction of the direction template itself corresponds to the angle of the direction.
  • the TDCD module 901 is specifically configured to:
  • the direction reliability corresponding to the direction template with the smallest angle energy is set as the reliability of the current direction corresponding to the pixel point.
  • the DWF module 902 is configured to acquire a corresponding DWF filtering template from a preset filter template set according to a current direction corresponding to each pixel point;
  • the filtering process is performed to obtain a DWF filtering result corresponding to each pixel.
  • the Mixing module 904 is configured to:
  • Alpha is the reliability of the current direction of each pixel
  • DWF is the DWF filtering result corresponding to each pixel
  • BF is the BF filtering result corresponding to each pixel.
  • the DWF filtering result and the BF filtering result of each pixel point may be respectively according to two by a multiplier and an adder.
  • the confidence of the current direction is mixed output.
  • the TDCD module 901, the DWF module 902, the BF module 903, and the Mixing module 904 may all be configured by a central processing unit (CPU) and a microprocessor located in the terminal.
  • CPU central processing unit
  • MPU Microprocessor Unit
  • DSP Digital Signal Processor
  • FPGA Field-Programmable Gate Array
  • the embodiment further provides a storage medium storing a computer program capable of implementing any one or more of the foregoing image denoising methods after being executed by the processor.
  • the computer storage medium may be various types of storage media, and may be preferably a non-transitory storage medium in this embodiment.
  • embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention can take the form of a hardware embodiment, a software embodiment, or a combination of software and hardware. Moreover, the present invention may employ computer-usable storage media (including but not limited to disks) in one or more of the computer-usable program code embodied therein. A form of computer program product embodied on a memory and optical storage, etc.).
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
  • the technical solution of the embodiment of the present invention performs direction-based weight filtering on the image edge, and performs bilateral filtering on other regions, and finally mixes the result output of the DWF and the BF according to the direction credibility, thereby not only effectively removing the Gauss existing in the video image. Noise, retaining the edge details of the video image, while also removing the directional noise caused by the direction interpolation, making the video image edge more Add smooth and clean.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)

Abstract

本发明实施例公开了一种图像去噪的方法和装置;该方法可以包括:将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度;按照所述每个像素点对应的当前方向以及预设的基于方向的权重滤波DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;将所述待处理图像的像素点按照预设的双边滤波BF模板进行滤波,获得每个像素点对应的BF滤波结果;按照所述每个像素点的DWF滤波结果和所述BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。本发明实施例还公开了一种计算机存储介质。

Description

一种图像去噪的方法、装置及计算机存储介质
相关申请的交叉引用
本申请基于申请号为201611073172.7、申请日为2016年11月29日的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此引入本申请作为参考。
技术领域
本发明涉及图像处理技术,尤其涉及一种图像去噪的方法、装置及计算机存储介质。
背景技术
在视频图像采集和传输过程中,经常会引入各种各样的噪声,其中最常见的就是高斯噪声,高斯噪声的存在使得视频图像变得闪烁,影响观看者的视觉感受。针对这种问题,需要对采集或传输过程中所引入的噪声进行去噪,让最终去噪之后的视频在观看时画面更干净,从而提升观看者的视觉感受和视频质量。
在视频图像处理领域,图像缩放是一种常用的关键技术,近年来兴起的基于方向的视频图像放大算法成为缩放领域的主要技术,然而基于方向的插值放大算法由于方向检测的不准确,经常会沿着物体边缘引入脉冲噪声,在视觉上给用户的感受就是物体的边缘有毛刺或者噪点存在,极大的影响了视频图像的感官质量。针对这种问题,需要对物体边缘存在的噪声进行去噪,让物体边缘变得更平滑干净
目前去除噪声的常用方法是采用低通滤波的方法,但是常用的低通滤波器存在如下问题:
首先,常见的低通滤波器是针对整个画面进行滤波,没有区分物体的边缘、细节和噪声;其次,即使部分低通滤波器对图像进行区域划分,能区分物体的边缘和噪声,但是这样的区分方式通常得到的区分结果要么是噪声,要么是边缘或者细节,也就是说区分的结果非0即1,这种区分方式过于武断,很容易噪声误检。
发明内容
为解决上述技术问题,本发明实施例期望提供一种图像去噪的方法、装置及计算机存储介质,能够在去除噪声的情况下保留图像的边缘细节,同时还能去除方向插值放大带来的方向噪声,使得视频图像边缘更加平滑。
为达到上述目的,本发明实施例的技术方案是这样实现的:
第一方面,本发明实施例提供了一种图像去噪的方法,所述方法包括:
将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度;
按照所述每个像素点对应的当前方向以及预设的基于方向的权重滤波(DWF,Direction-based Weighted Filter)模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;
将所述待处理图像的像素点按照预设的双边滤波(BF,Bilateral Filter)模板进行滤波,获得每个像素点对应的BF滤波结果;
按照所述每个像素点的DWF滤波结果和所述BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。
在一实施例中,所述方向模板集包括一个以上的方向模板;每个方向模板均有对应的方向角度,并且每个方向模板均有对应的索引号(index);每个方向模板的索引号与每个方向模板自身的方向角度对应。
在一实施例中,将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度,具体包括:
针对每个像素点,获取所述方向模板集中的每个方向模板各自对应的角度能量;
根据所述每个方向模板各自对应的角度能量确定所述每个方向模板各自对应的方向可信度;
获取所述角度能量最小的方向模板对应的模板索引号;
将所述角度能量最小的方向模板对应的角度方向设置为所述像素点对应的当前方向;
将所述角度能量最小的方向模板对应的方向可信度设置为所述像素点对应的当前方向的可信度。
在一实施例中,所述按照每个像素点对应的当前方向以及预设的DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果,具体包括:
根据所述每个像素点对应的当前方向从预设的滤波模板集内获取对应的DWF滤波模板;
按照预设的滤波策略以及所述每个像素点对应的DWF滤波模板进行滤波处理,获取每个像素点对应的DWF滤波结果。
在一实施例中,所述按照每个像素点的DWF滤波结果和BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像,具体包括:
按照下式进行混合输出,得到混合输出的结果output:
output=Alpha×DWF+(1-Alpha)×BF
其中,Alpha为所述每个像素点当前方向的可信度,DWF为所述每个像素点对应的DWF滤波结果,BF为所述每个像素点对应的BF滤波结果。
第二方面,本发明实施例提供了一种图像去噪装置,所述装置包括:方向及方向可信度检测(TDCD,Texture Direction and Confidence Detection)模块、DWF模块、BF模块和混合(Mixing)模块;其中,
所述TDCD模块,配置为将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度;
所述DWF模块,配置为按照所述每个像素点对应的当前方向以及预设的DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;
所述BF模块,配置为将所述待处理图像的像素点按照预设的BF模板进行滤波,获得每个像素点对应的BF滤波结果;
所述Mixing模块,配置为按照所述每个像素点的DWF滤波结果和所述BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。
在上述方案中,所述方向模板集包括一个以上的方向模板;每个方向模板均有对应的方向角度,并且每个方向模板均有对应的索引号index;每个方向模板的索引号与每个方向模板自身的方向角度对应。
在一实施例中,所述TDCD模块,具体配置为:
针对每个像素点,获取所述方向模板集中的每个方向模板各自对应的角度能量;以及,
根据所述每个方向模板各自对应的角度能量确定所述每个方向模板各自对应的方向可信度;以及,
获取所述角度能量最小的方向模板对应的模板索引号;以及,
将所述角度能量最小的方向模板对应的角度方向设置为所述像素点对应的当前方向;以及,
将所述角度能量最小的方向模板对应的方向可信度设置为所述像素点对应的当前方向的可信度。
在一实施例中,所述DWF模块,配置为根据所述每个像素点对应的当前方向从预设的滤波模板集内获取对应的DWF滤波模板;以及,
按照预设的滤波策略以及所述每个像素点对应的DWF滤波模板进行 滤波处理,获取每个像素点对应的DWF滤波结果。
在一实施例中,所述Mixing模块,配置为:
按照下式进行混合输出,得到混合输出的结果output:
output=Alpha×DWF+(1-Alpha)×BF
其中,Alpha为所述每个像素点当前方向的可信度,DWF为所述每个像素点对应的DWF滤波结果,BF为所述每个像素点对应的BF滤波结果。
第三方面,本发明实施例还提供了一种计算机存储介质,所述计算机存储介质中存储有计算机可执行指令,所述计算机可执行指令用于执行本发明实施例所述的图像去噪的方法。
本发明实施例提供了一种图像去噪的方法、装置及计算机存储介质,对图像边缘进行DWF,而对其他区域进行BF,最终根据方向可信度混合DWF和BF的结果输出,从而不仅能够有效去除视频图像中存在的高斯噪声,保留视频图像的边缘细节,同时还能去除方向插值放大带来的方向噪声,使得视频图像边缘更加平滑干净。
附图说明
图1为本发明实施例提供的一种图像去噪的方法流程示意图;
图2为本发明实施例提供的一种方向模板集的示意图;
图3为本发明实施例提供的一种获取当前方向及当前方向的可信度的流程示意图;
图4为本发明实施例提供的一种方向模板集对应的索引表示意图;
图5为本发明实施例提供的一种获取DWF滤波结果的流程示意图;
图6为本发明实施例提供的一种DWF滤波模板集的示意图;
图7为本发明实施例提供的一种DWF滤波模板集对应的索引表示意图;
图8为本发明实施例提供的一种BF模板示意图;
图9为本发明实施例提供的一种图像去噪的装置结构示意图;
图10为本发明实施例提供的一种混合模块的具体实现示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述。
本发明实施例的基本思想是提出一种基于方向检测及方向可信度估计的去噪算法,能够检测图像的边缘,并对图像边缘进行基于方向的权重滤波(DWF,Direction-based Weighted Filter),而对其他区域进行双边滤波(BF,Bilateral Filter),最终根据方向可信度混合DWF和BF的结果输出,从而不仅能够有效去除视频图像中存在的高斯噪声,保留视频图像的边缘细节,同时还能去除方向插值放大带来的方向噪声,使得视频图像边缘更加平滑干净。
基于上述基本思想,提出以下实施例。
实施例一
参见图1,其示出了本发明实施例提供的一种图像去噪的方法,该方法可以应用于需要对图像进行去噪处理的装置及设备,该方法可以包括:
S101:将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度;
S102:按照每个像素点对应的当前方向以及预设的基于方向的权重滤波DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;
S103:将待处理图像的像素点按照预设的双边滤波BF模板进行滤波,获得每个像素点对应的BF滤波结果;
S104:按照每个像素点的DWF滤波结果和BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。
对于图1所示的技术方案,需要说明的是,在步骤S101中,方向模板 集包括一个以上的方向模板;每个方向模板均有对应的方向角度,并且每个方向模板均有对应的索引号index;因此,每个方向模板的索引号与每个方向模板自身的方向角度对应;如图2所示,方向模板集中包括了16个方向模板,方向模板的索引号index从1至16;分别对应16个方向角度,在每个方向模板中,黑色方块表示当前进行处理的像素点,空白方块表示当前进行处理的像素点所对应的邻域像素点;每个方向模板对应的方向角度设置为当前黑色像素点方块所在行的最左边像素点方块与上一行最左边像素点方块之间的直线与水平方向的夹角。如图2中索引为2的方向模板,其对应的方向角度为11.25度。
基于上述针对步骤S101的说明,示例性地,对于步骤S101,参见图3,将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度,具体可以包括:
S1011:针对每个像素点,获取方向模板集中的每个方向模板各自对应的角度能量;
S1012:根据每个方向模板各自对应的角度能量确定每个方向模板各自对应的方向可信度;
S1013:获取所述角度能量最小的方向模板对应的模板索引号;
S1014:将所述角度能量最小的方向模板对应的角度方向设置为所述像素点对应的当前方向;
S1015:将所述角度能量最小的方向模板对应的方向可信度设置为所述像素点对应的当前方向的可信度。
需要说明的是,对于图3所述的示例,结合图2所示的方向模板集,在具体实现过程中,可以根据图2所示的方向模板集中的模板生成该方向模板集对应的索引表LUT_dir_pattern[16][2][16],该索引表具体如图4所示,索引表的第一维表示方向模板个数,索引表的第二维表示模板的坐标索引, 分为横坐标和纵坐标两个数组,索引表的第三维表示每个模板里面的像素个数,最大为16。在获得方向模板集对应的索引表LUT_dir_pattern[16][2][16]之后,图3所示的示例的具体实现过程,可以包括:
首先,按照式(1)计算每个方向模板的角度能量energy[n]:
Figure PCTCN2017112308-appb-000001
需要说明的是,n为从0开始的方向模板的索引号,因此,n表示的索引号与index所表示的索引号之间具有如下关系:n=index-1;
其中,number[16]={12,16,12,12,12,12,12,16,12,16,12,12,12,12,12,16},表示每个模板内像素个数,这样可以减少计算量。pi,j表示当前像素点,而i表示当前像素点位于图像中的列数,j表示当前像素点位于图像中的行数。可以理解地,每个模板的角度能量energy[n],可以通过其他运算式获得,这里只是用于列举一种具体获得的方案。
其次,按照方向模板的角度能量以及式(2)获取每个方向模板对应的方向可信度epsilon[n]:
Figure PCTCN2017112308-appb-000002
可以理解地,每个方向模板对应的方向可信度epsilon[n]也可以通过其他运算式获得,这里只是用于列举一种具体获得的方案。
在此,根据每个方向模板的角度能量energy[n]以及每个方向模板对应的方向可信度epsilon[n]对每个方向模板的角度能量energy[n]进行修正,具体的实现过程可以从式(3)或式(4)中优选一个:
energy′(n)=energy(n)×(1-epsilon[n])2  (3)
energy′(n)=energy(n)×(1-epsilon[n])  (4)
最后,从所有方向模板修正后的角度能量energy′[n]中选取最小值对应的方向模板作为当前方向,并获取该方向模板的索引号为Dir,并把该方向模板对应的可信度作为当前方向的可信度Alpha。具体如式(5)和式(6)所示:
Figure PCTCN2017112308-appb-000003
Figure PCTCN2017112308-appb-000004
其中,Index()表示取最小energy对应的n值。energy_thr为预设的配置参数,优选的默认值为150。
对于上述具体实现过程,为一种图像纹理方向及方向可信度检测(TDCD,Texture Direction and Confidence Detection)过程,因此,步骤S101所实现的也就是对待处理图像进行TDCD的过程。
示例性地,参见图5,对于步骤S102,按照每个像素点对应的当前方向以及预设的基于方向的权重滤波DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果,具体包括:
S1021:根据每个像素点对应的当前方向从预设的滤波模板集内获取对应的DWF滤波模板;
S1022:按照预设的滤波策略以及每个像素点对应的DWF滤波模板进行滤波处理,获取每个像素点对应的DWF滤波结果。
需要说明的是,对于图5所述的示例,在具体实现过程中,滤波模板集内的滤波模板可以对应于图2所示的方向模板集中的方向模板,滤波模板集中的滤波模板如图6所示,其中,滤波模板集内每个滤波模板所表示的角度值和方向模板集中每个方向模板的对应关系是方向0°对应方向模板1,以此类推,168.75°对应方向模板16。在滤波模板中,数字0表示当前处理点,其余表示邻域像素,数字大小表示在后续滤波时候的距离权重, 数字越大,权重越小。滤波模板内所有像素的坐标偏移索引direct_mask[16][2][9]如图7所表示。因此,当通过步骤S101获取得到每个像素点对应的当前方向之后,可以根据当前方向从滤波模板集中获取对应的滤波模板;在获得滤波模板之后,可以根据式(7)对每个像素点进行DWF滤波处理:
Figure PCTCN2017112308-appb-000005
其中,pi,j表示当前像素点,dis_LUT和dif_weight计算方式如下:
dis_LUT[9]={9,21,39,56,64,56,39,21,9}
Figure PCTCN2017112308-appb-000006
其中,noise_thr预设的配置参数,优选的默认值为20。
可以理解地,上述具体实现过程为一种DWF滤波过程。
示例性地,在具体实现过程中,对于步骤S103,将待处理图像的像素点按照预设的双边滤波BF模板进行滤波,获得每个像素点对应的BF滤波结果,可以包括:
首先确定预设的双边滤波BF模板,在本实施例中,BF模板如图8所示,其中,(0,0)表示当前像素点,其余表示相对当前像素点的偏移坐标,其中,第一个表示行偏移坐标,第二个表示列偏移坐标。
接着,按照式(8)对每个像素点进行滤波处理:
Figure PCTCN2017112308-appb-000007
其中,p0,0表示当前像素点,dis_LUT_BF和dis_weight_bf的计算方式如 下:
dis_LUT_BF[3][3]={{39,56,39},{56,64,56},{39,56,39}}
Figure PCTCN2017112308-appb-000008
示例性地,对于步骤S104,按照每个像素点的DWF滤波结果和BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像,具体可以包括:
按照下式进行混合输出,得到混合输出的结果output:
output=Alpha×DWF+(1-Alpha)×BF
其中,Alpha为每个像素点当前方向的可信度,DWF为每个像素点对应的DWF滤波结果,BF为每个像素点对应的BF滤波结果。
需要说明的是,每个像素点通过上述过程进行处理完毕后,就能够得到最终去噪完成后的图像。
本实施例提供了一种图像去噪的方法,能够检测图像的边缘,并对图像边缘进行基于方向的权重滤波(DWF,Direction-based Weighted Filter),而对其他区域进行双边滤波(BF,Bilateral Filter),最终根据方向可信度混合DWF和BF的结果输出,从而不仅能够有效去除视频图像中存在的高斯噪声,保留视频图像的边缘细节,同时还能去除方向插值放大带来的方向噪声,使得视频图像边缘更加平滑干净。
实施例二
基于前述实施例相同的技术构思,参见图9,其示出了本发明实施例提供的一种图像去噪的装置90,可以包括:TDCD模块901、DWF模块902、BF模块903和混合(Mixing)模块904;其中,
所述TDCD模块901,配置为将待处理图像的像素点按照预设的方向 模板集进行遍历,获取每个像素点对应的当前方向Dir及当前方向的可信度Alpha;
所述DWF模块902,配置为按照所述每个像素点对应的当前方向以及预设的DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;
所述BF模块903,配置为将所述待处理图像的像素点按照预设的双边滤波BF模板进行滤波,获得每个像素点对应的BF滤波结果;
所述Mixing模块904,配置为按照所述每个像素点的DWF滤波结果和所述BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。
示例性地,所述方向模板集包括一个以上的方向模板;每个方向模板均有对应的方向角度,并且每个方向模板均有对应的索引号index;每个方向模板的索引号与每个方向模板自身的方向角度对应。
示例性地,所述TDCD模块901,具体配置为:
针对每个像素点,获取所述方向模板集中的每个方向模板各自对应的角度能量;以及,
根据所述每个方向模板各自对应的角度能量确定所述每个方向模板各自对应的方向可信度;以及,
获取所述角度能量最小的方向模板对应的模板索引号;以及,
将所述角度能量最小的方向模板对应的角度方向设置为所述像素点对应的当前方向;以及,
将所述角度能量最小的方向模板对应的方向可信度设置为所述像素点对应的当前方向的可信度。
示例性地,所述DWF模块902,配置为根据所述每个像素点对应的当前方向从预设的滤波模板集内获取对应的DWF滤波模板;以及,
按照预设的滤波策略以及所述每个像素点对应的DWF滤波模板进行 滤波处理,获取每个像素点对应的DWF滤波结果。
示例性地,所述Mixing模块904,配置为:
按照下式进行混合输出,得到混合输出的结果output:
output=Alpha×DWF+(1-Alpha)×BF
其中,Alpha为所述每个像素点当前方向的可信度,DWF为所述每个像素点对应的DWF滤波结果,BF为所述每个像素点对应的BF滤波结果。
作为一种实施方式,对于Mixing模块904,其实现过程如图10所示,可以通过两个乘法器和一个加法器对所述每个像素点的DWF滤波结果和所述BF滤波结果按照各自的当前方向的可信度进行混合输出。
在实际应用中,所述TDCD模块901、所述DWF模块902、所述BF模块903和所述Mixing模块904,均可由位于终端中的中央处理器(CPU,Central Processing Unit)、微处理器(MPU,Microprocessor Unit)、数字信号处理器(DSP,Digital Signal Processor)、或现场可编程门阵列(FPGA,Field-Programmable Gate Array)等实现。
本实施例还提供一种存储介质,所述存储介质存储有计算机程序,所述计算机程序被处理器执行之后,能够实现前述任意一项或多项图像去噪的方法。
所述计算机存储介质可为各种类型的存储介质,在本实施例中可优选为非瞬间存储介质。
本领域技术人员应当理解,本实施例的存储介质中各程序的功能,可参照实施例所述的图像去噪的方法的相关描述而理解。
本领域内的技术人员应明白,本发明的实施例可提供为方法、系统、或计算机程序产品。因此,本发明可采用硬件实施例、软件实施例、或结合软件和硬件方面的实施例的形式。而且,本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘 存储器和光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器,使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中,使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品,该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上,使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理,从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
以上所述,仅为本发明的较佳实施例而已,并非用于限定本发明的保护范围。
工业实用性
本发明实施例的技术方案对图像边缘进行基于方向的权重滤波,而对其他区域进行双边滤波,最终根据方向可信度混合DWF和BF的结果输出,从而不仅能够有效去除视频图像中存在的高斯噪声,保留视频图像的边缘细节,同时还能去除方向插值放大带来的方向噪声,使得视频图像边缘更 加平滑干净。

Claims (11)

  1. 一种图像去噪的方法,所述方法包括:
    将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度;
    按照所述每个像素点对应的当前方向以及预设的基于方向的权重滤波DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;
    将所述待处理图像的像素点按照预设的双边滤波BF模板进行滤波,获得每个像素点对应的BF滤波结果;
    按照所述每个像素点的DWF滤波结果和所述BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。
  2. 根据权利要求1所述的方法,其中,所述方向模板集包括一个以上的方向模板;每个方向模板均有对应的方向角度,并且每个方向模板均有对应的索引号index;每个方向模板的索引号与每个方向模板自身的方向角度对应。
  3. 根据权利要求1所述的方法,其中,将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度,具体包括:
    针对每个像素点,获取所述方向模板集中的每个方向模板各自对应的角度能量;
    根据所述每个方向模板各自对应的角度能量确定所述每个方向模板各自对应的方向可信度;
    获取所述角度能量最小的方向模板对应的模板索引号;
    将所述角度能量最小的方向模板对应的角度方向设置为所述像素点对应的当前方向;
    将所述角度能量最小的方向模板对应的方向可信度设置为所述像素点 对应的当前方向的可信度。
  4. 根据权利要求1所述的方法,其中,所述按照每个像素点对应的当前方向以及预设的DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果,具体包括:
    根据所述每个像素点对应的当前方向从预设的滤波模板集内获取对应的DWF滤波模板;
    按照预设的滤波策略以及所述每个像素点对应的DWF滤波模板进行滤波处理,获取每个像素点对应的DWF滤波结果。
  5. 根据权利要求1所述的方法,其中,所述按照每个像素点的DWF滤波结果和BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像,具体包括:
    按照下式进行混合输出,得到混合输出的结果output:
    output=Alpha×DWF+(1-Alpha)×BF
    其中,Alpha为所述每个像素点当前方向的可信度,DWF为所述每个像素点对应的DWF滤波结果,BF为所述每个像素点对应的BF滤波结果。
  6. 一种图像去噪装置,所述装置包括:方向及方向可信度检测TDCD模块、基于方向的权重滤波DWF模块、双边滤波BF模块和混合Mixing模块;其中,
    所述TDCD模块,配置为将待处理图像的像素点按照预设的方向模板集进行遍历,获取每个像素点对应的当前方向及当前方向的可信度;
    所述DWF模块,配置为按照所述每个像素点对应的当前方向以及预设的DWF模板进行DWF滤波,获得每个像素点对应的DWF滤波结果;
    所述BF模块,配置为将所述待处理图像的像素点按照预设的BF模板进行滤波,获得每个像素点对应的BF滤波结果;
    所述Mixing模块,配置为按照所述每个像素点的DWF滤波结果和所 述BF滤波结果按照各自的当前方向的可信度进行混合输出,获得去噪后的图像。
  7. 根据权利要求6所述的装置,其中,所述方向模板集包括一个以上的方向模板;每个方向模板均有对应的方向角度,并且每个方向模板均有对应的索引号index;每个方向模板的索引号与每个方向模板自身的方向角度对应。
  8. 根据权利要求6所述的装置,其中,所述TDCD模块,具体配置为:
    针对每个像素点,获取所述方向模板集中的每个方向模板各自对应的角度能量;以及,
    根据所述每个方向模板各自对应的角度能量确定所述每个方向模板各自对应的方向可信度;以及,
    获取所述角度能量最小的方向模板对应的模板索引号;以及,
    将所述角度能量最小的方向模板对应的角度方向设置为所述像素点对应的当前方向;以及,
    将所述角度能量最小的方向模板对应的方向可信度设置为所述像素点对应的当前方向的可信度。
  9. 根据权利要求6所述的装置,其中,所述DWF模块,配置为根据所述每个像素点对应的当前方向从预设的滤波模板集内获取对应的DWF滤波模板;以及,
    按照预设的滤波策略以及所述每个像素点对应的DWF滤波模板进行滤波处理,获取每个像素点对应的DWF滤波结果。
  10. 根据权利要求6所述的装置,其中,所述Mixing模块,配置为:
    按照下式进行混合输出,得到混合输出的结果output:
    output=Alpha×DWF+(1-Alpha)×BF
    其中,Alpha为所述每个像素点当前方向的可信度,DWF为所述每个像 素点对应的DWF滤波结果,BF为所述每个像素点对应的BF滤波结果。
  11. 一种计算机存储介质,所述计算机存储介质中存储有计算机可执行指令,所述计算机可执行指令用于执行权利要求1至5任一项所述的图像去噪的方法。
PCT/CN2017/112308 2016-11-29 2017-11-22 一种图像去噪的方法、装置及计算机存储介质 WO2018099303A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611073172.7 2016-11-29
CN201611073172.7A CN108122204A (zh) 2016-11-29 2016-11-29 一种图像去噪的方法和装置

Publications (1)

Publication Number Publication Date
WO2018099303A1 true WO2018099303A1 (zh) 2018-06-07

Family

ID=62225690

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/112308 WO2018099303A1 (zh) 2016-11-29 2017-11-22 一种图像去噪的方法、装置及计算机存储介质

Country Status (2)

Country Link
CN (1) CN108122204A (zh)
WO (1) WO2018099303A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116977228B (zh) * 2023-09-25 2024-02-09 广东匠芯创科技有限公司 图像降噪方法、电子设备及存储介质
CN118096581A (zh) * 2024-04-23 2024-05-28 中国人民解放军空军军医大学 用于危急重症患者生命体征监测的图像处理方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110007973A1 (en) * 2009-07-10 2011-01-13 Samsung Electro-Mechanics Co., Ltd. Apparatus for removing image noise and method thereof
CN103345726A (zh) * 2013-06-14 2013-10-09 华为技术有限公司 图像去噪处理方法、装置和终端
CN105913396A (zh) * 2016-04-11 2016-08-31 湖南源信光电科技有限公司 一种噪声估计的图像边缘保持混合去噪方法
US20160253788A1 (en) * 2015-02-27 2016-09-01 Siliconfile Technologies Inc. Device for removing noise on image using cross-kernel type median filter and method therefor

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101248808B1 (ko) * 2011-06-03 2013-04-01 주식회사 동부하이텍 경계 영역의 잡음 제거 장치 및 방법
CN102222326B (zh) * 2011-06-28 2013-02-06 青岛海信信芯科技有限公司 一种基于单张低分辨率的图像去模糊方法及装置
US9208536B2 (en) * 2013-09-27 2015-12-08 Kofax, Inc. Systems and methods for three dimensional geometric reconstruction of captured image data
CN104915929A (zh) * 2014-03-14 2015-09-16 电子科技大学 一种图像滤波处理方法
CN104200442B (zh) * 2014-09-19 2017-11-21 西安电子科技大学 基于改进的canny边缘检测的非局部均值MRI图像去噪方法
CN106056555A (zh) * 2016-06-03 2016-10-26 宁波大红鹰学院 一种图像去噪方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110007973A1 (en) * 2009-07-10 2011-01-13 Samsung Electro-Mechanics Co., Ltd. Apparatus for removing image noise and method thereof
CN103345726A (zh) * 2013-06-14 2013-10-09 华为技术有限公司 图像去噪处理方法、装置和终端
US20160253788A1 (en) * 2015-02-27 2016-09-01 Siliconfile Technologies Inc. Device for removing noise on image using cross-kernel type median filter and method therefor
CN105913396A (zh) * 2016-04-11 2016-08-31 湖南源信光电科技有限公司 一种噪声估计的图像边缘保持混合去噪方法

Also Published As

Publication number Publication date
CN108122204A (zh) 2018-06-05

Similar Documents

Publication Publication Date Title
US10489892B2 (en) Method for signal processing
JP4727720B2 (ja) 画像処理方法および画像処理装置
US9959600B2 (en) Motion image compensation method and device, display device
US20150187051A1 (en) Method and apparatus for estimating image noise
CN109003233A (zh) 一种基于自适应权重全变分模型的图像去噪方法
WO2018099303A1 (zh) 一种图像去噪的方法、装置及计算机存储介质
KR20140109801A (ko) 3d이미지 품질을 향상시키는 방법과 장치
CN102118547A (zh) 图像加权滤波的方法
Zhang et al. An improved BM3D algorithm based on anisotropic diffusion equation
WO2017128646A1 (zh) 一种图像处理的方法及装置
CN112073718B (zh) 一种电视机花屏检测方法、装置、计算机设备及储存介质
CN106683129B (zh) 一种基于双边核回归的相对约减纹理分解方法及其装置
WO2020113824A1 (zh) 图像处理方法
Azetsu et al. Trilateral filter using rank order information of pixel value for mixed Gaussian and impulsive noise removal
JP2011112630A (ja) レーダ画像処理装置
US9336573B2 (en) Self-adaptive image edge correction device and method thereof
TW201510934A (zh) 影像銳化方法與影像處理裝置
JP6256680B2 (ja) 画像処理方法、画像処理装置、及び画像処理プログラム
JP4104475B2 (ja) 輪郭補正装置
JP2010171624A (ja) 輪郭補正回路及び補間画素生成回路
KR100905524B1 (ko) 확대된 영상의 흐려짐 선명화 장치 및 방법
CN116091360B (zh) 基于曲面拟合的非均匀性噪声消除方法和装置
WO2017036386A1 (zh) 一种视频去噪方法及装置、终端、存储介质
JP5855210B2 (ja) 情報処理装置、ライン状ノイズ低減処理方法、及びプログラム
CN117710250B (zh) 一种消除纤维镜成像蜂窝状结构的方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17877282

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17877282

Country of ref document: EP

Kind code of ref document: A1