WO2012145989A1 - 级联滤波器及其标定去噪强度的动态设定方法 - Google Patents

级联滤波器及其标定去噪强度的动态设定方法 Download PDF

Info

Publication number
WO2012145989A1
WO2012145989A1 PCT/CN2011/080033 CN2011080033W WO2012145989A1 WO 2012145989 A1 WO2012145989 A1 WO 2012145989A1 CN 2011080033 W CN2011080033 W CN 2011080033W WO 2012145989 A1 WO2012145989 A1 WO 2012145989A1
Authority
WO
WIPO (PCT)
Prior art keywords
filter
denoising
strength
intensity
image
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2011/080033
Other languages
English (en)
French (fr)
Inventor
邓志辉
张荣祥
查林
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Hikvision Software Co Ltd
Original Assignee
Hangzhou Hikvision Software Co Ltd
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 Hangzhou Hikvision Software Co Ltd filed Critical Hangzhou Hikvision Software Co Ltd
Publication of WO2012145989A1 publication Critical patent/WO2012145989A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/14—Picture signal circuitry for video frequency region
    • H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence

Definitions

  • the present invention relates to the field of video processing, and in particular to a dynamic setting technique for denoising strength of a cascading filter calibration.
  • the actual acquired image often contains a large amount of noise.
  • the filter is usually used to filter the image to reduce noise.
  • a cascading filter is a system that cascades multiple filters.
  • the image is filtered in multiple filters in a cascaded filter, usually achieving better filtering than a single filter.
  • the filter strength of each filter in the cascaded filter is controlled by a parameter called 'calibration denoising strength'.
  • the denoising intensity is best matched with the actual noise level of the input image. It is better to use a larger calibration denoising intensity for images with higher actual noise levels, and it is better to use images with lower actual noise levels. Smaller calibration denoising strength.
  • the prior art usually Set a fixed calibration denoising intensity for each filter in the cascaded filter.
  • the multi-stage filter uses the same noise level, so that some areas of the denoising result will be too smooth, and some areas have insufficient denoising. The denoising effect is not good.
  • an embodiment of the present invention provides a method for dynamically setting a denoising strength of a cascading filter, the cascading filter including at least a first filter and a second filter cascaded in front and rear.
  • the method includes the following steps:
  • the calibrated denoising strength of the second filter is set based on the calibrated denoising strength of the first filter and the estimated first filter denoising capability.
  • An embodiment of the present invention further provides a cascading filter including at least a first filter and a second filter cascaded in front and rear, and further comprising:
  • An estimating unit configured to estimate a first filter denoising capability according to a denoising result of the first filter
  • a calibration unit configured to set a calibration denoising intensity of the second filter according to the denoised intensity of the first filter and the estimated first filter denoising capability.
  • the adaptive adjustment of the denoising intensity of the subsequent filter is implemented for different input image images, thereby equalizing the denoising strength of the cascaded filters to obtain a better overall denoising effect. If the denoising intensity of the previous stage filter is high for an image, the denoising intensity of the latter stage filter will be automatically weakened; if the demodulation intensity of the previous stage is lower for another image, the latter one will be lower.
  • the stage filter automatically increases the denoising strength so that the overall denoising effect of the cascaded filter is closer to the preset calibration value.
  • the difference between the calibration denoising strength of the previous stage filter and the denoising ability estimation value is combined with one or more of the video motion information, the historical video information, and the image local texture information to jointly calculate the latter one.
  • the denoising intensity of the stage filter can further improve the overall denoising effect of the cascaded filter.
  • the cascaded filter is set to a hybrid configuration of the time domain filter and the spatial domain filter, which can further improve the overall effect of the denoising. Some noise is more suitable for filtering in the time domain. Some noise is more suitable for filtering in the airspace. It can be used with a time domain filter and a spatial domain filter to have better denoising effect. On this basis, it is an innovation of this patent to use a dynamic method to set the denoising strength between different types of filters cascaded.
  • the technique of this patent is used.
  • the scheme can automatically set a higher calibration denoising intensity for the spatial filter of the latter stage, and increase the denoising intensity in the airspace, which can often achieve better results. If the previous stage estimates that the denoising ability is high, the lower calibration denoising strength is automatically set to the latter stage according to the technical solution of the patent to avoid excessive filtering.
  • time-space domain filter is costly or requires a large amount of resources, and a time-space domain filter is used with a time or spatial domain filter to achieve a better overall under the premise of lower cost or less resources. effect.
  • FIG. 1 is a schematic flow chart of a dynamic setting method for denoising intensity of a cascading filter according to a first embodiment of the present invention
  • FIG. 2 is a schematic diagram of a FIR operator in the first embodiment of the present invention.
  • FIG. 3 is a schematic flow chart of a dynamic setting method for denoising intensity of a cascading filter according to a second embodiment of the present invention
  • FIG. 4 is a schematic flow chart of a dynamic setting method for denoising intensity of a cascading filter according to a third embodiment of the present invention.
  • FIG. 5 is a schematic structural diagram of a cascade filter according to a fourth embodiment of the present invention.
  • a first embodiment of the present invention relates to a method for dynamically setting a denoising strength of a cascaded filter.
  • the cascading filter calibrates a dynamic setting method of denoising intensity.
  • the cascading filter includes at least a first filter and a second filter cascaded in front and rear.
  • the first is the first filter and the second is the second filter.
  • more filters may be included in the cascaded filter, such as three, four, five, etc., in this embodiment, only the denoising between any two adjacent filters is described.
  • Dynamic setting method of intensity Knowing the dynamic setting method between two adjacent filters can be easily analogized to each filter.
  • a filter A, B, C, and D form a cascade filter, and the dynamic setting method disclosed in the embodiment of the present invention is used between A and B, between B and C, and between C and D, respectively.
  • A is the first filter and B is the second filter.
  • B is the first filter
  • C is the second filter.
  • C is the first filter
  • D is the second filter.
  • the dynamic setting method in the embodiment of the present invention may be used between every two filters, or the dynamic setting method in the embodiment of the present invention may be used only between partial filters in the cascading filter.
  • FIG. 1 is a schematic flow chart of a method for dynamically setting the denoising strength of the cascaded filter.
  • step 101 the first filter denoising capability is estimated based on the denoising result of the first filter.
  • the denoising ability of the first filter can be estimated from the noise image input to the first filter and the denoised image output from the first filter.
  • a noise image is a true image in the true sense, or a local data block in the entire image.
  • a special algorithm can also be used to estimate the denoising ability of the first filter based only on the output image of the first filter.
  • the following is an example of estimating the denoising ability of two pre-stage filters (ie, the first filter).
  • the first method is called the SAD method
  • the second method is called FIR operator extraction method.
  • FIR refers to finite impulse response filter ( Finite Impulse Response ).
  • the operation of the pre-stage filter denoising ability estimation module is illustrated by taking the local data block of the image as an example. Assume that the image of the local block of the denoising of the previous stage filter is I1, and the block of the output image of the denoised image is I2.
  • the operator used in the FIR operator extraction method has a frequency domain response similar to the noise characteristic, and is not limited to the FIR provided by the embodiment of the present invention. operator.
  • the previous stage filter denoising ability estimation method is not limited thereto, and any method capable of estimating the filter denoising ability can be employed.
  • the first filter denoising capability is subtracted from the estimated denoising strength of the first filter to obtain a difference.
  • the 'calibration denoising intensity' referred to in various embodiments of the present invention may be a scalar indicating the intensity of denoising the entire image as a whole, or a matrix indicating the local denoising intensity of the image.
  • the calibration denoising strength of the first filter and the estimated difference of the first filter denoising ability are used as the estimated values of the denoising strength of the second filter.
  • the method of estimating the denoising intensity of the second filter is not limited thereto, and any method capable of estimating the denoising intensity of the second filter can be employed.
  • step 103 Setting the denoising intensity of the second filter according to the difference.
  • the difference may be directly used as the calibration denoising intensity of the second filter, or a specific operation may be performed according to the difference, and the operation result is used as the calibration denoising intensity of the second filter.
  • the steps 102 and 103 are based on the calibrated denoising strength of the first filter and step 101 The first filter denoising ability estimated in the middle, and a specific way of setting the denoising intensity of the second filter.
  • K Estimated first filter denoising capability / calibrated denoising strength of the first filter
  • calibrated denoising intensity of the second filter ( 1 - K ) * first filter denoising capability, and so on.
  • the calibration denoising strength of the second filter is jointly determined according to the difference value and the reference factor, wherein the reference factor is any one or any combination of the following factors:
  • Video motion information historical video information, and image local texture information.
  • the difference between the calibration denoising strength of the previous stage filter and its denoising ability estimate, combined with one or more of the video motion information, the historical video information, and the image local texture information, the latter filter is jointly calculated.
  • the denoising intensity of the calibration can further improve the overall denoising effect of the cascaded filter.
  • the cascaded filter includes at least one time domain filter and one spatial domain filter.
  • the cascaded filter is set to a hybrid configuration of the time domain filter and the spatial domain filter, which can further improve the overall effect of denoising. Some noise is more suitable for filtering in the time domain. Some noise is more suitable for filtering in the airspace. It can be used with a time domain filter and a spatial domain filter to have better denoising effect. On this basis, it is an innovation of this patent to use a dynamic method to set the denoising strength between different types of filters cascaded.
  • the technique of this patent is used.
  • the scheme can automatically set a higher calibration denoising intensity for the spatial filter of the latter stage, and increase the denoising intensity in the airspace, which can often achieve better results. If the previous stage estimates that the denoising ability is high, the lower calibration denoising strength is automatically set to the latter stage according to the technical solution of the patent to avoid excessive filtering. The same is true for the case where the previous stage is a spatial domain filter and the latter stage is a time domain filter.
  • the cascaded filter includes a time-space domain filter, the other being a time domain filter or a spatial domain filter.
  • time-space domain filter is costly or requires a lot of resources.
  • a time-space domain filter is used with a time or spatial domain filter to achieve a better overall effect under the premise of lower cost or less resources.
  • the cascaded filter can also include multiple spatiotemporal filters, even consisting of spatiotemporal filters.
  • the cascaded filters are all of the same type of filter, for example, both time domain filters, spatial domain filters, or both time domain filters, the technical solution of the present invention can also be used.
  • a second embodiment of the present invention relates to a dynamic setting method for denoising intensity of a cascading filter calibration.
  • image 3 It is a schematic flow chart of the dynamic setting method of the denoising intensity of the cascade filter.
  • the second embodiment is improved on the basis of the first embodiment, and the main improvement is that the video motion information is considered in calculating the calibration denoising strength of the second filter.
  • step 301 motion information m of the image partial block is calculated.
  • the motion calibration coefficient k is obtained from the motion calibration table T from the motion information m of the image local block.
  • the motion calibration table T is a table that is used to calibrate the relationship between motion size and denoising strength.
  • the first filter denoising ability L is estimated based on the denoising result of the first filter.
  • step 304 using the first filter's denoising intensity sigma1 minus the estimated first filter denoising ability L , get the difference (sigma1 - L).
  • the denoised intensity of the second filter is set according to the difference. Specifically, the motion calibration coefficient k and the difference The product of (sigma1 - L) is used as the calibration denoising strength sigma2 of the second filter.
  • this technical solution can be used to adaptively strengthen or weaken the denoising strength of the locality to prevent insufficient denoising in some areas. And some areas are too smooth, improving the overall denoising effect.
  • a third embodiment of the present invention relates to a dynamic setting method for denoising intensity of a cascading filter calibration.
  • Figure 4 It is a schematic flow chart of the dynamic setting method of the denoising intensity of the cascade filter.
  • the third embodiment also considers video motion information, and uses a statistical method based on the second embodiment to obtain a more accurate calibration denoising strength.
  • step 401 motion information of the image partial block is calculated.
  • step 402 The motion calibration coefficient is obtained from the motion calibration table by the motion information of the image local block, and the motion calibration table is used to calibrate the relationship between the motion magnitude and the denoising strength.
  • the first filter denoising capability is estimated based on the denoising result of the first filter.
  • step 404 The first filter denoising capability is subtracted from the estimated denoising strength of the first filter to obtain a difference.
  • step 405 the product of the motion calibration coefficient and the difference is used as the denoising intensity of the image partial block calibrated by the second filter.
  • step 406 The denoising intensity of all the calibrated image partial blocks of the second filter is counted, and the mean value thereof is used as the calibration denoising intensity of the second filter.
  • the mean value thereof is used as the calibration denoising intensity of the second filter.
  • other algorithms such as taking the median can calculate the denoised intensity of the second filter.
  • step 103 A sub-step of setting the denoising strength of the second filter based on the difference.
  • the method embodiments of the present invention can all be implemented in software, hardware, firmware, and the like. Regardless of whether the invention is implemented in software, hardware, or firmware, the instruction code can be stored in any type of computer-accessible memory (eg, permanent or modifiable, volatile or non-volatile, solid state Or non-solid, fixed or replaceable media, etc.).
  • the instruction code can be stored in any type of computer-accessible memory (eg, permanent or modifiable, volatile or non-volatile, solid state Or non-solid, fixed or replaceable media, etc.).
  • the memory can be, for example, programmable array logic ( Programmable Array Logic (referred to as 'PAL'), Random Access Memory ('RAM') Programmable Read Only Memory (PROM '), Read-Only Memory (Read-Only Memory) , referred to as 'ROM'), electrically erasable programmable read-only memory (Electrically Erasable Programmable ROM, referred to as 'EEPROM '), disk, CD, Digital Versatile Disc ('DVD').
  • programmable array logic Programmable Array Logic
  • 'RAM' Random Access Memory
  • PROM ' Programmable Read Only Memory
  • Read-Only Memory Read-Only Memory
  • 'EEPROM Electrically erasable programmable read-only memory
  • disk CD
  • DVD'DVD' Digital Versatile Disc
  • a fourth embodiment of the invention relates to a cascade filter.
  • Figure 5 It is a schematic diagram of the structure of the cascaded filter.
  • the cascaded filter includes at least a first filter and a second filter cascaded in front and rear, and further comprising: between the first and second filters:
  • an estimating unit configured to estimate a first filter denoising capability according to a denoising result of the first filter.
  • the calibration unit may further include:
  • Difference calculation subunit for using the first filter denoising strength of the first filter minus the first filter denoising capability of the estimation unit output to obtain a difference
  • the calibration intensity calculation and setting subunit is configured to set the calibration denoising intensity of the second filter according to the difference calculated by the difference calculation subunit.
  • the cascading filter can be a time domain filter and a spatial domain filter, or a time-space domain filter, and the other is a time domain filter or a spatial domain filter.
  • the technical solution comprises two basic units, one of which is an estimation unit (estimation unit) for denoising capability of the previous stage filter (first filter), and the second is denoising for the next stage filter (second filter) Strength calibration unit (calibration unit).
  • the calibrated denoising strength can be a scalar representing the intensity of the overall denoising of the entire image, or a matrix representing the local denoising strength of the image.
  • the technical solution does not depend on the specific implementation of the previous stage filter and the next stage filter (whether it is a time domain filter, a spatial domain filter or a time domain filter), and the input and output of each stage of the filter Can be any video format, for example RGB, YUV, Bayer Wait.
  • This technical solution can be used in video denoising to estimate the overall noise intensity or local noise intensity of the filter image in the cascaded filter bank in real time, and to facilitate the flexible design of the filter of the cascaded filter bank. It is beneficial for the filter to adjust the denoising strength according to the local noise intensity of the image, thereby equalizing the denoising of each level of the filter and obtaining a better overall denoising effect.
  • the estimating unit extracts the information of the image by using at least the noisy input image including the previous stage filter and the previous stage filter to denoise the image.
  • the purpose of this unit is to estimate the actual denoising ability of the previous stage filter reasonably and effectively, and provide the basis and reference for the calibration of the denoising intensity of the next stage filter; according to the way of noise transmission before and after the filter, the unit
  • the output is a scalar that represents the ability to denoise the entire image as a whole, or a matrix that represents the local denoising ability of the image.
  • the calibration unit utilizes the actual denoising ability of the previous stage filter estimated by the estimation unit and the calibration denoising intensity of the previous stage filter, and cooperates with the auxiliary optional video motion information, historical video information, and image local texture information.
  • One or more pieces of information calibrate the denoising strength required by the next stage filter as the denoising strength of the next stage filter.
  • the denoised intensity of the output of the unit can also be a scalar indicating the intensity of the overall denoising of the entire image, or a matrix representing the local denoising strength of the image.
  • the first filter and the second filter are cascaded to denoise, and a noise transmission module composed of an estimation unit and a calibration unit is used to improve the denoising capability of each filter.
  • the noise image input to the first filter and the denoised image outputted by the denoising of the first filter are used together as an input of the estimation unit, thereby calculating the ability of the first filter to denoise, which requires special explanation.
  • the noise image and the denoising image can be a whole image in the true sense or a partial data block in the image. And the image can be in any data format, for example RGB, YUV, Bayer, single-channel grayscale images, and more.
  • the first, second, and third embodiments are method embodiments corresponding to the present embodiment, and the present embodiment can be implemented in cooperation with the first, second, and third embodiments.
  • the related technical details mentioned in the first, second, and third embodiments are still effective in the present embodiment, and are not described herein again in order to reduce repetition. Accordingly, the related art details mentioned in the present embodiment can also be applied to the first, second, and third embodiments.
  • each unit mentioned in the embodiments of the present invention is a logical unit.
  • a logical unit may be a physical unit, a part of a physical unit, or multiple physical entities.
  • the combined implementation of the elements, the physical implementation of these logical units themselves is not the most important, the combination of the functions implemented by these logical units is the key to solving the technical problems raised by the present invention.
  • the above-mentioned various device embodiments of the present invention do not introduce a unit that is not closely related to solving the technical problem proposed by the present invention, which does not indicate that the above device implementation does not have other unit.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Picture Signal Circuits (AREA)
  • Image Processing (AREA)

Abstract

本发明涉及视频处理领域，公开了一种级联滤波器及其标定去噪强度的动态设定方法。本发明中，对前一级滤波器的去噪能力进行估计，用前一级滤波器的标定去噪强度减去估计结果得到差值，再根据该差值设定后一级滤波器的标定去噪强度，可以针对不同的输入图像实现后一级滤波器去噪强度的自适应调整，从而均衡级联的各级滤波器的去噪强度以获得更好的整体去噪效果。优选地，级联滤波器中既有时域滤波器也有空域滤波器。

Description

级联滤波器及其标定去噪强度的动态设定方法 技术领域
本发明涉及视频处理领域，特别涉及级联滤波器标定去噪强度的动态设定技术。
背景技术
在视频监控等具体的应用场景中，实际采集到的图像中往往含有大量的噪声，为了能方便图像的应用或后期的处理，通常采用滤波器对图像进行滤波以降低噪声。
滤波器有很多种类，如时域滤波器、空域滤波器、时空域滤波器等等。不过无论是哪一个种类，单个滤波器的滤波效果往往是有限的，因此有人使用级联滤波器对图像进行滤波。
级联滤波器是指将多个滤波器级联起来的一种系统。图像在级联滤波器中会先后在多个滤波器中进行滤波，通常可以达到比单个滤波器更好的滤波效果。
级联滤波器中每一个滤波器的滤波强度是由一个参数控制的，该参数被称为'标定去噪强度'。理论上标定去噪强度最好与所输入的图像的实际噪声水平相匹配，对实际噪声水平较高的图像最好使用较大的标定去噪强度，对实际噪声水平较低的图像最好使用较小的标定去噪强度。
但是，在实际应用场景中，图像的实际噪声水平往往无法得知，或者是在变化中的，而且噪声的类型也会在变化中，或者图像不同区域具有不同的噪声水平，所以现有技术通常为级联滤波器中的每一个滤波器设定一个固定的标定去噪强度，这样的话多级滤波器采用同一个噪声水平，使得去噪结果有些区域会过于平滑，有些区域去噪不足，整体去噪效果不佳。
 技术问题
本发明的目的在于提供一种级联滤波器及其标定去噪强度的动态设定方法，能够均衡各级滤波器的去噪强度以获得更好的整体去噪效果。
技术解决方案
为解决上述技术问题，本发明的实施方式提供了一种级联滤波器标定去噪强度的动态设定方法，该级联滤波器至少包括前后级联的第一滤波器和第二滤波器，方法包括以下步骤：
根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计；
根据第一滤波器的标定去噪强度和估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度。
本发明的实施方式还提供了一种级联滤波器，至少包括前后级联的第一滤波器和第二滤波器，还包括：
估计单元，用于根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计；
标定单元，用于根据第一滤波器的标定去噪强度和估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度。
 有益效果
本发明实施方式与现有技术相比，主要区别及其效果在于：
对前一级滤波器去噪能力进行估计，用前一级滤波器的标定去噪强度减去估计结果得到差值，再根据该差值设定后一级滤波器的标定去噪强度，可以针对不同的输入图像象实现后一级滤波器去噪强度的自适应调整，从而均衡级联的各级滤波器的去噪强度以获得更好的整体去噪效果。如果针对某个图像，前一级滤波器去噪强度较高，后一级滤波器的去噪强度就会自动减弱一些；如果针对另一个图像，前一级滤波去噪强度较低，后一级滤波器就自动加大去噪强度，从而使级联滤波器的整体去噪效果更为接近预先设定的标定值。
进一步地，将前一级滤波器的标定去噪强度与其去噪能力估计值的差值，结合视频运动信息、历史视频信息、图像局部纹理信息中的一项或多项，共同计算得到后一级滤波器的标定去噪强度，可以进一步提高级联滤波器的整体去噪效果。
进一步地，在动态调整标定去噪强度的基础上，将级联滤波器设置成时域滤波器和空域滤波器的混合配置，可以进一步提高去噪的整体效果。有一些噪声比较适合在时域过滤，有一些噪声比较适合在空域过滤，搭配使用时域滤波器和空域滤波器，可以有更好的去噪效果。在此基础上，在级联的不同类型的滤波器之间使用动态的方法设定标定去噪强度是本专利的一个创新。例如，对于前一级是时域滤波器，后一级是空域滤波器的情况，如果前一级估计下来去噪能力不高，说明在时域去噪效果不佳，则使用本专利的技术方案可以自动为后一级的空域滤波器设定较高的标定去噪强度，在空域加大去噪强度，往往可以达到较好的效果。如果前一级估计下来去噪能力较高，则根据本专利的技术方案对后一级自动设定较低的标定去噪强度，以免滤波过度。
进一步地，时空域滤波器成本较高或需要占用较多的资源，使用一个时空域滤波器搭配时间或空域滤波器，在较低成本或较少资源的前提下，也能达到较好的整体效果。
附图说明
图 1 是本发明第一实施方式中一种级联滤波器标定去噪强度的动态设定方法的流程示意图；
图 2 是本发明第一实施方式中一种 FIR 算子的示意图；
图 3 是本发明第二实施方式中一种级联滤波器标定去噪强度的动态设定方法的流程示意图；
图 4 是本发明第三实施方式中一种级联滤波器标定去噪强度的动态设定方法的流程示意图；
图 5 是本发明第四实施方式中一种级联滤波器的结构示意图。
 本发明的最佳实施方式
在以下的叙述中，为了使读者更好地理解本申请而提出了许多技术细节。但是，本领域的普通技术人员可以理解，即使没有这些技术细节和基于以下各实施方式的种种变化和修改，也可以实现本申请各权利要求所要求保护的技术方案。
为使本发明的目的、技术方案和优点更加清楚，下面将结合附图对本发明的实施方式作进一步地详细描述。
本发明第一实施方式涉及一种级联滤波器标定去噪强度的动态设定方法。该级联滤波器标定去噪强度的动态设定方法该级联滤波器至少包括前后级联的第一滤波器和第二滤波器。
为了更容易理解本发明的技术方案，本发明的各实施方式大都以两个滤波器进行说明，但实际上可以方便地扩展到更多个相互级联的滤波器，只要在每两个相邻滤波器之间都采用本发明的技术方案即可，两个滤波器可以是相邻的，也可以两者之间存在间隔。
如果级联滤波器中只有两个滤波器，则在前的是第一滤波器，在后的是第二滤波器。当然，级联滤波器中也可以包括更多的滤波器，如三个、四个、五个等等，此时本实施方式中只是描述其中任意两个相邻滤波器之间的标定去噪强度的动态设定方法。知道了两个相邻滤波器之间的动态设定方法，就可以方便地类推到各个滤波器。例如，滤波器 A 、 B 、 C 、 D 组成一个级联滤波器，则 A 和 B 之间， B 和 C 之间， C 和 D 之间分别采用本发明实施方式所公开的动态设定方法。在 A 和 B 之间， A 是第一滤波器， B 是第二滤波器。在 B 和 C 之间， B 是第一滤波器， C 是第二滤波器。在 C 和 D 之间， C 是第一滤波器， D 是第二滤波器。
可以是每两个滤波器之间都采用本发明实施方式中的动态设定方法，也可以只是级联滤波器中的部分滤波器之间采用本发明实施方式中的动态设定方法。
图 1 是该级联滤波器标定去噪强度的动态设定方法的流程示意图。
在步骤 101 中，根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计。
可以根据输入第一滤波器的噪声图像和第一滤波器输出的去噪图像估计第一滤波器的去噪能力。噪声图像是真正意义上的一整幅图像，或者整幅图像中的局部数据块。当然，在特定场景下，也可以采用特殊的算法仅根据第一滤波器的输出图像来估计第一滤波器的去噪能力。
下面例举两种前一级滤波器（即第一滤波器）去噪能力的估计方法，方法一称为作差求 SAD 法，方法二称为 FIR 算子提取法。 FIR 是指有限冲激响应滤波器（ Finite Impulse Response ）。并以图像的局部数据块为例说明前一级滤波器去噪能力估计模块的操作。假设前一级滤波器去噪的图像局部块数据为 I1 ，去噪后的输出图像局部块数据为 I2 ，若采用 SAD 法，前一级滤波器去噪能力表示为 L=||I1- I2||1/N ， N 为 I1 和 I2 局部数据块像素的个数；若采用 FIR 算子提取法，前一级滤波器去噪能力表示为 L=|F*(I1-I2)| ， F 为 FIR 算子， * 表示卷积，图 2 给出了一种 FIR 算子的例子。
FIR 算子提取法中采用的算子具有和噪声特性类似的频域响应即可，并不限于本发明实施方式提供的 FIR 算子。并且前一级滤波器去噪能力估计方法也不限于此，任何能够估计出滤波器去噪能力的方法都可以被采用。
此后进入步骤 102 ，使用第一滤波器的标定去噪强度减去估计得到的第一滤波器去噪能力，得到差值。本发明各实施方式中所称的'标定去噪强度'可以是一个表示对整幅图像整体去噪的强度的标量，也可以是一个表示图像局部去噪强度的矩阵。
本实施方式中，采用第一滤波器的标定去噪强度以及估计得到的第一滤波器去噪能力的差值作为第二滤波器的标定去噪强度的估计值。并且第二滤波器的标定去噪强度的估计方法也不限于此，任何能够估计出第二滤波器的标定去噪强度的方法都可以被采用。
此后进入步骤 103 ，根据差值设定第二滤波器的标定去噪强度。可以是直接将差值作为第二滤波器的标定去噪强度，也可以根据该差值进行特定的运算，将运算结果作为第二滤波器的标定去噪强度。
所述步骤 102 和 103 是 根据第一滤波器的标定去噪强度和步骤 101 中估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度的一种具体方式。除了这种实现方式之外，还可以有多种等同效果的实现方式，例如，计算 K= 估计得到的第一滤波器去噪能力 / 第一滤波器的标定去噪强度，第二滤波器的标定去噪强度＝（ 1 － K ） * 第一滤波器去噪能力，等等。
在本发明的一个优选实例中，根据差值和参考因素共同决定第二滤波器的标定去噪强度，其中参考因素是以下各项因素中的任意一项或任意组合：
视频运动信息、历史视频信息、图像局部纹理信息。
当然，也可以不用参考因素，只凭该前一级滤波器的标定去噪强度与其去噪能力估计值的差值设定后一级滤波器（第二滤波器）的标定去噪强度。
对前一级滤波器去噪能力进行估计，用前一级滤波器的标定去噪强度减去估计结果得到差值，再根据该差值设定后一级滤波器的标定去噪强度，可以针对不同的输入图像实现后一级滤波器去噪强度的自适应调整，从而均衡级联的各级滤波器的去噪强度以获得更好的整体去噪效果。如果针对某个图像，前一级滤波器去噪强度较高，后一级滤波器的去噪强度就会自动减弱一些。如果针对另一个图像，前一级滤波去噪强度较低，后一级滤波器就会自动加大去噪强度，从而使级联滤波器的整体去噪效果更为接近预先设定的标定值。
将前一级滤波器的标定去噪强度与其去噪能力估计值的差值，结合视频运动信息、历史视频信息、图像局部纹理信息中的一项或多项，共同计算得到后一级滤波器的标定去噪强度，可以进一步提高级联滤波器的整体去噪效果。
在本发明的一个优选例子中，级联滤波器中至少包括一个时域滤波器和一个空域滤波器。
在动态调整标定去噪强度的基础上，将级联滤波器设置成时域滤波器和空域滤波器的混合配置，可以进一步提高去噪的整体效果。有一些噪声比较适合在时域过滤，有一些噪声比较适合在空域过滤，搭配使用时域滤波器和空域滤波器，可以有更好的去噪效果。在此基础上，在级联的不同类型的滤波器之间使用动态的方法设定标定去噪强度是本专利的一个创新。例如，对于前一级是时域滤波器，后一级是空域滤波器的情况，如果前一级估计下来去噪能力不高，说明在时域去噪效果不佳，则使用本专利的技术方案可以自动为后一级的空域滤波器设定较高的标定去噪强度，在空域加大去噪强度，往往可以达到较好的效果。如果前一级估计下来去噪能力较高，则根据本专利的技术方案对后一级自动设定较低的标定去噪强度，以免滤波过度。对于前一级是空域滤波器，后一级是时域滤波器的情况分析也是类似。
本发明的另一个优选例子中，级联滤波器中包括一个时空域滤波器，其它为时域滤波器或空域滤波器。
时空域滤波器成本较高或需要占用较多的资源，使用一个时空域滤波器搭配时间或空域滤波器，在较低成本或较少资源的前提下，也能达到较好的整体效果。
此外，可以理解，级联滤波器中也可以包括多个时空域滤波器，甚至都由时空域滤波器组成。
当然，如果级联滤波器中都是同一种类型的滤波器，例如都是时域滤波器、都是空域滤波器、或者都是时空域滤波器，也可以使用本发明的技术方案。
本发明第二实施方式涉及一种级联滤波器标定去噪强度的动态设定方法。图 3 是该级联滤波器标定去噪强度的动态设定方法的流程示意图。
第二实施方式在第一实施方式的基础上进行了改进，主要改进之处在于：在计算第二滤波器的标定去噪强度时，考虑了视频运动信息。
在步骤 301 中，计算图像局部块的运动信息 m 。
此后进入步骤 302 ，由图像局部块的运动信息 m 从运动标定表 T 中查取运动标定系数 k 。运动标定表 T 是一张表，用于标定运动大小与去噪强度的关系。
此后进入步骤 303 ，根据第一滤波器的去噪结果对第一滤波器去噪能力 L 进行估计。
此后进入步骤 304 ，使用第一滤波器的标定去噪强度 sigma1 减去估计得到的第一滤波器去噪能力 L ，得到差值 (sigma1 - L) 。
此后进入步骤 305 ，根据差值设定第二滤波器的标定去噪强度。具体地说，将运动标定系数 k 与差值 (sigma1 - L) 的乘积作为第二滤波器的标定去噪强度 sigma2 。
因为是针对图像局部块进行标定去噪强度的动态设定，如果图像不同区域具有不同的噪声水平，利用本技术方案可以自适应地对局部加强或减弱标定去噪强度，防止有些区域去噪不足，而有些区域又过于平滑的问题，提高整体去噪效果。
本发明第三实施方式涉及一种级联滤波器标定去噪强度的动态设定方法。图 4 是该级联滤波器标定去噪强度的动态设定方法的流程示意图。
第三实施方式也考虑了视频运动信息，并在第二实施方式基础上采用统计的方法以得到更为准确的标定去噪强度。
在步骤 401 中，计算图像局部块的运动信息。
此后进入步骤 402 ，由图像局部块的运动信息从运动标定表中查取运动标定系数，该运动标定表用于标定运动大小与去噪强度的关系。
此后进入步骤 403 ，根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计。
此后进入步骤 404 ，使用第一滤波器的标定去噪强度减去估计得到的第一滤波器去噪能力，得到差值。
此后进入步骤 405 ，将运动标定系数与差值的乘积作为第二滤波器标定的图像局部块的去噪强度。
此后进入步骤 406 ，统计第二滤波器所有标定的图像局部块的去噪强度，采用其均值作为第二滤波器的标定去噪强度。当然，也可以不采用均值，而是其它的算法（如取中值）来计算第二滤波器的标定去噪强度。
上述步骤 405 和 406 相当于步骤 103 根据差值设定第二滤波器的标定去噪强度的子步骤。
本发明的各方法实施方式均可以以软件、硬件、固件等方式实现。不管本发明是以软件、硬件、还是固件方式实现，指令代码都可以存储在任何类型的计算机可访问的存储器中（例如永久的或者可修改的，易失性的或者非易失性的，固态的或者非固态的，固定的或者可更换的介质等等）。同样，存储器可以例如是可编程阵列逻辑（ Programmable Array Logic ，简称' PAL '） 、随机存取存储器（ Random Access Memory ，简称' RAM '） 、可编程只读存储器（ Programmable Read Only Memory ，简称' PROM '） 、只读存储器（ Read-Only Memory ，简称' ROM '） 、电可擦除可编程只读存储器（ Electrically Erasable Programmable ROM ，简称' EEPROM '） 、磁盘、光盘、数字通用光盘（ Digital Versatile Disc ，简称' DVD '）等等。
本发明第四实施方式涉及一种级联滤波器。图 5 是该级联滤波器的结构示意图。该级联滤波器至少包括前后级联的第一滤波器和第二滤波器，在第一和第二滤波器之间还包括：
估计单元，用于根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计。
标定单元， 用于根据第一滤波器的标定去噪强度和所述估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度。
在本发明的一个优选实例中，该标定单元可以进一步包括：
差值计算子单元：用于使用第一滤波器的标定去噪强度减去估计单元输出的第一滤波器去噪能力，得到差值，
标定强度计算和设置子单元，用于根据差值计算子单元计算得到的差值设定第二滤波器的标定去噪强度。
当然，标定单元还可以有其它的实现方式，例如，计算 K= 估计得到的第一滤波器去噪能力 / 第一滤波器的标定去噪强度，第二滤波器的标定去噪强度＝（ 1 － K ） * 第一滤波器去噪能力，等等。
级联滤波器中可以是一个时域滤波器和一个空域滤波器，也可以是一个时空域滤波器，其它为时域滤波器或空域滤波器。
本技术方案包括两个基本单元，其一为前一级滤波器（第一滤波器）去噪能力的估计单元（估计单元），其二为下一级滤波器（第二滤波器）去噪强度的标定单元（标定单元）。标定的去噪强度可以是一个表示对整幅图像整体去噪的强度的标量，也可以是一个表示图像局部去噪强度的矩阵。该技术方案不依赖于前一级滤波器和下一级滤波器的具体实现方式（无论是时域滤波器、空域滤波器还是时空域滤波器皆可），并且各级滤波器的输入和输出可以是任意视频格式，例如 RGB 、 YUV 、 Bayer 等。这种技术方案可以用在视频去噪中，实时地估计出级联滤波器组中的滤波器图像的整体噪声强度或者局部噪声强度，且有利于级联滤波器组的滤波器的灵活设计，有利于滤波器根据图像局部噪声强度调节去噪强弱，从而均衡各级滤波器的去噪并获得更好的整体去噪效果。
估计单元至少利用包含前一级滤波器的含噪声输入图像和前一级滤波器去噪后输出图像的信息。该单元的目的是合理有效地估计出前一级滤波器的实际去噪能力，并为下一级滤波器去噪强度的标定提供依据和参考；根据前后滤波器噪声传递的方式不同，此单元的输出是一个表示对整幅图像整体去噪的能力的标量，或者是一个表示图像局部去噪能力的矩阵。
标定单元利用估计单元估计的前一级滤波器的实际去噪能力以及前一级滤波器的标定去噪强度，并配合辅助的可选的视频运动信息、历史视频信息以及图像局部纹理信息等中的一项或几项信息，标定出下一级滤波器需要的去噪强度，以此做为下一级滤波器的去噪强度。同理，该单元输出的标定去噪强度也可以是一个表示对整幅图像整体去噪的强度的标量，或者是一个表示图像局部去噪强度的矩阵。
如图 4 所示，第一滤波器和第二滤波器级联去噪，其间采用由估计单元和标定单元组成的噪声传递模块提高各滤波器的去噪能力。在本实施例中，输入第一滤波器的噪声图像和第一滤波器的去噪后输出的去噪图像共同作为估计单元的输入，以此计算第一滤波器去噪的能力，需要特别说明的是噪声图像和去噪图像可以是真正意义上的一整幅图像，也可以是图像中的局部数据块。并且图像可以是任意数据格式，例如 RGB 、 YUV 、 Bayer 、单通道灰度图像等等。
第一、第二和第三实施方式是与本实施方式相对应的方法实施方式，本实施方式可与第一、第二和第三实施方式互相配合实施。第一、第二和第三实施方式中提到的相关技术细节在本实施方式中依然有效，为了减少重复，这里不再赘述。相应地，本实施方式中提到的相关技术细节也可应用在第一、第二和第三实施方式中。
需要说明的是，本发明各设备实施方式中提到的各单元都是逻辑单元，在物理上，一个逻辑单元可以是一个物理单元，也可以是一个物理单元的一部分，还可以以多个物理单元的组合实现，这些逻辑单元本身的物理实现方式并不是最重要的，这些逻辑单元所实现的功能的组合是才解决本发明所提出的技术问题的关键。此外，为了突出本发明的创新部分，本发明上述各设备实施方式并没有将与解决本发明所提出的技术问题关系不太密切的单元引入，这并不表明上述设备实施方式并不存在其它的单元。
虽然通过参照本发明的某些优选实施方式，已经对本发明进行了图示和描述，但本领域的普通技术人员应该明白，可以在形式上和细节上对其作各种改变，而不偏离本发明的精神和范围。
 本发明的实施方式
工业实用性
序列表自由内容

Claims (11)

1 ．一种级联滤波器标定去噪强度的动态设定方法，该级联滤波器至少包括前后级联的第一滤波器和第二滤波器，其特征在于，所述方法包括以下步骤：
根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计；
根据第一滤波器的标定去噪强度和所述估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度。
2 ．根据权利要求 1 所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，所述根据第一滤波器的标定去噪强度和所述估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度的步骤包括以下子步骤：
使用第一滤波器的标定去噪强度减去所述估计得到的第一滤波器去噪能力，得到差值；
根据所述差值设定第二滤波器的标定去噪强度。
3 ．根据权利要求 2 所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，所述根据差值设定第二滤波器的标定去噪强度的步骤中，
根据所述差值和参考因素共同决定第二滤波器的标定去噪强度，其中所述参考因素是以下各项因素中的任意一项或任意组合：
视频运动信息、历史视频信息、图像局部纹理信息。
4 ．根据权利要求 3 所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，所述标定去噪强度是一个表示对整幅图像整体去噪的强度的标量，或者是一个表示图像局部去噪强度的矩阵。
5 ．根据权利要求 4 所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，所述级联滤波器中至少包括一个时域滤波器和一个空域滤波器。
6 ．根据权利要求 4 所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，所述级联滤波器中包括一个时空域滤波器，其它为时域滤波器或空域滤波器。
7 ．根据权利要求 4 所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，所述根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计的步骤中，
根据输入第一滤波器的噪声图像和第一滤波器输出的去噪图像估计第一滤波器的去噪能力；
所述噪声图像是整幅图像，或者整幅图像中的局部数据块。
8 ．根据权利要求 1 至 7 中任意一项所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，在所述得到差值的步骤之前还包括以下步骤：
计算图像局部块的运动信息；
由所述图像局部块的运动信息从运动标定表中查取运动标定系数，该运动标定表用于标定运动大小与去噪强度的关系；
所述根据差值设定第二滤波器的标定去噪强度的步骤中，
将所述运动标定系数与所述差值的乘积作为第二滤波器的标定去噪强度。
9 ．根据权利要求 1 至 7 中任意一项所述的级联滤波器标定去噪强度的动态设定方法，其特征在于，在所述得到差值的步骤之前还包括以下步骤：
计算图像局部块的运动信息；
由所述图像局部块的运动信息从运动标定表中查取运动标定系数，该运动标定表用于标定运动大小与去噪强度的关系；
所述根据差值设定第二滤波器的标定去噪强度的步骤进一步包括以下子步骤：
将所述运动标定系数与所述差值的乘积作为第二滤波器标定的图像局部块的去噪强度；
统计第二滤波器所有标定的图像局部块的去噪强度，采用其均值作为第二滤波器的标定去噪强度。
10 ．一种级联滤波器，至少包括前后级联的第一滤波器和第二滤波器，其特征在于，还包括：
估计单元，用于根据第一滤波器的去噪结果对第一滤波器去噪能力进行估计；
标定单元，用于根据第一滤波器的标定去噪强度和所述估计得到的第一滤波器去噪能力，设定第二滤波器的标定去噪强度。
11 ．根据权利要求 10 所述的级联滤波器，其特征在于，
所述标定单元进一步包括：
差值计算子单元：用于使用第一滤波器的标定去噪强度减去所述估计单元输出的第一滤波器去噪能力，得到差值，
标定强度计算和设置子单元，用于根据所述差值计算子单元计算得到的差值设定第二滤波器的标定去噪强度。
所述级联滤波器中至少包括一个时域滤波器和一个空域滤波器；或者，
所述级联滤波器中包括一个时空域滤波器，其它为时域滤波器或空域滤波器。
PCT/CN2011/080033 2011-04-29 2011-09-22 级联滤波器及其标定去噪强度的动态设定方法 Ceased WO2012145989A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110112920.9 2011-04-29
CN201110112920.9A CN102170520B (zh) 2011-04-29 2011-04-29 级联滤波器及其标定去噪强度的动态设定方法

Publications (1)

Publication Number Publication Date
WO2012145989A1 true WO2012145989A1 (zh) 2012-11-01

Family

ID=44491481

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2011/080033 Ceased WO2012145989A1 (zh) 2011-04-29 2011-09-22 级联滤波器及其标定去噪强度的动态设定方法

Country Status (2)

Country Link
CN (1) CN102170520B (zh)
WO (1) WO2012145989A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113485010A (zh) * 2021-07-16 2021-10-08 南开大学 一种基于光学透镜的图像加密物理光学系统及方法
CN113674316A (zh) * 2021-08-04 2021-11-19 浙江大华技术股份有限公司 一种视频降噪方法和装置及设备

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102170520B (zh) * 2011-04-29 2013-12-25 杭州海康威视数字技术股份有限公司 级联滤波器及其标定去噪强度的动态设定方法
CN104061908B (zh) * 2014-06-13 2016-06-01 北京空间机电研究所 一种红外遥感器背景电平漂移消除系统及其方法
CN105898109A (zh) * 2015-01-26 2016-08-24 北京英潮元吉科技有限公司 视频线路滤波器
CN106296609B (zh) * 2016-08-09 2023-09-12 广东盈动高科自动化有限公司 用于容栅的全波和数字滤波去噪方法和电路
CN106815821B (zh) * 2017-01-23 2020-08-28 上海兴芯微电子科技有限公司 近红外图像的去噪方法和装置
CN113643209B (zh) * 2021-08-25 2025-01-10 Oppo广东移动通信有限公司 图像降噪处理方法、装置、存储介质与电子设备
CN114529483B (zh) * 2022-02-10 2024-12-06 Oppo广东移动通信有限公司 数据处理方法、装置、终端和可读存储介质
CN118258454B (zh) * 2024-05-31 2024-07-30 福建澳泰自动化设备有限公司 一种基于声层析断面检测的水流量检测系统
CN119671160B (zh) * 2024-11-29 2025-12-16 邦宁数字技术股份有限公司 基于人工智能及物联网的巡检服务管理系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1514649A (zh) * 2003-08-11 2004-07-21 深圳市中太多媒体通信有限公司 一种用于视频信号压缩的图像前处理方法
CN101110903A (zh) * 2007-08-31 2008-01-23 湖北科创高新网络视频股份有限公司 一种视频数据实时去噪的方法和系统
CN101795352A (zh) * 2009-02-03 2010-08-04 矽统科技股份有限公司 动态噪声滤波装置及西格玛滤波方法
CN102170520A (zh) * 2011-04-29 2011-08-31 杭州海康威视软件有限公司 级联滤波器及其标定去噪强度的动态设定方法

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6937292B1 (en) * 1992-04-22 2005-08-30 Samsung Electronics Co., Ltd. Ghost cancellation reference signal with bessel chirps and PN sequences, and TV receiver using such signal
JPH1185166A (ja) * 1997-09-04 1999-03-30 Fujitsu Ltd 能動騒音制御装置
GB9807211D0 (en) * 1998-04-03 1998-06-03 Nds Ltd Filtering of digital video signals
US6904444B2 (en) * 2001-04-12 2005-06-07 The United States Of America As Represented By The Secretary Of The Navy Pseudo-median cascaded canceller
EP1976122A1 (en) * 2007-03-31 2008-10-01 Sony Deutschland Gmbh Adaptive filter device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1514649A (zh) * 2003-08-11 2004-07-21 深圳市中太多媒体通信有限公司 一种用于视频信号压缩的图像前处理方法
CN101110903A (zh) * 2007-08-31 2008-01-23 湖北科创高新网络视频股份有限公司 一种视频数据实时去噪的方法和系统
CN101795352A (zh) * 2009-02-03 2010-08-04 矽统科技股份有限公司 动态噪声滤波装置及西格玛滤波方法
CN102170520A (zh) * 2011-04-29 2011-08-31 杭州海康威视软件有限公司 级联滤波器及其标定去噪强度的动态设定方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113485010A (zh) * 2021-07-16 2021-10-08 南开大学 一种基于光学透镜的图像加密物理光学系统及方法
CN113674316A (zh) * 2021-08-04 2021-11-19 浙江大华技术股份有限公司 一种视频降噪方法和装置及设备

Also Published As

Publication number Publication date
CN102170520B (zh) 2013-12-25
CN102170520A (zh) 2011-08-31

Similar Documents

Publication Publication Date Title
WO2012145989A1 (zh) 级联滤波器及其标定去噪强度的动态设定方法
CN106375740B (zh) 生成rgb图像的方法、装置和系统
WO2022108061A1 (en) Method and device for joint denoising and demosaicing using neural network
US11481873B2 (en) Method and apparatus for image processing
CN101770636B (zh) 图像噪声去除方法及装置
CN110493583B (zh) 图像处理方法、装置、电子设备及计算机可读存储介质
KR101389562B1 (ko) 이미지 처리 장치 및 방법
TW200931963A (en) Image signal processing circuit, image pickup apparatus and image signal processing method as well as computer program
WO2020056567A1 (zh) 图像处理方法、装置、电子设备及可读存储介质
CN104751441A (zh) 图像噪声估测的方法与装置
CN107454285A (zh) 图像噪声估计方法及装置与图像提取装置
KR100735561B1 (ko) 이미지 센서로부터 발생되는 잡음을 저감하는 방법 및 장치
CN108122201A (zh) 一种拜耳插值拉链效应去除方法
KR20070078463A (ko) 이미지 센서 노이즈 저감 장치 및 방법
CN104680484B (zh) 一种图像增强的方法及装置
NO20190396A1 (en) Isp bias-compensating noise reduction systems and methods
CN105809677B (zh) 一种基于双边滤波器的图像边缘检测方法及系统
CN105991900B (zh) 噪声检测方法和去噪方法
TWI670708B (zh) 顏色偏移校正方法及裝置
WO2015088106A1 (ko) 영상 노이즈 제거 방법 및 시스템
US8054348B2 (en) Noise reduction device and digital camera
CN117830238A (zh) 一种图像偏色检测方法、装置、电子设备及存储介质
CN115619640A (zh) 图像处理方法及装置、终端、芯片、芯片模组
CN103533320B (zh) 一种Bayer格式图像联合去噪插值的方法和系统
JP2004173060A (ja) ノイズ除去方法、撮像装置およびノイズ除去プログラム

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: 11864191

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 24/03/2014)

122 Ep: pct application non-entry in european phase

Ref document number: 11864191

Country of ref document: EP

Kind code of ref document: A1