WO2022233251A1 - 图像处理方法、装置、电子设备和存储介质 - Google Patents

图像处理方法、装置、电子设备和存储介质 Download PDF

Info

Publication number
WO2022233251A1
WO2022233251A1 PCT/CN2022/089154 CN2022089154W WO2022233251A1 WO 2022233251 A1 WO2022233251 A1 WO 2022233251A1 CN 2022089154 W CN2022089154 W CN 2022089154W WO 2022233251 A1 WO2022233251 A1 WO 2022233251A1
Authority
WO
WIPO (PCT)
Prior art keywords
filtering
preset
value
parameter
image
Prior art date
Application number
PCT/CN2022/089154
Other languages
English (en)
French (fr)
Inventor
门泽华
Original Assignee
影石创新科技股份有限公司
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 影石创新科技股份有限公司 filed Critical 影石创新科技股份有限公司
Publication of WO2022233251A1 publication Critical patent/WO2022233251A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20024—Filtering details
    • G06T2207/20032—Median filtering

Definitions

  • the present application relates to the technical field of image processing, and in particular, to an image processing method, apparatus, electronic device and storage medium.
  • a spatial domain filtering model or a temporal domain filtering model can be used to process the image data, so that the processed image is smoother; for example, the above filtering model can be a Gaussian filtering model.
  • the filtering effect of the Gaussian filter model changes with the change of the model parameters. For example, when the sliding window in the Gaussian filter model is larger, the filtering effect is better.
  • preset constraints are usually set in the Gaussian filtering model to constrain the filtered data. When a larger sliding window is used to filter the image data, the filtering results corresponding to some time points or spatial points may not satisfy the preset constraint conditions.
  • the length of the sliding window of the Gaussian filtering model needs to be reduced, resulting in poor filtering effect of image data.
  • An image processing method comprising:
  • Smoothing is performed on the candidate values of the first filtering parameter corresponding to the plurality of sampling points to obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • Filter processing is performed on the image to be processed according to a plurality of sampling points corresponding to the target value of the first filtering parameter, and a processing result of the image to be processed is output.
  • the noise corresponding to the filtering result obtained by the preset filtering model changes monotonically with the change of the first filtering parameter.
  • the above-mentioned filtering processing of the image to be processed according to the preset value of the first filtering parameter, respectively determining that multiple sampling points of the image to be processed correspond to candidate values of the first filtering parameter including:
  • the initial value is determined as the sampling point corresponding to the candidate value of the first filtering parameter
  • the filtering result corresponding to the candidate value satisfies the preset constraint condition and is the closest to the initial value.
  • the above-mentioned searching for candidate values in the preset values includes:
  • the initial value is adjusted according to the preset gradient until the filtering result corresponding to the adjusted initial setting value satisfies the preset constraint condition;
  • the preset value corresponding to the filtering result satisfying the preset constraint condition is determined as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the above-mentioned searching for candidate values in the preset values includes:
  • a binary search method is used in the preset value to search for the candidate value of the sampling point corresponding to the first filtering parameter.
  • the above-mentioned filtering processing of the image to be processed according to the preset value of the first filtering parameter, respectively determining that multiple sampling points of the image to be processed correspond to candidate values of the first filtering parameter including:
  • the noise of the filtering results corresponding to the N preset values is determined, and the preset value with the lowest noise is selected from the preset value of the sampling point corresponding to the candidate value of the first filtering parameter.
  • the above-mentioned smoothing processing is performed on the candidate values of the first filtering parameter corresponding to the multiple sampling points, and obtaining the multiple sampling points corresponding to the target value of the first filtering parameter includes:
  • the fitting value corresponding to each sampling point in the parameter fitting curve is determined as the sampling point corresponding to the target value of the first filtering parameter.
  • the preset filtering model is a Gaussian filtering model
  • the first filtering parameter is a sliding window or a standard deviation of the Gaussian filtering model.
  • An image processing device comprising:
  • the determining module is configured to filter the image to be processed according to the preset value of the first filtering parameter, and respectively determine that a plurality of sampling points of the image to be processed correspond to the candidate values of the first filtering parameter; the filtering results at the sampling points satisfy the predetermined It is assumed that among the plurality of preset values corresponding to the preset constraints of the filtering model, the noise of the filter result corresponding to the candidate value is lower than the noise of the filter result corresponding to other preset values;
  • a processing module configured to perform smoothing processing on the candidate values of the first filtering parameter corresponding to the plurality of sampling points, and obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • the filtering module is used for filtering the image to be processed according to the target value of the first filtering parameter corresponding to the plurality of sampling points, and outputting the processing result of the image to be processed.
  • An electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:
  • Smoothing is performed on the candidate values of the first filtering parameter corresponding to the plurality of sampling points to obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • Filter processing is performed on the image to be processed according to a plurality of sampling points corresponding to the target value of the first filtering parameter, and a processing result of the image to be processed is output.
  • Smoothing is performed on the candidate values of the first filtering parameter corresponding to the plurality of sampling points to obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • Filter processing is performed on the image to be processed according to a plurality of sampling points corresponding to the target value of the first filtering parameter, and a processing result of the image to be processed is output.
  • the electronic device performs filtering processing on the image to be processed according to the preset value of the first filtering parameter, and respectively determines that multiple sampling points of the image to be processed correspond to the alternatives of the first filtering parameter among the multiple preset values corresponding to the filtering results of the sampling points satisfying the preset constraints of the preset filtering model, the noise of the filtering results corresponding to the above-mentioned candidate values is lower than the noise of the filtering results corresponding to other preset values. Then, the candidate values corresponding to the first filtering parameters are smoothed for the multiple sampling points, and the obtained multiple sampling points correspond to the target values of the first filtering parameters; finally, according to the multiple sampling points corresponding to the first The target value of the filtering parameter performs filtering processing on the image to be processed, and outputs the processing result of the image to be processed.
  • the electronic device determines that multiple sampling points correspond to candidate values of the first filtering parameter, the value of the first filtering parameter with the best filtering effect that satisfies the preset constraint conditions can be determined for each sampling point, so that each sampling point can The best filtering effect is obtained when the filtering result satisfies the preset constraint conditions; further, because the electronic device performs smoothing processing on each candidate value to obtain the target value corresponding to each sampling point, so that each sampling point is determined by the target value. After the data is processed, the obtained image processing result is smoother and the filtering effect is better.
  • FIG. 1 is a schematic flowchart of an image processing method in one embodiment
  • FIG. 2 is a schematic flowchart of an image processing method in one embodiment
  • FIG. 3 is a schematic flowchart of an image processing method in one embodiment
  • FIG. 5 is a schematic diagram of an image processing method in one embodiment
  • FIG. 6 is a schematic diagram of an image processing method in one embodiment
  • FIG. 7 is a structural block diagram of an image processing apparatus in one embodiment
  • FIG. 8 is a structural block diagram of an image processing apparatus in one embodiment
  • FIG. 9 is a structural block diagram of an image processing apparatus in one embodiment
  • FIG. 10 is a structural block diagram of an image processing apparatus in one embodiment
  • FIG. 11 is an internal structure diagram of an electronic device in one embodiment.
  • the image processing method provided by the present application can be applied to electronic equipment, and the electronic equipment can perform data processing on image data to reduce noise in the image data.
  • the above electronic devices can be, but are not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.
  • the above-mentioned electronic equipment may also be imaging equipment such as cameras and video cameras; the above-mentioned cameras may be, but are not limited to, ordinary cameras, pocket cameras, anti-shake cameras, and virtual reality (Virtual Reality) cameras.
  • Reality VR for short panoramic camera, action camera, and consumer-grade or professional-grade panoramic camera, etc.
  • an image processing method is provided, and the method is applied to the electronic device in FIG. 1 as an example for description, including:
  • the above image to be processed may be data collected by an electronic device, or may be data sent by the electronic device receiving data from other devices, which is not limited herein.
  • the above-mentioned image to be processed may be a single-frame image, or may be a video including multiple frames.
  • the above-mentioned sampling points may be temporal sampling points in the image to be processed, or may be spatial sampling points in the image to be processed, which are not limited herein.
  • an image lower than a single shot may be the data of the time sampling point corresponding to the shooting moment; for the video, the data of the above-mentioned spatial sampling point may be the data collected at different times for the same spatial position.
  • the above-mentioned multiple sampling points may be each sampling point in the image to be processed, or may be one sampling point in one of the sampling intervals in the image to be processed.
  • the above-mentioned preset filtering model may perform spatial domain filtering processing on the to-be-processed image, or may perform temporal-domain filtering processing on the to-be-processed image to reduce noise in the to-be-processed image.
  • the above noise may be redundant interference information in the image to be processed, which affects the display effect of the image.
  • the above-mentioned preset filter model may be a Gaussian filter, a median filter, or an average filter, etc.
  • the type of the above-mentioned preset filter model is not limited herein.
  • the above-mentioned Gaussian filter may be a one-dimensional Gaussian filter or a two-dimensional Gaussian filter.
  • the above-mentioned first filtering parameter may be a model parameter in a preset filtering model, and is related to the filtering effect of the preset filtering model, such as a sliding window in a median filter.
  • the above filtering effect can be the noise in the image processing result obtained after the image to be processed is processed by the preset filtering model, or the noise reduction ratio of the image to be processed by the preset filtering model, or the information of the image processing result.
  • the noise ratio is not limited here; the better the above filtering effect, the lower the noise in the filtered image processing result.
  • the noise corresponding to the filtering result obtained by the preset filtering model changes monotonically with the change of the first filtering parameter.
  • the noise of the filtering result obtained by the above preset filtering model may change linearly with the change of the first filtering parameter, or may exhibit a nonlinear monotonic change with the change of the first filtering parameter, which is not limited herein.
  • the above-mentioned first filtering parameter may be positively related to the filtering effect of the preset filtering model, or may be negatively related to the filtering effect of the preset filtering model; it is not limited herein.
  • the preset filtering model is a Gaussian filtering model
  • the first filtering parameter is a sliding window or a standard deviation of the Gaussian filtering model.
  • the preset value of the above-mentioned first filter parameter may be the preset range of the first filter parameter, or may include a plurality of discrete values, which is not limited here.
  • the preset value of the sliding window of the Gaussian filter model can be a sliding window interval, such as [A1, A2]; it can also be a sliding window value, such as A1, A2, A3... An and other values; for the preset value
  • the specific form is not limited here.
  • the electronic device can filter the data of multiple sampling points in the image to be processed according to the preset value of the first filtering parameter, and obtain the filtering result corresponding to the sampling point; further, the electronic device can determine the sampling point according to the filtering result.
  • the purpose of the above filtering process is to determine the candidate values corresponding to the sampling points.
  • the above-mentioned candidate value is one of the preset values of the first filtering parameter.
  • the corresponding candidate value is The noise of the filtering result is lower than the noise of the filtering results corresponding to other preset values; that is, after filtering the data of the sampling point with the alternative value, the best filtering result can be obtained under the condition that the filtering result satisfies the preset constraints. filter effect.
  • the electronic device may select candidate values according to the filtering results corresponding to each preset value by traversing, or may search for the aforementioned candidate values in the preset values according to the preset search method, which is not limited herein.
  • the above-mentioned preset constraint conditions may be preset in the electronic device, or may be obtained by adjusting the image processing result of the previous image, and are used to constrain the image processed by the preset filter model;
  • the constraint condition can be the upper limit of the filtering result corresponding to each sampling point, the lower limit of the filtering result corresponding to each sampling point, or the constraint of the variation range of the data of the sampling point before and after processing by the preset filtering model, so that After the preset filtering model processes the data of the sampling points, the data cannot be too different from the data before the processing.
  • the variation of the candidate values with the positions of the sampling points may be discontinuous, and the electronic device may perform smoothing processing on the candidate values. Specifically, the electronic device can connect the candidate values of each sampling point according to the positional relationship of the sampling points, determine the discontinuous points in the connected curve, and adjust the alternative values of the remaining sampling points around the discontinuous sampling point. value to make the curve smooth.
  • the electronic device may take the discontinuous point as a starting point and adjust the candidate values of the remaining sampling points in sequence in a near-to-far manner to obtain the target value of the first filter parameter adjusted for each sampling point. It should be noted that after processing the data of the sampling point by using the target value, the obtained filtering result needs to satisfy the preset constraint conditions.
  • S103 Perform filtering processing on the image to be processed according to the target values of the first filtering parameters corresponding to the plurality of sampling points, and output a processing result of the image to be processed.
  • the electronic device can use the target value corresponding to the sampling point to filter the data of the sampling point, obtain the filtering result corresponding to the data of each sampling point, and output the processing result corresponding to the image to be processed.
  • the electronic device performs filtering processing on the image to be processed according to the preset value of the first filtering parameter, and respectively determines that multiple sampling points of the image to be processed correspond to the candidate values of the first filtering parameter; the filtering results at the sampling points Among the multiple preset values corresponding to the preset constraint conditions of the preset filter model, the noise of the filter result corresponding to the above-mentioned candidate value is lower than the noise of the filter result corresponding to the other preset values; then, for the multiple sampling points Corresponding to the candidate value of the first filtering parameter, smoothing is performed to obtain a plurality of sampling points corresponding to the target value of the first filtering parameter; finally, the image to be processed is corresponding to the target value of the first filtering parameter according to the plurality of sampling points Filter processing is performed, and the processing result of the image to be processed is output.
  • the electronic device determines that multiple sampling points correspond to candidate values of the first filtering parameter, the value of the first filtering parameter with the best filtering effect that satisfies the preset constraint conditions can be determined for each sampling point, so that each sampling point can The best filtering effect is obtained when the filtering result satisfies the preset constraint conditions; further, because the electronic device performs smoothing processing on each candidate value to obtain the target value corresponding to each sampling point, so that each sampling point is determined by the target value. After the data is processed, the obtained image processing result is smoother and the filtering effect is better.
  • FIG. 2 is a schematic flowchart of an image processing method in an embodiment. This embodiment relates to a way for an electronic device to determine a candidate value.
  • the foregoing S101 includes:
  • the electronic device may determine which preset value corresponds to the preset value according to the preset value of the first filtering parameter of the preset filtering model The filtered result has the lowest noise. Specifically, the electronic device can obtain the monotonic variation relationship between the noise of the filtering result and the first filtering parameter. For example, the above monotonic variation relationship is that the larger the first filtering parameter, the better the filtering effect.
  • the electronic device can set the maximum value of the preset values Determined as the initial value of the first filtering parameter, that is, the global optimal parameter; each sampling point in the image to be processed can obtain the best filtering effect by using this initial value, and the obtained filtering result has the lowest noise, but there may be some
  • the filtering results corresponding to the sampling points do not satisfy the preset constraints.
  • the electronic device may determine the minimum value among the preset values as the initial value of the first filtering parameter.
  • the electronic device may set the first filtering parameter in the preset filtering model to the initial value, and then perform filtering processing on the data corresponding to each sampling point in the plurality of sampling points, respectively, Obtain the initial processing result of the sampling point.
  • the electronic device may match the initial processing result of the sampling point with the preset constraint condition corresponding to the sampling point, and determine whether the initial processing result of the sampling point satisfies the preset constraint condition. If the initial processing result of the sampling point satisfies the preset constraint condition, the electronic device may determine the initial value as the candidate value of the first filtering parameter corresponding to the sampling point.
  • the electronic device may search the preset value for the corresponding filter result that satisfies the preset constraint conditions according to the monotonic variation relationship between the noise of the filtering result and the first filtering parameter. Set the value, and obtain the candidate value of the best filtering effect under the preset constraints.
  • the electronic device can determine the candidate value according to the monotonous variation relationship between the noise of the filtering result and the first filtering parameter when searching for the candidate value For example, if the above-mentioned monotonic variation relationship is that the larger the first filtering parameter, the lower the noise of the filtering result, the electronic device can use the next preset value smaller than the initial value to filter the data of the sampling point, and then determine the next Whether the filter result corresponding to a preset value satisfies the preset constraint condition; if the above-mentioned monotonic variation relationship is that the smaller the first filter parameter, the lower the noise of the filter result, the next preset value greater than the initial value can be used to pair the data of the sampling point Perform filtering processing, and then determine whether the filtering result corresponding to the next preset value satisfies the preset constraint condition.
  • the electronic device adjusts the initial value according to the preset gradient until the filtering result corresponding to the adjusted initial value satisfies the preset constraint condition;
  • the preset value is determined as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the above-mentioned first filtering parameter is a sliding window in the Gaussian filter model
  • the preset value of the above-mentioned sliding window may include a plurality of window lengths arranged at equal intervals, namely A1, A2, A3...A100; and, the Gaussian filter model The larger the sliding window of , the lower the noise of the filtering result.
  • the electronic device can set A100 as the initial value. If one of the sampling points uses A100, the filtering result does not meet the preset constraints of the sampling point, then the electronic device can use A99 to filter the data of the sampling point. If A99 corresponds to If the filtering result satisfies the preset constraint condition, then A99 may be determined as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the electronic device may also use an initial value as a starting point, and use a dichotomy method to search for candidate values of the sampling point corresponding to the first filtering parameter in the preset value.
  • a dichotomy method to search for candidate values of the sampling point corresponding to the first filtering parameter in the preset value.
  • the A80 is further used to filter the data of the sampling point; if the filtering result corresponding to the A90 does not meet the preset constraint conditions, the electronic device can further use the search between A80 and A90 to satisfy the preset constraint conditions.
  • the preset value of the constraint condition if the filter result corresponding to A90 satisfies the preset constraint condition, the electronic device can search for the preset value that satisfies the preset constraint condition between A90 and A100 until the result that satisfies the preset constraint condition is obtained.
  • the preset value with lower noise is determined as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the electronic device on the basis of determining the initial value, can quickly obtain the candidate value of each sampling point corresponding to the first filter parameter according to the monotonic change relationship between the first filter parameter and the noise of the filter result, Therefore, the filtering result of each sampling point can obtain the best filtering effect under the condition that the preset constraint condition is satisfied.
  • FIG. 3 is a schematic flowchart of an image processing method in an embodiment. This embodiment relates to a way for an electronic device to determine a candidate value.
  • the foregoing S101 includes:
  • the electronic device may, for each sampling point, in the preset value of the first filtering parameter, filter out a plurality of preset values in which the filtering result corresponding to the sampling point satisfies the preset constraint condition. set value. Specifically, when the preset value of the first filtering parameter is a plurality of discrete values, the electronic device may traverse all the preset values of the first filtering parameter for each sampling point of the image to be processed, and obtain a condition that satisfies the preset constraint. The N preset values of .
  • the electronic device can select a plurality of preset values in the preset value interval at preset intervals, and then, for each sampling point of the image to be processed, traverse the For the above-mentioned multiple preset values, N preset values satisfying the preset constraint conditions are obtained.
  • the electronic device may further determine the noise of the filtering results corresponding to the N preset values, and select the preset value with the lowest noise from the preset value as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the electronic device can traverse each preset value, and filter out N whose filtering results satisfy the preset constraints.
  • a preset value is obtained, and the value of the first filtering parameter with the best filtering effect under the preset constraint condition is obtained by further quantifying and comparing the filtering effect, so that a better image processing effect can be obtained.
  • FIG. 4 is a schematic flowchart of an image processing method in an embodiment. This embodiment relates to a way for an electronic device to determine a candidate value.
  • the foregoing S103 includes:
  • the electronic device may determine the sampling point corresponding to the candidate value different from the initial value as the control point according to the relationship between the candidate value and the initial value. If the first filter parameter of the preset filter model is larger, the better the filter effect is, and the above-mentioned alternative values different from the initial value are alternative values smaller than the initial value; the smaller the first filter parameter of the preset filter model, the better the filter effect. , the above-mentioned alternative values that are different from the initial value are alternative values that are greater than the initial value.
  • the electronic device can keep the candidate values of the control points unchanged, and curve the candidate values of other sampling points.
  • the fitting process is performed to obtain a parameter fitting curve.
  • the above-mentioned curve fitting processing may be linear fitting processing or quadratic curve fitting processing, which is not limited herein.
  • the candidate values of each sampling point can be as shown in Figure 5, the abscissa represents the sampling point, and the ordinate represents the value of the first filter parameter; the discontinuous points in the figure are control points, and the candidate values in Figure 5 are plotted.
  • the parameter fitting curve shown in Figure 6 can be obtained.
  • the fitting value of each sampling point is within the preset value range of the first filtering parameter. That is to say, if the larger the first filter parameter of the preset filtering model is, the better the filtering effect is.
  • the fitting value of the control point is the same as the candidate value, and the fitting values of the remaining sampling points are the same as the candidate value. It can be smaller than the initial value or equal to the initial value; if the smaller the first filtering parameter of the preset filtering model, the better the filtering effect, among the fitting values of the above sampling points, the fitting value of the control point is the same as the alternative value,
  • the fitted values of the remaining sampling points can be greater than or equal to the initial value.
  • the electronic determines the fitting value corresponding to each sampling point in the parameter fitting curve as the sampling point corresponding to the target value of the first filtering parameter; and then according to the plurality of sampling points corresponding to the first filtering parameter A target value of a filtering parameter is used to filter the image to be processed, and a processing result of the image to be processed is output.
  • the electronic device may sequentially use the target value of each sampling point to perform filtering processing on the data of the sampling point according to the arrangement sequence of the sampling points, and output the processing result of the image to be processed.
  • the electronic device adopts the above curve fitting method to obtain the target value of each sampling point corresponding to the first filtering parameter, which may be that the filtering result of each sampling point has the lowest noise when the preset constraint conditions are satisfied, and can also make the target value of each sampling point corresponding to the first filtering parameter.
  • the image processing result of the processed image is smoother, the filtering effect is better, and a better image display effect is obtained.
  • steps in the flowcharts of FIGS. 2-4 are shown in sequence according to the arrows, these steps are not necessarily executed in the sequence shown by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in FIGS. 2-4 may include multiple steps or multiple stages. These steps or stages are not necessarily executed and completed at the same time, but may be executed at different times. The execution of these steps or stages The order is also not necessarily sequential, but may be performed alternately or alternately with other steps or at least a portion of the steps or phases within the other steps.
  • an image processing apparatus including:
  • the determining module 10 is configured to filter the image to be processed according to the preset value of the first filtering parameter, and respectively determine that multiple sampling points of the image to be processed correspond to candidate values of the first filtering parameter; the filtering results at the sampling points satisfy Among the multiple preset values corresponding to the preset constraints of the preset filter model, the noise of the filter result corresponding to the candidate value is lower than the noise of the filter result corresponding to other preset values;
  • the processing module 20 is configured to perform smoothing processing on the candidate values of the first filtering parameter corresponding to the plurality of sampling points, and obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • the filtering module 30 is configured to perform filtering processing on the image to be processed according to a plurality of sampling points corresponding to the target value of the first filtering parameter, and output a processing result of the image to be processed.
  • the noise corresponding to the filtering result obtained by using the preset filtering model changes monotonically with the change of the first filtering parameter.
  • the foregoing determining module 10 includes:
  • a first determining unit 101 configured to determine a preset value corresponding to the lowest noise as the initial value of the first filtering parameter according to the monotonic variation relationship between the noise of the filtering result and the first filtering parameter;
  • the processing unit 102 is configured to, for each sampling point of the to-be-processed image, perform filtering processing on the data of the sampling point by using an initial value to obtain an initial processing result of the sampling point;
  • the second determining unit 103 is configured to determine the initial value as the candidate value of the sampling point corresponding to the first filtering parameter when the initial processing result satisfies the preset constraint condition;
  • the search unit 104 is configured to search for an alternative value in the preset value when the initial processing result does not meet the preset constraint condition; the filtering result corresponding to the alternative value satisfies the preset constraint condition, and is consistent with the preset constraint condition.
  • the initial value is the closest.
  • the above-mentioned search unit 104 is specifically configured to: on the basis of the initial value, adjust the initial value according to the preset gradient, until the filtering result corresponding to the adjusted initial value satisfies the preset value Constraints; the preset values corresponding to the filtering results satisfying the preset constraints are determined as the sampling points corresponding to the candidate values of the first filtering parameters.
  • the above-mentioned search unit 104 is specifically configured to: take the initial value as the starting point, and use the dichotomy method to search for the candidate value of the sampling point corresponding to the first filtering parameter in the preset value.
  • the foregoing determining module 10 includes:
  • Traversing unit 105 configured to traverse all preset values of the first filter parameter for each sampling point of the image to be processed, and obtain N preset values that satisfy preset constraints;
  • the selection unit 106 is configured to determine the noise of the filtering results corresponding to the N preset values, and select the preset value with the lowest noise among the sampling points corresponding to the candidate value of the first filtering parameter.
  • the foregoing processing module 20 includes:
  • the third determination unit 201 is configured to determine the sampling point corresponding to the candidate value different from the initial value as the control point;
  • the fitting unit 202 is configured to keep the candidate values of the control points unchanged, and perform curve fitting processing on the candidate values of other sampling points to obtain a parameter fitting curve;
  • the fourth determining unit 203 is configured to determine the fitting value corresponding to each sampling point in the parameter fitting curve as the sampling point corresponding to the target value of the first filtering parameter.
  • the preset filtering model is a Gaussian filtering model
  • the first filtering parameter is a sliding window or a standard deviation of the Gaussian filtering model
  • the image processing apparatus provided above can execute the above image processing method embodiments, and the implementation principles and technical effects thereof are similar, and details are not described herein again.
  • Each module in the above-mentioned image processing apparatus may be implemented in whole or in part by software, hardware, and combinations thereof.
  • the above modules can be embedded in or independent of the processor in the electronic device in the form of hardware, or stored in the memory in the electronic device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
  • an electronic device in one embodiment, the electronic device may be a server, and its internal structure diagram may be as shown in FIG. 11 .
  • the electronic device includes a processor, memory, and a network interface connected by a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities.
  • the memory of the electronic device includes a non-volatile storage medium and an internal memory.
  • the nonvolatile storage medium stores an operating system, a computer program, and a database.
  • the internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium.
  • the database of the electronic device is used to store data processing data.
  • the network interface of the electronic device is used to communicate with an external terminal through a network connection.
  • the computer program implements an image processing method when executed by a processor.
  • FIG. 11 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the electronic device to which the solution of the present application is applied.
  • the specific electronic device may be Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.
  • an electronic device including a memory and a processor, a computer program is stored in the memory, and the processor implements the following steps when executing the computer program:
  • Smoothing is performed on the candidate values of the first filtering parameter corresponding to the plurality of sampling points to obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • Filter processing is performed on the image to be processed according to a plurality of sampling points corresponding to the target value of the first filtering parameter, and a processing result of the image to be processed is output.
  • the noise corresponding to the filtering result obtained by using the preset filtering model changes monotonically with the change of the first filtering parameter.
  • the following step is further implemented: according to the monotonic variation relationship between the noise of the filtering result and the first filtering parameter, determining a preset value corresponding to the lowest noise as the first filtering parameter For each sampling point of the image to be processed, use the initial value to filter the data of the sampling point to obtain the initial processing result of the sampling point; if the initial processing result satisfies the preset constraints, the initial value is determined as The sampling point corresponds to the candidate value of the first filtering parameter; if the initial processing result does not satisfy the preset constraint condition, search the candidate value in the preset value; the filter result corresponding to the candidate value satisfies the preset constraint condition, and is the closest to the initial value.
  • the processor further implements the following steps when executing the computer program: on the basis of the initial value, adjusting the initial value according to a preset gradient until the filtering result corresponding to the adjusted initial value satisfies the preset constraint condition; The preset value corresponding to the filtering result satisfying the preset constraint condition is determined as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the processor when the processor executes the computer program, the processor further implements the following steps: starting with an initial value, and searching for candidate values of the sampling point corresponding to the first filtering parameter in the preset value by using a dichotomy method.
  • the processor further implements the following steps when executing the computer program: for each sampling point of the image to be processed, traverse all preset values of the first filtering parameter, and obtain N preset values that satisfy preset constraints ; determine the noise of the filtering results corresponding to the N preset values, and select the preset value with the lowest noise among them as the candidate value of the sampling point corresponding to the first filtering parameter.
  • the processor executes the computer program, the following steps are further implemented: determining a sampling point corresponding to a candidate value different from the initial value as a control point; keeping the candidate value of the control point unchanged, for other sampling points Perform curve fitting processing on the candidate values of , to obtain a parameter fitting curve; the fitting value corresponding to each sampling point in the parameter fitting curve is determined as the sampling point corresponding to the target value of the first filtering parameter.
  • the preset filtering model is a Gaussian filtering model
  • the first filtering parameter is a sliding window or a standard deviation of the Gaussian filtering model.
  • a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:
  • Smoothing is performed on the candidate values of the first filtering parameter corresponding to the plurality of sampling points to obtain the target value of the first filtering parameter corresponding to the plurality of sampling points;
  • Filter processing is performed on the image to be processed according to a plurality of sampling points corresponding to the target value of the first filtering parameter, and a processing result of the image to be processed is output.
  • the noise corresponding to the filtering result obtained by using the preset filtering model changes monotonically with the change of the first filtering parameter.
  • the following step is further implemented: according to the monotonic variation relationship between the noise of the filtering result and the first filtering parameter, determine the preset value corresponding to the lowest noise as the first filtering parameter The initial value of the parameter; for each sampling point of the image to be processed, the initial value is used to filter the data of the sampling point to obtain the initial processing result of the sampling point; if the initial processing result satisfies the preset constraints, the initial value is determined.
  • the filter result corresponding to the candidate value satisfies the preset constraints, and is the closest to the initial value.
  • the following steps are further implemented: on the basis of the initial value, the initial value is adjusted according to a preset gradient, until the filtering result corresponding to the adjusted initial value satisfies the preset constraint condition; The preset value corresponding to the filtering result satisfying the preset constraint condition is determined as the sampling point corresponding to the candidate value of the first filtering parameter.
  • the following steps are further implemented: starting with an initial value, and searching for candidate values of the sampling point corresponding to the first filtering parameter in the preset value by a binary method.
  • the following steps are further implemented: for each sampling point of the image to be processed, traverse all preset values of the first filtering parameter, and obtain N presets that satisfy the preset constraints value; determine the noise of the filtering results corresponding to the N preset values, and select the preset value with the lowest noise among them as the candidate value of the sampling point corresponding to the first filtering parameter.
  • the following steps are further implemented: determining a sampling point corresponding to a candidate value different from the initial value as a control point; keeping the candidate value of the control point unchanged, and performing other sampling
  • the candidate value of the point is subjected to curve fitting processing to obtain a parameter fitting curve; the fitting value corresponding to each sampling point in the parameter fitting curve is determined as the sampling point corresponding to the target value of the first filtering parameter.
  • the preset filtering model is a Gaussian filtering model
  • the first filtering parameter is a sliding window or a standard deviation of the Gaussian filtering model.
  • Non-volatile memory may include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory or optical memory, etc.
  • Volatile memory may include random access memory (Random Access Memory) Access Memory, RAM) or external cache memory.
  • RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (Dynamic Random Access Memory). Access Memory, DRAM), etc.

Landscapes

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

Abstract

本申请涉及一种图像处理方法、装置、电子设备和存储介质，电子设备根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，上述备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。采用上述方法可以提升滤波效果，并使得图像处理结果更平滑，滤波效果更好。

Description

图像处理方法、装置、电子设备和存储介质 技术领域
本申请涉及图像处理技术领域，特别是涉及一种图像处理方法、装置、电子设备和存储介质。
 背景技术
电子设备对图像数据进行处理的过程中，通常需要对获取的图像数据进行滤波处理，降低图像数据中的噪声信号。
传统方法中，可以采用空间域滤波模型或时间域滤波模型对图像数据进行处理，使得处理后的图像更平滑；例如，上述滤波模型可以是高斯滤波模型。一般情况下，高斯滤波模型的滤波效果随模型参数的变化而变化，例如当高斯滤波模型中的滑动窗口越大，滤波效果越好。为了实现一定的滤波效果，高斯滤波模型中通常设置预设约束条件，对滤波处理后的数据进行约束。当采用较大的滑动窗口对图像数据进行滤波处理时，可能导致部分时间点或空间点对应的滤波结果不满足预设约束条件。
需要说明的是，在上述背景技术部分公开的信息仅用于加强对本申请的背景的理解，因此可以包括不构成对本领域普通技术人员已知的现有技术的信息。
 技术问题
通常情况下，为了使得所有时间点或空间点对应的滤波结果满足预设约束条件，需要降低高斯滤波模型的滑动窗口的长度，导致图像数据的滤波效果差。
 技术解决方案
基于此，有必要针对上述技术问题，提供一种能够提升滤波效果的图像处理方法、装置、电子设备和存储介质。
一种图像处理方法，上述方法包括：
根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
在其中一个实施例中，通过预设滤波模型获得的滤波结果对应的噪声，随第一滤波参数的变化而单调变化。
在其中一个实施例中，上述根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值，包括：
根据滤波结果的噪声与所述第一滤波参数的单调变化关系，将最低噪声对应的预设值确定为所述第一滤波参数的初始值；
针对待处理图像的每一个采样点，采用初始值对采样点的数据进行滤波处理，获得采样点的初始处理结果；
若初始处理结果满足预设约束条件，则将初始值确定为采样点对应于第一滤波参数的备选值；
若初始处理结果不满足预设约束条件，则在预设值中搜索备选值；备选值对应的滤波结果满足预设约束条件，且与初始值距离最近。
在其中一个实施例中，上述在所述预设值中搜索备选值，包括：
在初始值的基础上，按照预设梯度调整初始值，直至采用调整后的初始设值对应的滤波结果满足预设约束条件；
将满足预设约束条件的滤波结果对应的预设值，确定为采样点对应于第一滤波参数的备选值。
在其中一个实施例中，上述在所述预设值中搜索备选值，包括：
以初始值为起点，在预设值中采用二分法搜索采样点对应于第一滤波参数的备选值。
在其中一个实施例中，上述根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值，包括：
针对待处理图像的每一个采样点，遍历第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值；
确定N个预设值对应的滤波结果的噪声，从中选择噪声最低的预设值为采样点对应于第一滤波参数的备选值。
在其中一个实施例中，上述对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值，包括：
将与初始值不同的备选值对应的采样点，确定为控制点；
保持控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线；
将参数拟合曲线中各个采样点对应的拟合值，确定为采样点对应于对第一滤波参数的目标值。
在其中一个实施例中，上述预设滤波模型为高斯滤波模型，第一滤波参数为高斯滤波模型的滑动窗口或标准偏差。
一种图像处理装置，上述装置包括：
确定模块，用于根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
处理模块，用于对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
滤波模块，用于根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
一种电子设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现以下步骤：
根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现以下步骤：
根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
技术效果
上述图像处理方法、装置、电子设备和存储介质，电子设备根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，上述备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；然后，对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；最后，根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。由于电子设备确定了多个采样点对应于第一滤波参数的备选值，从而可以针对每个采样点确定满足预设约束条件的滤波效果最佳的第一滤波参数的值，使得各个采样点的滤波结果满足预设约束条件的情况下，获得最佳的滤波效果；进一步地，由于电子设备对各个备选值进行平滑处理获得各个采样点对应的目标值，使得通过目标值对各个采样点的数据进行处理后，获得的图像处理结果更平滑，滤波效果更好。
 附图说明
图1为一个实施例中图像处理方法的流程示意图；
图2为一个实施例中图像处理方法的流程示意图；
图3为一个实施例中图像处理方法的流程示意图；
图4为另一个实施例中图像处理方法的流程示意图；
图5为一个实施例中图像处理方法的示意图；
图6为一个实施例中图像处理方法的示意图；
图7为一个实施例中图像处理装置的结构框图；
图8为一个实施例中图像处理装置的结构框图；
图9为一个实施例中图像处理装置的结构框图；
图10为一个实施例中图像处理装置的结构框图；
图11为一个实施例中电子设备的内部结构图。
 本发明的实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请提供的图像处理方法，可以应用于电子设备，电子设备可以对图像数据进行数据处理，降低图像数据中的噪声。上述电子设备可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备。上述电子设备还可以是相机、摄像机等影像设备；上述相机可以但不限于是普通相机、口袋相机、防抖相机、虚拟现实（Virtual Reality，简称VR）全景相机、运动相机以及消费级或专业级全景相机等。
在一个实施例中，如图1所示，提供了一种图像处理方法，以该方法应用于图1中的电子设备为例进行说明，包括：
S101、根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声。
上述待处理图像可以是电子设备采集到的数据，也可以是电子设备接收其他设备发送的数据，在此不做限定。上述待处理图像可以是单帧图像，也可以是包含多个帧的视频。上述采样点可以是待处理图像中的时间采样点，也可以是待处理图像中的空间采样点，在此不做限定。例如，低于单次拍摄的图像，可以是拍摄时刻对应的时间采样点的数据；对于视频，上述空间采样点的数据可以是针对同一个空间位置在不同的时刻采集到的数据。上述多个采样点可以是待处理图像中的每个采样点，也可以是待处理图像中其中一个采样区间中的一个采样点。
其中，上述预设滤波模型可以对待处理图像进行空间域滤波处理，也可以对待处理图像进行时间域滤波处理，降低待处理图像中的噪声。上述噪声可以是待处理图像中多余的干扰信息，影响图像的显示效果。上述预设滤波模型可以是高斯滤波器，也可以是中值滤波器，还可以是均值滤波器等，对于上述预设滤波模型的类型在此不做限定。其中，上述高斯滤波器可以是一维高斯滤波器，也可以是二维高斯滤波器。
上述第一滤波参数可以是预设滤波模型中的模型参数，与预设滤波模型的滤波效果相关，例如中值滤波器中的滑动窗口。上述滤波效果可以是预设滤波模型对待处理图像进行处理后，获得的图像处理结果中的噪声，也可以是预设滤波模型对待处理图像进行处理的降噪比例，还可以是图像处理结果的信噪比，在此不作限定；上述滤波效果越好，表征滤波后的图像处理结果中的噪声越低。可选地，通过预设滤波模型获得的滤波结果对应的噪声，随所述第一滤波参数的变化而单调变化。上述预设滤波模型获得的滤波结果的噪声，可以随第一滤波参数的变化而线性变化，也可以随第一滤波参数的变化呈现非线性的单调变化，在此不做限定。上述第一滤波参数可以与预设滤波模型的滤波效果正相关，也可以与预设滤波模型的滤波效果负相关；在此不做限定。可选地，上述预设滤波模型为高斯滤波模型的情况下，上述第一滤波参数为高斯滤波模型的滑动窗口或标准偏差。
 上述第一滤波参数的预设值可以是第一滤波参数的预设范围，也可以是包含多个离散值，在此不做限定。例如，高斯滤波模型的滑动窗口的预设值可以是滑动窗口区间，例如[A1，A2]；也可以是滑动窗口值，例如A1、A2、A3……An等多个值；对于预设值的具体形式在此不做限定。
电子设备可以根据第一滤波参数的预设值，对待处理图像中的多个采样点的数据分别进行滤波处理，获得该采样点对应的滤波结果；进一步地，电子设备可以根据滤波结果确定采样点对应于第一滤波参数的备选值。上述滤波处理的目的为确定采样点对应的备选值。上述备选值为第一滤波参数的预设值中的一个值，在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，该备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；也就是说，采用备选值对该采样点的数据进行滤波处理后，在滤波结果满足预设约束条件下，可以获得最佳的滤波效果。
电子设备可以通过遍历的方式，根据各个预设值对应的滤波结果选择备选值，也可以按照预设搜索方式在预设值中搜索上述备选值，在此不做限定。
其中，上述预设约束条件可以是电子设备中预设的，也可以是随上一图像的图像处理结果进行调整获得的，用于对预设滤波模型处理后的图像进行约束的条件；上述预设约束条件可以是各个采样点对应的滤波结果的上限，也可以是各个采样点对应的滤波结果的下限，还可以是该采样点的数据经过预设滤波模型处理前后的变化范围的约束，使得预设滤波模型对采样点的数据进行处理之后，与处理之前的数据不能相差太大。
S102、对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值。
在获得多个采样点对应于第一滤波参数的备选值之后， 备选值随采样点位置的变化可能是不连续的，电子设备既可以对备选值进行平滑处理。具体地，电子设备可以按照采样点的位置关系，将各个采样点的备选值进行连接，确定出连接后的曲线中的不连续点，通过调整不连续的采样点周围其余采样点的备选值，使得曲线平滑。电子设备可以以不连续点为起点，采用由近及远的方式，依次调整其余采样点的备选值，获得各个采样点调整的第一滤波参数的目标值。需要说明的是，采用目标值对该采样点的数据进行处理后，获得的滤波结果需要满足预设约束条件。
S103、根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
在上述步骤的基础上，电子设备可以分别采用采样点对应的目标值，对该采样点的数据进行滤波处理，获得各个采样点数据对应的滤波结果，输出该待处理图像对应的处理结果。
上述图像处理方法，电子设备根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，上述备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；然后，对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；最后，根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。由于电子设备确定了多个采样点对应于第一滤波参数的备选值，从而可以针对每个采样点确定满足预设约束条件的滤波效果最佳的第一滤波参数的值，使得各个采样点的滤波结果满足预设约束条件的情况下，获得最佳的滤波效果；进一步地，由于电子设备对各个备选值进行平滑处理获得各个采样点对应的目标值，使得通过目标值对各个采样点的数据进行处理后，获得的图像处理结果更平滑，滤波效果更好。
图2为一个实施例中图像处理方法的流程示意图，本实施例涉及电子设备确定备选值的一种方式，在上述实施例的基础上，如图2所示，上述S101包括：
S201、根据滤波结果的噪声与第一滤波参数的单调变化关系，将最低噪声对应的预设值确定为第一滤波参数的初始值。
在滤波结果对应的噪声随所述第一滤波参数的变化而单调变化的情况下，电子设备可以根据预设滤波模型的第一滤波参数的预设值，确定预设值中哪个预设值对应的滤波结果的噪声最低。具体地，电子设备可以获取滤波结果的噪声与第一滤波参数的单调变化关系，例如上述单调变化关系为第一滤波参数越大滤波效果越好时，电子设备可以将预设值中的最大值确定为第一滤波参数的初始值，也就是全局最佳参数；待处理图像中的各个采样点采用该初始值均可以获得最佳的滤波效果，获得的滤波结果的噪声最低，但是可能存在部分采样点对应的滤波结果不满足预设约束条件。相应地，上述变化关系为第一滤波参数越小滤波效果越好时，电子设备可以将预设值中的最小值确定为第一滤波参数的初始值。
S202、针对待处理图像的每一个采样点，采用初始值对采样点的数据进行滤波处理，获得采样点的初始处理结果。
在获得第一滤波参数的初始值之后，电子设备可以将预设滤波模型中的第一滤波参数设置为该初始值，然后对多个采样点中每个采样点对应的数据分别进行滤波处理，获得采样点的初始处理结果。
S203、若初始处理结果满足预设约束条件，则将初始值确定为采样点对应于第一滤波参数的备选值。
进一步地，针对每个采样点，电子设备可以将采样点的初始处理结果与该采样点对应的预设约束条件进行匹配，确定该采样点的初始处理结果是否满足预设约束条件。若该采样点的初始处理结果满足预设约束条件，电子设备可以将该初始值确定为该采样点对应的第一滤波参数的备选值。
S204、若初始处理结果不满足预设约束条件，则在预设值中搜索备选值；备选值对应的滤波结果满足预设预设约束条件，且与初始值距离最近。
若该采样点的初始处理结果不满足预设约束条件，电子设备可以根据滤波结果的噪声随第一滤波参数的单调变化关系，在预设值中搜索对应的滤波结果满足预设约束条件的预设值，并且在满足预设约束条件下获得最佳滤波效果的备选值。
由于预设滤波模型获得的滤波结果的噪声随第一滤波参数的变化而单调变化，电子设备在搜索备选值时可以根据滤波结果的噪声随第一滤波参数的单调变化关系，确定备选值的搜索方向；例如，若上述单调变化关系为第一滤波参数越大滤波结果的噪声越低，电子设备可以采用小于初始值的下一个预设值对采样点的数据进行滤波处理，然后确定下一个预设值对应的滤波结果是否满足预设约束条件；若上述单调变化关系为第一滤波参数越小滤波结果的噪声越低，可以采用大于初始值的下一个预设值对采样点的数据进行滤波处理，然后确定下一个预设值对应的滤波结果是否满足预设约束条件。具体地，电子设备在初始值的基础上，按照预设梯度调整初始值，直至采用调整后的初始值对应的滤波结果满足预设约束条件；然后，将满足预设约束条件的滤波结果对应的预设值，确定为采样点对应于第一滤波参数的备选值。
例如，上述第一滤波参数为高斯滤波模型中的滑动窗口，上述滑动窗口的预设值可以包括等间距排列的多个窗口长度，分别为A1、A2、A3……A100；并且，高斯滤波模型的滑动窗口越大，滤波结果的噪声越低。电子设备可以将A100设置为初始值，若其中一个采样点采用A100时滤波结果不满足该采样点的预设约束条件，那么电子设备可以采用A99对该采样点的数据进行滤波处理，若A99对应的滤波结果满足预设约束条件，那么可以将A99确定为该采样点对应于第一滤波参数的备选值。
可选地，在搜索备选值时，电子设备还可以以初始值为起点，在预设值中采用二分法搜索采样点对应于第一滤波参数的备选值。继续以上述高斯滤波模型为例，电子设备在其中一个采样点采用A100时滤波结果不满足该采样点的预设约束条件时，可以采用A80对该采样点的数据进行滤波处理，若A80对应的滤波结果满足预设约束条件，进一步地采用A80对该采样点的数据进行滤波处理；若A90对应的滤波结果不满足预设约束条件，电子设备可以进一步在A80和A90之间采用搜索满足预设约束条件的预设值，若A90对应的滤波结果满足预设约束条件，电子设备可以在A90和A100之间采用搜索满足预设约束条件的预设值，直至获得满足预设约束条件的结果的噪声越低的预设值，并将其确定为该采样点对应于第一滤波参数的备选值。
上述图像处理方法，电子设备在确定初始值的基础上，可以根据第一滤波参数与滤波结果的噪声之间的单调变化关系，快速获得每个采样点对应于第一滤波参数的备选值，从而使得每个采样点的滤波结果在满足预设约束条件的情况下，获得最佳的滤波效果。
图3为一个实施例中图像处理方法的流程示意图，本实施例涉及电子设备确定备选值的一种方式，在上述实施例的基础上，如图3所示，上述S101包括：
S301、针对待处理图像的每一个采样点，遍历第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值。
在确定备选值的另一种方式中，电子设备可以针对每个采样点，在第一滤波参数的预设值中，筛选出该采样点对应的滤波结果满足预设约束条件的多个预设值。具体地，第一滤波参数的预设值为多个离散值的情况下，电子设备可以针对待处理图像的每一个采样点，遍历第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值。第一滤波参数的预设值为预设值区间的情况下，电子设备可以以预设间隔在上述预设值区间中选择多个预设值，然后针对待处理图像的每一个采样点，遍历上述多个预设值，获得满足预设约束条件的N个预设值。
S302、确定N个预设值对应的滤波结果的噪声，从中选择噪声最低的预设值为采样点对应于第一滤波参数的备选值。
在上述步骤的基础上，电子设备可以进一步确定N个预设值对应的滤波结果的噪声，并从中选择噪声最低的预设值为采样点对应于第一滤波参数的备选值。
上述图像处理方法，预设滤波模型的滤波效果随第一滤波参数单调变化或非单调变化的情况下，电子设备均可以通过遍历各个预设值，并筛选出滤波结果满足预设约束条件的N个预设值，进一步通过对滤波效果的量化比较，获得满足预设约束条件下滤波效果最佳的第一滤波参数的值，进而可以获得更好的图像处理效果。
图4为一个实施例中图像处理方法的流程示意图，本实施例涉及电子设备确定备选值的一种方式，在上述实施例的基础上，如图4所示，上述S103包括：
S401、将与初始值不同的备选值对应的采样点，确定为控制点。
在获得各个采样点对应的备选值的基础上，电子设备可以根据备选值与初始值之间的关系，将与初始值不同的备选值对应的采样点确定为控制点。若预设滤波模型的第一滤波参数越大滤波效果越好，上述与初始值不同的备选值为小于初始值的备选值；预设滤波模型的第一滤波参数越小滤波效果越好，上述与初始值不同的备选值为大于初始值的备选值。
S402、保持控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线。
为了使得各个采样点的备选值可以连续变化，并且各个采样点的滤波结果均满足预设约束条件，电子设备可以保持控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线。上述曲线拟合处理可以是线性拟合处理，也可以是二次曲线拟合处理，在此不做限定。各个采样点的备选值可以如图5所示，横坐标表示采样点，纵坐标表示第一滤波参数的值；图中的不连续点为控制点，对图5中的备选值进行曲线拟合处理后，可以获得图6所示的参数拟合曲线。
电子设备获得的参数拟合曲线中，各个采样点的拟合值均位于第一滤波参数的预设值范围内。也就是说，若预设滤波模型的第一滤波参数越大滤波效果越好，上述各个采样点的拟合值中，控制点的拟合值与备选值相同，其余采样点的拟合值可以小于初始值，也可以等于初始值；若预设滤波模型的第一滤波参数越小滤波效果越好，上述各个采样点的拟合值中，控制点的拟合值与备选值相同，其余采样点的拟合值可以大于初始值，也可以等于初始值。
S403、将参数拟合曲线中各个采样点对应的拟合值，确定为采样点对应于对第一滤波参数的目标值。
在上述步骤的基础上，电子将参数拟合曲线中各个采样点对应的拟合值，确定为采样点对应于对第一滤波参数的目标值；然后根据多个采样点对应于对所述第一滤波参数的目标值对待处理图像进行滤波处理，输出所述待处理图像的处理结果。电子设备可以按照采样点的排列顺序，依次采用各采样点的目标值对该采样点的数据进行滤波处理，输出待处理图像的处理结果。
上述图像处理方法，电子设备采用上述曲线拟合方式获得各个采样点对应于第一滤波参数的目标值，既可以是各个采样点的滤波结果在满足预设约束条件下噪声最低，又可以使得待处理图像的图像处理结果更平滑，滤波效果更好，获得更好的图像显示效果。
应该理解的是，虽然图2-4的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2-4中的至少一部分步骤可以包括多个步骤或者多个阶段，这些步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤中的步骤或者阶段的至少一部分轮流或者交替地执行。
在一个实施例中，如图7所示，提供了一种图像处理装置，包括：
确定模块10，用于根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
处理模块20，用于对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
滤波模块30，用于根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
在一个实施例中，在上述实施例的基础上，通过预设滤波模型获得的滤波结果对应的噪声，随所述第一滤波参数的变化而单调变化。
在一个实施例中，在上述实施例的基础上，如图8所示，上述确定模块10包括：
第一确定单元101，用于根据滤波结果的噪声与所述第一滤波参数的单调变化关系，将最低噪声对应的预设值确定为所述第一滤波参数的初始值；
处理单元102，用于针对待处理图像的每一个采样点，采用初始值对采样点的数据进行滤波处理，获得采样点的初始处理结果；
第二确定单元103，用于在初始处理结果满足预设约束条件的情况下，将初始值确定为采样点对应于第一滤波参数的备选值；
搜索单元104，用于在初始处理结果不满足预设约束条件的情况下，在所述预设值中搜索备选值；所述备选值对应的滤波结果满足预设约束条件，且与所述初始值距离最近。
在一个实施例中，在上述实施例的基础上，上述搜索单元104具体用于：初始值的基础上，按照预设梯度调整初始值，直至采用调整后的初始值对应的滤波结果满足预设约束条件；将满足预设约束条件的滤波结果对应的预设值，确定为采样点对应于第一滤波参数的备选值。
在一个实施例中，在上述实施例的基础上，上述搜索单元104具体用于：以初始值为起点，在预设值中采用二分法搜索采样点对应于第一滤波参数的备选值。
在一个实施例中，在上述实施例的基础上，如图9所示，上述确定模块10包括：
遍历单元105，用于针对待处理图像的每一个采样点，遍历第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值；
选择单元106，用于确定N个预设值对应的滤波结果的噪声，从中选择噪声最低的预设值为采样点对应于第一滤波参数的备选值。
在一个实施例中，在上述实施例的基础上，如图10所示，上述处理模块20包括：
第三确定单元201，用于将与初始值不同的备选值对应的采样点，确定为控制点；
拟合单元202，用于保持控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线；
第四确定单元203，用于将参数拟合曲线中各个采样点对应的拟合值，确定为采样点对应于对第一滤波参数的目标值。
在一个实施例中，在上述实施例的基础上，预设滤波模型为高斯滤波模型，第一滤波参数为高斯滤波模型的滑动窗口或标准偏差。
上述提供的图像处理装置，可以执行上述图像处理方法实施例，其实现原理和技术效果类似，在此不再赘述。
关于图像处理装置的具体限定可以参见上文中对于图像处理方法的限定，在此不再赘述。上述图像处理装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于电子设备中的处理器中，也可以以软件形式存储于电子设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中，提供了一种电子设备，该电子设备可以是服务器，其内部结构图可以如图11所示。该电子设备包括通过系统总线连接的处理器、存储器和网络接口。其中，该电子设备的处理器用于提供计算和控制能力。该电子设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该电子设备的数据库用于存储数据处理数据。该电子设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种图像处理方法。
本领域技术人员可以理解，图11中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的电子设备的限定，具体的电子设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在一个实施例中，提供了一种电子设备，包括存储器和处理器，存储器中存储有计算机程序 ，该处理器执行计算机程序时实现以下步骤：
根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
在一个实施例中，通过预设滤波模型获得的滤波结果对应的噪声，随所述第一滤波参数的变化而单调变化。
在一个实施例中，处理器执行计算机程序时还实现以下步骤：根据滤波结果的噪声与所述第一滤波参数的单调变化关系，将最低噪声对应的预设值确定为所述第一滤波参数的初始值；针对待处理图像的每一个采样点，采用初始值对采样点的数据进行滤波处理，获得采样点的初始处理结果；若初始处理结果满足预设约束条件，则将初始值确定为采样点对应于第一滤波参数的备选值；若初始处理结果不满足预设约束条件，则在所述预设值中搜索备选值；所述备选值对应的滤波结果满足预设约束条件，且与所述初始值距离最近。
在一个实施例中，处理器执行计算机程序时还实现以下步骤：在初始值的基础上，按照预设梯度调整初始值，直至采用调整后的初始值对应的滤波结果满足预设约束条件；将满足预设约束条件的滤波结果对应的预设值，确定为采样点对应于第一滤波参数的备选值。
在一个实施例中，处理器执行计算机程序时还实现以下步骤：以初始值为起点，在预设值中采用二分法搜索采样点对应于第一滤波参数的备选值。
在一个实施例中，处理器执行计算机程序时还实现以下步骤：针对待处理图像的每一个采样点，遍历第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值；确定N个预设值对应的滤波结果的噪声，从中选择噪声最低的预设值为采样点对应于第一滤波参数的备选值。
在一个实施例中，处理器执行计算机程序时还实现以下步骤：将与初始值不同的备选值对应的采样点，确定为控制点；保持控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线；将参数拟合曲线中各个采样点对应的拟合值，确定为采样点对应于对第一滤波参数的目标值。
在一个实施例中，预设滤波模型为高斯滤波模型，第一滤波参数为高斯滤波模型的滑动窗口或标准偏差。
本实施例提供的电子设备，其实现原理和技术效果与上述方法实施例类似，在此不再赘述。
在一个实施例中，提供了一种计算机可读存储介质，其上存储有计算机程序，计算机程序被处理器执行时实现以下步骤：
根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定待处理图像的多个采样点对应于第一滤波参数的备选值；在采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
对多个采样点对应于第一滤波参数的备选值进行平滑处理，获得多个采样点对应于对第一滤波参数的目标值；
根据多个采样点对应于对第一滤波参数的目标值对待处理图像进行滤波处理，输出待处理图像的处理结果。
在一个实施例中，通过预设滤波模型获得的滤波结果对应的噪声，随所述第一滤波参数的变化而单调变化。
在一个实施例中，计算机程序被处理器执行时还实现以下步骤：根据滤波结果的噪声与所述第一滤波参数的单调变化关系，将最低噪声对应的预设值确定为所述第一滤波参数的初始值；针对待处理图像的每一个采样点，采用初始值对采样点的数据进行滤波处理，获得采样点的初始处理结果；若初始处理结果满足预设约束条件，则将初始值确定为采样点对应于第一滤波参数的备选值；若初始处理结果不满足预设约束条件，则在所述预设值中搜索备选值；所述备选值对应的滤波结果满足预设约束条件，且与所述初始值距离最近。
在一个实施例中，计算机程序被处理器执行时还实现以下步骤：在初始值的基础上，按照预设梯度调整初始值，直至采用调整后的初始值对应的滤波结果满足预设约束条件；将满足预设约束条件的滤波结果对应的预设值，确定为采样点对应于第一滤波参数的备选值。
在一个实施例中，计算机程序被处理器执行时还实现以下步骤：以初始值为起点，在预设值中采用二分法搜索采样点对应于第一滤波参数的备选值。
在一个实施例中，计算机程序被处理器执行时还实现以下步骤：针对待处理图像的每一个采样点，遍历第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值；确定N个预设值对应的滤波结果的噪声，从中选择噪声最低的预设值为采样点对应于第一滤波参数的备选值。
在一个实施例中，计算机程序被处理器执行时还实现以下步骤：将与初始值不同的备选值对应的采样点，确定为控制点；保持控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线；将参数拟合曲线中各个采样点对应的拟合值，确定为采样点对应于对第一滤波参数的目标值。
在一个实施例中，预设滤波模型为高斯滤波模型，第一滤波参数为高斯滤波模型的滑动窗口或标准偏差。
本实施例提供的计算机存储介质，其实现原理和技术效果与上述方法实施例类似，在此不再赘述。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和易失性存储器中的至少一种。非易失性存储器可包括只读存储器（Read-Only Memory，ROM）、磁带、软盘、闪存或光存储器等。易失性存储器可包括随机存取存储器（Random Access Memory，RAM）或外部高速缓冲存储器。作为说明而非局限，RAM可以是多种形式，比如静态随机存取存储器（Static Random Access Memory，SRAM）或动态随机存取存储器（Dynamic Random Access Memory，DRAM）等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (11)

  1. 一种图像处理方法，其特征在于，所述方法包括：
    根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定所述待处理图像的多个采样点对应于所述第一滤波参数的备选值；在所述采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，所述备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
    对所述多个采样点对应于所述第一滤波参数的备选值进行平滑处理，获得所述多个采样点对应于对所述第一滤波参数的目标值；
    根据所述多个采样点对应于对所述第一滤波参数的目标值对所述待处理图像进行滤波处理，输出所述待处理图像的处理结果。
  2. 根据权利要求1所述的方法，其特征在于，通过预设滤波模型获得的滤波结果对应的噪声，随所述第一滤波参数的变化而单调变化。
  3. 根据权利要求2所述的方法，其特征在于，所述根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定所述待处理图像的多个采样点对应于所述第一滤波参数的备选值，包括：
    根据滤波结果的噪声与所述第一滤波参数的单调变化关系，将最低噪声对应的预设值确定为所述第一滤波参数的初始值；
    针对所述待处理图像的每一个采样点，采用所述初始值对所述采样点的数据进行滤波处理，获得所述采样点的初始处理结果；
    若所述初始处理结果满足预设约束条件，则将所述初始值确定为所述采样点对应于所述第一滤波参数的备选值；
    若所述初始处理结果不满足预设约束条件，则在所述预设值中搜索备选值；所述备选值对应的滤波结果满足预设约束条件，且与所述初始值距离最近。
  4. 根据权利要求3所述的方法，其特征在于，所述在所述预设值中搜索备选值，包括：
    在所述初始值的基础上，按照预设梯度调整所述初始值，直至采用调整后的初始值对应的滤波结果满足所述预设约束条件；
    将满足所述预设约束条件的滤波结果对应的预设值，确定为所述采样点对应于所述第一滤波参数的备选值。
  5. 根据权利要求3所述的方法，其特征在于，所述在所述预设值中搜索备选值，包括：
    以所述初始值为起点，在所述预设值中采用二分法搜索所述采样点对应于所述第一滤波参数的备选值。
  6. 根据权利要求1所述的方法，其特征在于，所述根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定所述待处理图像的多个采样点对应于所述第一滤波参数的备选值，包括：
    针对所述待处理图像的每一个采样点，遍历所述第一滤波参数的所有预设值，获得满足预设约束条件的N个预设值；
    确定N个预设值对应的滤波结果的噪声，从中选择噪声最低的预设值为所述采样点对应于所述第一滤波参数的备选值。
  7. 根据权利要求3-5任一项所述的方法，其特征在于，所述对所述多个采样点对应于所述第一滤波参数的备选值进行平滑处理，获得所述多个采样点对应于对所述第一滤波参数的目标值，包括：
    将与所述初始值不同的备选值对应的采样点，确定为控制点；
    保持所述控制点的备选值不变，对其他采样点的备选值进行曲线拟合处理，获得参数拟合曲线；
    将所述参数拟合曲线中各个采样点对应的拟合值，确定为所述采样点对应于对所述第一滤波参数的目标值。
  8. 根据权利要求1-6任一项所述的方法，其特征在于，所述预设滤波模型为高斯滤波模型，所述第一滤波参数为所述高斯滤波模型的滑动窗口或标准偏差。
  9. 一种图像处理装置，其特征在于，所述装置包括：
    确定模块，用于根据第一滤波参数的预设值对待处理图像进行滤波处理，分别确定所述待处理图像的多个采样点对应于所述第一滤波参数的备选值；在所述采样点的滤波结果满足预设滤波模型的预设约束条件下对应的多个预设值中，所述备选值对应的滤波结果的噪声低于其他预设值对应的滤波结果的噪声；
    处理模块，用于对所述多个采样点对应于所述第一滤波参数的备选值进行平滑处理，获得所述多个采样点对应于对所述第一滤波参数的目标值；
    滤波模块，用于根据所述多个采样点对应于对所述第一滤波参数的目标值对所述待处理图像进行滤波处理，输出所述待处理图像的处理结果。
  10. 一种电子设备，包括存储器和处理器，所述存储器存储有计算机程序，其特征在于，所述处理器执行所述计算机程序时实现权利要求1至8中任一项所述的方法的步骤。
  11. 一种计算机可读存储介质，其上存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现权利要求1至8中任一项所述的方法的步骤。
PCT/CN2022/089154 2021-05-06 2022-04-26 图像处理方法、装置、电子设备和存储介质 WO2022233251A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110488365.3 2021-05-06
CN202110488365.3A CN115311147A (zh) 2021-05-06 2021-05-06 图像处理方法、装置、电子设备和存储介质

Publications (1)

Publication Number Publication Date
WO2022233251A1 true WO2022233251A1 (zh) 2022-11-10

Family

ID=83853809

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/089154 WO2022233251A1 (zh) 2021-05-06 2022-04-26 图像处理方法、装置、电子设备和存储介质

Country Status (2)

Country Link
CN (1) CN115311147A (zh)
WO (1) WO2022233251A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140086473A1 (en) * 2012-09-26 2014-03-27 Nidec Sankyo Corporation Image processing device, an image processing method and a program to be used to implement the image processing
CN103823219A (zh) * 2014-03-14 2014-05-28 中国科学院电子学研究所 自适应迭代的非局部干涉合成孔径雷达干涉相位滤波方法
CN110349080A (zh) * 2019-06-10 2019-10-18 北京迈格威科技有限公司 一种图像处理方法及装置
CN110956594A (zh) * 2019-11-27 2020-04-03 北京金山云网络技术有限公司 图像滤波方法、装置、电子设备及存储介质
CN111784733A (zh) * 2020-07-06 2020-10-16 深圳市安健科技股份有限公司 图像处理方法、装置、终端和计算机可读存储介质
CN112651902A (zh) * 2021-02-10 2021-04-13 江苏启阳半导体技术有限公司 一种图像处理中的多帧滤波实现方法及其系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140086473A1 (en) * 2012-09-26 2014-03-27 Nidec Sankyo Corporation Image processing device, an image processing method and a program to be used to implement the image processing
CN103823219A (zh) * 2014-03-14 2014-05-28 中国科学院电子学研究所 自适应迭代的非局部干涉合成孔径雷达干涉相位滤波方法
CN110349080A (zh) * 2019-06-10 2019-10-18 北京迈格威科技有限公司 一种图像处理方法及装置
CN110956594A (zh) * 2019-11-27 2020-04-03 北京金山云网络技术有限公司 图像滤波方法、装置、电子设备及存储介质
CN111784733A (zh) * 2020-07-06 2020-10-16 深圳市安健科技股份有限公司 图像处理方法、装置、终端和计算机可读存储介质
CN112651902A (zh) * 2021-02-10 2021-04-13 江苏启阳半导体技术有限公司 一种图像处理中的多帧滤波实现方法及其系统

Also Published As

Publication number Publication date
CN115311147A (zh) 2022-11-08

Similar Documents

Publication Publication Date Title
US20240104378A1 (en) Dynamic quantization of neural networks
US10187627B2 (en) Multi-lens based capturing apparatus and method
US20150043826A1 (en) Image processing apparatus, image processing method, and program
WO2019085709A1 (zh) 一种应用于卷积神经网络的池化处理的方法及系统
EP2165311A1 (en) Information processing method and apparatus, program, and storage medium
US20210150733A1 (en) Data processing method, apparatus, and system
KR20220059194A (ko) 대상 객체에 적응적인 객체 추적 방법 및 장치
US10318816B2 (en) Method, system and apparatus for segmenting an image set to generate a plurality of event clusters
Song et al. Dual alternating direction method of multipliers for inverse imaging
EP3971781A1 (en) Method and apparatus with neural network operation
CN115705651A (zh) 视频运动估计方法、装置、设备和计算机可读存储介质
WO2022233251A1 (zh) 图像处理方法、装置、电子设备和存储介质
CN110856014B (zh) 动态图像生成方法、装置、电子设备及存储介质
CN110689565B (zh) 一种深度图确定的方法、装置及电子设备
WO2023207899A1 (zh) 视频的防抖处理方法、装置、存储介质及电子设备
CN111225144A (zh) 视频拍摄方法、装置、电子设备及计算机存储介质
WO2023010701A1 (en) Image generation method, apparatus, and electronic device
Wu et al. Video stabilisation with total warping variation model
CN112804444B (zh) 一种视频处理方法、装置、计算设备及存储介质
KR102240403B1 (ko) 이미지 평활화 방법 및 이미지 평활화 장치
CN113538269A (zh) 图像处理方法及装置、计算机可读存储介质和电子设备
CN111143589A (zh) 一种图像处理方法及装置、存储介质
WO2022213372A1 (zh) 图像去雾方法、装置、电子设备和计算机可读介质
CN112565595B (zh) 图像抖动消除方法、装置、电子设备和存储介质
US20240161230A1 (en) Method and apparatus for processing image based on neural network

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: 22798592

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: 22798592

Country of ref document: EP

Kind code of ref document: A1