WO2022183902A1 - 图像清晰度确定方法、装置、设备及存储介质 - Google Patents
图像清晰度确定方法、装置、设备及存储介质 Download PDFInfo
- Publication number
- WO2022183902A1 WO2022183902A1 PCT/CN2022/076274 CN2022076274W WO2022183902A1 WO 2022183902 A1 WO2022183902 A1 WO 2022183902A1 CN 2022076274 W CN2022076274 W CN 2022076274W WO 2022183902 A1 WO2022183902 A1 WO 2022183902A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- dark channel
- preset
- target image
- sharpness
- evaluation index
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000011156 evaluation Methods 0.000 claims abstract description 63
- 238000004590 computer program Methods 0.000 claims description 11
- 238000010801 machine learning Methods 0.000 claims description 5
- 230000010354 integration Effects 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 4
- 238000011109 contamination Methods 0.000 description 3
- 239000000428 dust Substances 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 239000004519 grease Substances 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000002834 transmittance Methods 0.000 description 2
- 241001156002 Anthonomus pomorum Species 0.000 description 1
- 230000003749 cleanliness Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Definitions
- the present application relates to the field of computer technology, for example, to a method, apparatus, device, and storage medium for determining image sharpness.
- the sharpness of the image is an important indicator to measure the quality of the image, which can better correspond to the subjective feeling of people. Valid information, and generally does not meet the user's aesthetic requirements.
- the present application provides an image sharpness determination method, apparatus, device, and storage medium, which can optimize the image sharpness determination scheme in the related art.
- a method for determining image sharpness comprising:
- evaluation index information corresponding to the target image based on the dark channel value wherein the evaluation index information includes at least one of dark channel integral information, exposure degree information and brightness information;
- the evaluation index information is input into a preset sharpness determination model, and the sharpness of the target image is determined according to an output result of the preset sharpness determination model.
- a device for determining image clarity comprising:
- the dark channel value acquisition module is set to acquire the dark channel value corresponding to each pixel in the target image
- An evaluation index information determination module configured to determine evaluation index information corresponding to the target image based on the dark channel value, wherein the evaluation index information includes at least one of dark channel integral information, exposure degree information and brightness information;
- the sharpness determination module is configured to input the evaluation index information into a preset sharpness determination model, and determine the sharpness of the target image according to an output result of the preset sharpness determination model.
- a computer device including a memory, a processor, and a computer program stored on the memory and running on the processor, and when the processor executes the computer program, the image definition provided by the embodiments of the present application is realized. Determine the method.
- a computer-readable storage medium is also provided, on which a computer program is stored, and when the computer program is executed by a processor, implements the image definition determination method provided by the embodiments of the present application.
- FIG. 1 is a schematic flowchart of a method for determining image clarity provided by an embodiment of the present application
- FIG. 2 is a schematic flowchart of another method for determining image clarity provided by an embodiment of the present application.
- Fig. 3 is a kind of image comparison schematic diagram provided by the embodiment of this application.
- FIG. 4 is a schematic flowchart of another image sharpness determination method provided by an embodiment of the present application.
- FIG. 5 is a schematic diagram of a screen interface provided by an embodiment of the present application.
- FIG. 6 is a structural block diagram of an apparatus for determining image definition provided by an embodiment of the present application.
- FIG. 7 is a structural block diagram of a computer device according to an embodiment of the present application.
- FIG. 1 is a schematic flowchart of a method for determining image sharpness provided by an embodiment of the present application.
- the method may be executed by an apparatus for determining image sharpness, where the apparatus may be implemented by software and/or hardware, and may be integrated in computer equipment.
- the method includes the following operations.
- Step 101 Obtain the dark channel value corresponding to each pixel in the target image.
- the source of the target image is not limited, and it can come from an image collected by an image acquisition device such as a camera (it can be a photo generated after shooting, a preview image or a cached image, etc.), or it can come from a computer device Locally stored images, but also from images acquired over the network.
- the above images from different sources can be called initial images
- the target image can contain all the image content in the initial image (for example, the initial image is used as the target image), and the target image can also contain part of the image content in the initial image (for example, for the initial image).
- the region of interest in the image is intercepted to obtain the target image, and the location of the region of interest can be set according to actual needs).
- the dark channel refers to a grayscale image composed of the minimum value of the red, green and blue (RGB) three-channel values of the image.
- the dark channel is a basic assumption that when there are objects or surfaces with darker color in the image, their color channel should have a very low value, and the corresponding dark channel value should be low.
- the target image contains multiple pixels, and each pixel can correspond to a set of RGB three-channel values.
- the RGB three-channel values represent the size of the red, green, and blue color components of the pixel, and the dark channel value can refer to RGB.
- the dark channel value corresponding to the current pixel may be determined respectively, and then the dark channel value corresponding to all pixels in the target image may be obtained.
- Step 102 Determine evaluation index information corresponding to the target image based on the dark channel value, wherein the evaluation index information includes at least one of dark channel integral information, exposure degree information, and brightness information.
- the value of the dark channel corresponding to the darker object or surface in the image should be lower, and for an image, its sharpness may be affected by the object being photographed. , shooting environment and shooting equipment and other influences, when the color of the shooting object is dark, the shooting environment is foggy, etc., and the camera is dirty, etc., the image will not be clear enough. After research, it is found that the shooting environment is foggy and camera. In the image taken under conditions such as dirt, the values of the dark channel corresponding to the foggy part and the part corresponding to the dirty part are also low. Therefore, the unclear part can be characterized based on the dark channel value, and the dark channel After the corresponding operation is performed on the value, the obtained information can be used as the evaluation index information of the image.
- the evaluation index information includes at least one of dark channel integral information, exposure level information, and brightness information.
- the evaluation index information can include any two or three of the above, and other information related to the dark channel value can also be added as the evaluation index information (such as the minimum dark channel value, etc.), other information unrelated to the dark channel value can also be added as evaluation index information (for example, converting RGB format to YUV format, where Y represents brightness, U and V represent chroma, and Y in this format The value of the channel determines the brightness information), which is not limited.
- the dark channel integral information may include the number of pixels included in a certain preset interval of one or more dark channel values;
- the exposure degree information may include the exposure degree represented by the dark channel value, such as underexposure. , normal exposure or overexposure, etc., the representation method is not limited, and an appropriate calculation method can be selected according to the actual situation;
- the brightness information can include the image brightness represented by the dark channel value, and the representation method is not limited, and the appropriate calculation method can be selected according to the actual situation. calculation method.
- Step 103 Input the evaluation index information into a preset sharpness determination model, and determine the sharpness of the target image according to the output result of the preset sharpness determination model.
- a training sample set can be prepared in advance, the training sample set can contain a large number of images, the clarity of each image is marked as a sample label by means of manual annotation, and the preset model is trained by using the training sample set, and then the result is obtained.
- a preset sharpness determination model In this step, the evaluation index information corresponding to the target image is input into the preset sharpness determination model, so as to determine the sharpness of the target image.
- the preset model may be a preset machine learning model or other models, and the structure of the model and weight parameters are not limited, and may be selected according to actual conditions.
- the sample label can be a sharpness score, and the range can be freely set, such as an integer from 0 to 10, 0 means the least clear, that is, the lowest level of clarity, 10 means the clearest, that is, the highest level of clarity.
- the image definition determination method acquires the dark channel value corresponding to each pixel in the target image, and determines the evaluation index information corresponding to the target image based on the dark channel value, wherein the evaluation index information includes dark channel integral information, For at least one of exposure degree information and brightness information, the evaluation index information is input into a preset sharpness determination model, and the sharpness of the target image is determined according to an output result of the preset sharpness determination model.
- the evaluation index information can be reasonably and accurately determined according to the dark channel value corresponding to each pixel, and the evaluation index information can be input into the corresponding preset sharpness determination After the model, accurate sharpness evaluation results can be obtained, thereby improving the accuracy of the image sharpness determination scheme.
- the determining the evaluation index information corresponding to the target image based on the dark channel value includes: for at least two first preset dark channels For each first preset dark channel value interval in the value interval, count the number of pixels included in the current first preset dark channel value interval, and obtain the dark channel integral corresponding to the current first preset dark channel value interval value; the dark channel integration information corresponding to the target image is determined according to at least two dark channel integration values.
- different dark channel value intervals can reflect the brightness and darkness of the image, and the total number of pixels in multiple intervals can reflect the distribution of areas with different brightness and darkness in the image, which is conducive to accurate image clarity. Evaluate.
- the number of the first preset dark channel value intervals and the interval range of each first preset dark channel value interval can be set according to actual conditions.
- the at least two dark channel integral values may refer to at least two selected from all the calculated dark channel integral values, such as selecting at least two maximum values; the at least two dark channel integral values may refer to all the calculated dark channel integral values, That is, the dark channel integral values corresponding to the at least two first preset dark channel value intervals respectively.
- the union of at least two first preset dark channel value intervals is 0 to 255.
- the determining the evaluation index information corresponding to the target image based on the dark channel value includes: calculating a second preset dark channel value interval including The sum of the dark channel values corresponding to the pixels of The second preset dark channel value interval is used to indicate overexposure; the exposure level information corresponding to the target image is determined according to the quotient of the sum and the total. In this way, the overexposure will affect the clarity of the image. By studying the dark channel value corresponding to the overexposed area, the second preset dark channel value interval can be reasonably determined. The average value of the dark channel values of the pixel points can accurately characterize the severity of the overexposure situation, which in turn facilitates accurate evaluation of image sharpness.
- the quotient of the sum and the total number may be used as the exposure level information, or other operations, such as normalization operations, may be performed on the basis of the quotient, and the operation result may be used as the exposure level information corresponding to the target image.
- the determining the evaluation index information corresponding to the target image based on the dark channel value includes: calculating a median or an average value of all dark channel values ; Determine the brightness information corresponding to the target image according to the median or the average value.
- the preset definition determination model is obtained by acquiring a training sample set, wherein each training sample in the training sample set includes training sample data and a training sample label, and the training sample data
- the training sample image and the evaluation index information corresponding to the training sample image are included, and the training sample label includes the sharpness score corresponding to the training sample image;
- the preset machine learning model is trained by using the training sample set, and the corresponding preset sharpness score is obtained. degree determination model. In this way, when labeling the training samples, the score is used to quantify the sharpness.
- the output result of the preset sharpness determination model is closer to the corresponding sharpness score when the sample is labeled, so as to be able to Get the sharpness score of the target image more accurately.
- the method before acquiring the dark channel value corresponding to each pixel in the target image, the method further includes: acquiring the target image collected by the camera.
- the sharpness of the image collected by the camera can be evaluated in real time, which is beneficial to the device where the camera is located or the user to adjust the device parameters, camera parameters or the state of the camera in time according to the sharpness evaluation result, so as to capture a clearer image. .
- the solution in the embodiment of this application can be used to remind the user to wipe the camera and improve the light transmittance of the lens. , to improve the picture quality.
- the method further includes: if it is determined that the camera is currently in a dirty state according to the sharpness of the target image , the dirty reminder operation is performed.
- the clarity of the image captured by the camera can reflect the cleanliness of the lens to a certain extent. If the clarity is poor, it may be that the lens is dirty with finger grease or dust. It can be reminded in time, which is beneficial to the device. Or the user can take corresponding measures in time to remove the contamination and improve the image quality.
- the form of performing the contamination reminder operation is not limited.
- the dirty reminder instruction can be output; if the reminder object is the user, the device where the camera is located can be controlled to remind the user in a preset reminder mode, such as displaying the camera in a dirty state on the display screen.
- Reminder text or reminder icon, etc. for example, voice reminder.
- the determining the sharpness of the target image according to the output result of the preset sharpness determination model includes: determining the sharpness of the target image according to the output result of the preset sharpness determining model determining that the camera is currently in a dirty state according to the sharpness of the target image, including: when the sharpness score of the target image is lower than a preset score threshold, determining that the camera is currently dirty state.
- the preset scoring threshold is determined according to the training sample labels in the training sample set corresponding to the preset definition determination model. With this arrangement, a threshold value for determining whether or not it is in a dirty state can be appropriately set.
- the manner of determining the preset score threshold is not limited, for example, it may be the average or median of the labels of the training samples, and so on.
- FIG. 2 is a schematic flowchart of another method for determining image clarity provided by an embodiment of the present application, which can be applied to a scene in which the clarity of a captured image is insufficient due to a dirty state of the camera, as shown in FIG. 2 ,
- the method may include the following operations.
- Step 201 Acquire a target image captured by a camera.
- Step 202 Obtain the dark channel value corresponding to each pixel in the target image.
- I C represents the target image in the RGB data format
- C represents the RGB three-channel
- I dark represents the corresponding dark channel
- the dark channel is the grayscale image of the minimum value in the RGB three channels of the target image. It can be assumed that the area with a value close to 0 (which can be considered as black) in the dark channel has no white fog caused by dirt, and the area with a value close to 255 (which can be considered as white) contains white fog. For a pixel, the corresponding dark channel value can be denoted as d.
- Step 203 For each first preset dark channel value interval in the at least two first preset dark channel value intervals, count the number of pixels contained in the current first preset dark channel value interval to obtain the current first dark channel value interval.
- the dark channel integral value may also be referred to as a dark channel integral map, and the calculation method is as follows:
- FIG. 3 is a schematic diagram of image comparison provided by an embodiment of the present application.
- the leftmost image represents the target image
- the middle image represents the grayscale image corresponding to the dark channel
- the rightmost image represents the corresponding dark channel integral map Image.
- the dark area, normal exposure area, highlight area and white fog area are filled with different colors respectively, and the distribution of areas with different light and dark degrees of the target image can be seen.
- Part of the white fog area is circled.
- Step 204 Calculate the average value of the dark channel values corresponding to the pixels included in the second preset dark channel value interval, and normalize the average value to obtain exposure degree information corresponding to the target image.
- the second preset dark channel value interval is used to represent overexposure, which can be set according to actual conditions, for example, it can be [250, 255).
- Exposure rate can be calculated using the following expression:
- ⁇ x ⁇ [s,t) d indicates that the dark channel value in the grayscale image I dark belongs to the sum of the dark channel values of the pixel x between the value range [s, t), and N indicates that the dark channel value belongs to the value range
- the number of pixels x between [s, t), [s, t) can be [250, 255).
- Step 205 Calculate the median of all dark channel values to obtain brightness information corresponding to the target image.
- the median of the dark channel values can be regarded as the average brightness of the target image, that is, the brightness information corresponding to the target image.
- Step 206 Input all dark channel integral values, exposure level information, and brightness information into a preset sharpness determination model, and determine a sharpness score of the target image according to the output result of the preset sharpness determination model.
- training sample images may be collected, and the training sample images may be manually marked, that is, the training sample images may be scored for clarity, and the score value range is [0, 10].
- Calculate the dark channel integral map, overexposure rate and average brightness for the training sample images take ⁇ dark channel integral map, overexposure rate, average brightness, sharpness score ⁇ as the training sample set, and input it into the machine learning model for training, Get the preset sharpness determination model.
- Step 207 If it is determined that the camera is currently in a dirty state according to the sharpness score of the target image, perform a dirty reminder operation.
- a target image captured by a camera is acquired, a corresponding dark channel integral map, overexposure rate and average brightness are calculated according to the dark channel value of each pixel in the target image, and input to the preset
- the sharpness score of the target image can be quickly and accurately obtained, and then it can be judged whether the camera is in a dirty state according to the sharpness score.
- FIG. 4 is a schematic flowchart of another method for determining image clarity provided by an embodiment of the present application, which can be applied to various scenarios for identifying and reminding the dirty state of a camera, such as a photo-taking scenario, a video-taking scenario, and a live-streaming scenario. As shown in FIG. 4, the method may include the following operations.
- Step 401 Determine whether the camera of the mobile terminal is turned on, if yes, execute step 402; otherwise, execute step 401 repeatedly.
- Step 402 judging whether a dirty reminder has appeared on the screen of the mobile terminal, if yes, end the process; otherwise, go to step 403 .
- Step 403 Acquire the target image collected by the camera, and determine the dark channel value corresponding to each pixel in the target image.
- the target image here can be, for example, a preview image in a photographing scene, or a real-time image in a photographing scene and a live broadcast scene, and the like.
- Step 404 Calculate a preset number of dark channel integral values, overexposure rate and average brightness according to the dark channel value.
- Step 405 Input all the dark channel integral values, overexposure ratio and average brightness into the preset sharpness determination model, and determine the sharpness score of the target image according to the output result of the preset sharpness determination model.
- Step 406 determine whether the mobile terminal is in a shooting state, if yes, go to step 407 ; otherwise, return to go to step 403 .
- Step 403 may be executed after a preset time interval, so as to reduce the power consumption of the mobile terminal.
- Step 407 Determine whether the sharpness score is less than the preset score threshold, if so, go to Step 408; otherwise, return to Step 403.
- the preset scoring threshold may be determined according to the average value of the training sample labels in the training sample set corresponding to the preset definition determination model.
- Step 408 displaying a dirty reminder on the screen.
- FIG. 5 is a schematic diagram of a screen interface provided by an embodiment of the present application.
- the prompt text may be, for example, "Your lens is dirty, please wipe it and continue shooting".
- the solution provided by the embodiment of the present application can detect that the camera is dirty, and notify the user to wipe the camera in the form of a pop-up window, so as to solve the problem of image quality loss caused by the dirty camera.
- only screen brightness is used as an evaluation index, and it is greatly dependent on hardware devices, and can only be applied to a specific mobile terminal, requiring more computing resources.
- the solution adopted in the embodiment of the present application uses the dark channel integral map, the overexposure rate, and the screen brightness level as evaluation indicators, which can improve the accuracy of the contamination reminder, reduce the false detection rate and missed detection rate of the reminder, and can be applied
- the device hardware requirements are relatively low, no interaction with hardware is required, and more computing resources are not required, and the application range is wider.
- FIG. 6 is a structural block diagram of an apparatus for determining image sharpness provided by an embodiment of the present application.
- the apparatus can be implemented by software and/or hardware, and can generally be integrated in computer equipment.
- the image sharpness can be determined by executing a method for determining image sharpness. Spend.
- the device includes the following modules.
- the dark channel value obtaining module 601 is set to obtain the dark channel value corresponding to each pixel in the target image
- the evaluation index information determination module 602 is configured to determine the evaluation index information corresponding to the target image based on the dark channel value, wherein the evaluation index information includes at least one of dark channel integral information, exposure degree information and brightness information ;
- the sharpness determination module 603 is configured to input the evaluation index information into a preset sharpness determination model, and determine the sharpness of the target image according to the output result of the preset sharpness determination model.
- the image clarity determination device acquires the dark channel value corresponding to each pixel in the target image, and determines the evaluation index information corresponding to the target image based on the dark channel value, wherein the evaluation index information includes dark channel integral information, exposure At least one of the degree information and the brightness information, the evaluation index information is input into the preset sharpness determination model, and the sharpness of the target image is determined according to the output result of the preset sharpness determination model.
- the evaluation index information can be reasonably and accurately determined according to the dark channel value corresponding to each pixel, and the evaluation index information can be input into the corresponding preset sharpness determination After the model, accurate sharpness evaluation results can be obtained, thereby improving the accuracy of the image sharpness determination scheme.
- FIG. 7 is a structural block diagram of a computer device according to an embodiment of the present application.
- the computer device 700 includes a memory 701, a processor 702, and a computer program that is stored in the memory 701 and can run on the processor 702.
- the processor 702 executes the computer program, the image definition determination provided by the embodiments of the present application is implemented. method.
- the embodiments of the present application further provide a storage medium containing computer-executable instructions, where the computer-executable instructions, when executed by a computer processor, are used to execute the image definition determination method provided by the embodiments of the present application.
- the storage medium may be a non-transitory storage medium.
- the image sharpness determination apparatus, device, and storage medium provided in the above embodiments can execute the image sharpness determination method provided by any embodiment of the present application, and have corresponding functional modules and effects for implementing the method.
- the method for determining image sharpness provided by any embodiment of the present application.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Multimedia (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Image Analysis (AREA)
- Studio Devices (AREA)
Abstract
本申请提供了图像清晰度确定方法、装置、设备及存储介质。其中,该图像清晰度确定方法包括:获取目标图像中每个像素对应的暗通道值;基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种;将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
Description
本申请要求在2021年03月02日提交中国专利局、申请号为202110231007.4的中国专利申请的优先权,该申请的全部内容通过引用结合在本申请中。
本申请涉及计算机技术领域,例如涉及图像清晰度确定方法、装置、设备及存储介质。
图像的清晰度是衡量图像质量优劣的重要指标,它能够较好的与人的主观感受相对应,若图像的清晰度较低,则用户的主观感受是图像比较模糊,难以获取图像中的有效信息,且一般不符合用户的审美要求。
存在很多场景需要对图像的清晰度进行评价,相继也出现了很多用于确定图像清晰度的方案,较为常用的包括基于多种函数的确定方案,如布伦纳(Brenner)梯度函数、灰度方差函数以及能量梯度函数等等。然而,相关技术中的方案确定方式通常比较简单,准确度有待提高。
发明内容
本申请提供了图像清晰度确定方法、装置、设备及存储介质,可以优化相关技术中的图像清晰度确定方案。
提供了一种图像清晰度确定方法,该方法包括:
获取目标图像中每个像素对应的暗通道值;
基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种;
将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
还提供了一种图像清晰度确定装置,该装置包括:
暗通道值获取模块,设置为获取目标图像中每个像素对应的暗通道值;
评估指标信息确定模块,设置为基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种;
清晰度确定模块,设置为将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
还提供了一种计算机设备,包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序,所述处理器执行所述计算机程序时实现如本申请实施例提供的图像清晰度确定方法。
还提供了一种计算机可读存储介质,其上存储有计算机程序,该计算机程序被处理器执行时实现如本申请实施例提供的图像清晰度确定方法。
图1为本申请实施例提供的一种图像清晰度确定方法的流程示意图;
图2为本申请实施例提供的又一种图像清晰度确定方法的流程示意图;
图3为本申请实施例提供的一种图像对比示意图;
图4为本申请实施例提供的另一种图像清晰度确定方法的流程示意图;
图5为本申请实施例提供的一种屏幕界面示意图;
图6为本申请实施例提供的一种图像清晰度确定装置的结构框图;
图7为本申请实施例提供的一种计算机设备的结构框图。
下面结合附图和实施例对本申请进行说明。此处所描述的实施例仅仅用于解释本申请,而非对本申请的限定。为了便于描述,附图中仅示出了与本申请相关的部分而非全部结构。此外,在不冲突的情况下,本申请中的实施例及实施例中的特征可以相互组合。
图1为本申请实施例提供的一种图像清晰度确定方法的流程示意图,该方法可以由图像清晰度确定装置执行,其中该装置可由软件和/或硬件实现,可集成在计算机设备中。如图1所示,该方法包括如下操作。
步骤101、获取目标图像中每个像素对应的暗通道值。
本申请实施例中,对目标图像的来源不做限定,可以来自摄像头等图像采集装置采集的图像(可以是拍摄后生成的照片,也可以是预览图像或缓存图像等),也可以来自计算机设备本地存储的图像,还可以来自通过网络获取的图像。可以将上述不同来源的图像称为初始图像,目标图像可以包含初始图像中的全部图像内容(例如将初始图像作为目标图像),目标图像还可以包含初始图像中的部分图像内容(例如对初始图像中的感兴趣区域进行截取,得到目标 图像,感兴趣区域所处位置可以根据实际需求设置)。
暗通道指按图像的红绿蓝(Red Green Blue,RGB)三通道值中取得最小值构成的灰度图。暗通道是一个基本假设,这个假设认为,图像中存在颜色较暗的物体或者表面时,它们的颜色通道中应具有一个很低的值,对应的暗通道的值应是较低的。
目标图像中包含多个像素,每一个像素都可以对应一组RGB三通道值,RGB三通道值分别表示了像素的红、绿、蓝等三个颜色的分量的大小,暗通道值可以指RGB三个通道值中的最小通道值。例如,对于像素A,R通道值为20,G通道值为30,B通道值为50,则像素A对应的暗通道值为R通道值的数值,即20;对于像素B,R通道值为60,G通道值为30,B通道值为50,则像素B对应的暗通道值为G通道值的数值,即30。本步骤中,可以针对目标图像中的每个像素,分别确定当前像素对应的暗通道值,进而得到目标图像中所有像素分别对应的暗通道值。
步骤102、基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种。
示例性的,基于上述暗通道的假设,图像中存在的颜色较暗的物体或者表面对应的暗通道的值应是较低的,而对于一张图像来说,其清晰度可能会受到拍摄物体、拍摄环境以及拍摄设备等多方面的影响,当拍摄物体颜色较暗时、拍摄环境有雾等、以及摄像头存在脏污等,均会导致图像不够清晰,经过研究发现,拍摄环境有雾以及摄像头存在脏污等情况下拍摄的图像中,对应雾的部分和对应脏污的部分的暗通道的值也是较低的,因此,可以基于暗通道值来对不清晰的部分进行表征,对暗通道值进行相应的运算后,得到的信息可作为图像的评估指标信息。
示例性的,评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种。为了提升评估指标的全面性,使得评估结果更加准确,评估指标信息可以包括上述任意两种或包括上述三种,还可以增加其他的与暗通道值相关的信息作为评估指标信息(如最小暗通道值等),也可以增加其他的与暗通道值无关的信息作为评估指标信息(例如将RGB格式转变成YUV格式,其中,Y表示明亮度,U和V表示色度,以该格式中的Y通道的值确定亮度信息),不做限定。
示例性的,暗通道积分信息可以包括在一定的预先设置的一个或多个暗通道值区间内包含的像素点的数量;曝光程度信息可以包括利用暗通道值来表征的曝光程度,如曝光不足、曝光正常或曝光过度等,表征方式不做限定,可以 根据实际情况选择适当的计算方式;亮度信息可以包括利用暗通道值来表征的图像亮度,表征方式不做限定,可以根据实际情况选择适当的计算方式。
步骤103、将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
示例性的,可以预先准备训练样本集合,训练样本集合可以包含大量的图像,采用人工标注等方式标注每个图像的清晰度作为样本标签,并利用训练样本集合对预设模型进行训练,进而得到预设清晰度确定模型,在本步骤中,将目标图像对应的评估指标信息输入至预设清晰度确定模型,进而确定目标图像的清晰度。其中,预设模型可以是预设机器学习模型,也可以是其他模型,模型的结构以及权重参数等不做限定,可根据实际情况进行选取。其中,样本标签可以是清晰度评分,范围可以自由设置,如0至10中的整数,0表示最不清晰,也即清晰度最低等级,10表示最清晰,也即清晰度最高等级。
本申请实施例中提供的图像清晰度确定方法,获取目标图像中每个像素对应的暗通道值,基于暗通道值确定目标图像对应的评估指标信息,其中,评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种,将评估指标信息输入至预设清晰度确定模型,并根据预设清晰度确定模型的输出结果确定所述目标图像的清晰度。通过采用上述技术方案,针对需要进行清晰度评估的目标图像,可以根据其中每个像素对应的暗通道值来合理准确地确定评估指标信息,在将评估指标信息输入到相应的预设清晰度确定模型后,可以得出准确的清晰度评估结果,从而提升图像清晰度确定方案的准确性。
在一些实施例中,当所述评估指标信息包括暗通道积分信息时,所述基于所述暗通道值确定所述目标图像对应的评估指标信息,包括:针对至少两个第一预设暗通道值区间中的每个第一预设暗通道值区间,统计当前第一预设暗通道值区间内包含的像素点的数量,得到所述当前第一预设暗通道值区间对应的暗通道积分值;根据至少两个暗通道积分值确定所述目标图像对应的暗通道积分信息。这样设置,不同暗通道值区间可以体现出图像的明暗程度等情况,多个区间内的像素点总数量可以体现出图像的不同明暗程度的区域的分布情况,可有利于对图像清晰度进行准确评估。其中,第一预设暗通道值区间的数量以及每个第一预设暗通道值区间的区间范围可以根据实际情况进行设置。至少两个暗通道积分值可以指从所有计算得到的暗通道积分值中选取至少两个,例如选取至少两个最大值;至少两个暗通道积分值可以指所有计算得到的暗通道积分值,也即至少两个第一预设暗通道值区间分别对应的暗通道积分值。可选的,所述至少两个第一预设暗通道值区间中的任意两个第一预设暗通道值区间不存在交集。至少两个第一预设暗通道值区间的并集为0至255。
在一些实施例中,当所述评估指标信息包括曝光程度信息时,所述基于所述暗通道值确定所述目标图像对应的评估指标信息,包括:计算第二预设暗通道值区间内包含的像素点对应的暗通道值的总和,其中,所述第二预设暗通道值区间用于表示曝光过度;统计所述第二预设暗通道值区间内包含的像素点的总数,其中,所述第二预设暗通道值区间用于表示曝光过度;根据所述总和与所述总数的商确定所述目标图像对应的曝光程度信息。这样设置,曝光过度的情况会影响图像的清晰度,通过研究曝光过度区域对应的暗通道值可以合理地确定出第二预设暗通道值区间,利用第二预设暗通道值区间内包含的像素点的暗通道值的均值,可以准确地表征曝光过度情况的严重程度,进而有利于对图像清晰度进行准确评估。其中,可以将总和与总数的商作为曝光程度信息,也可以在该商的基础上进行其他运算,如归一化运算,并将运算结果作为目标图像对应的曝光程度信息。
在一些实施例中,当所述评估指标信息包括亮度信息时,所述基于所述暗通道值确定所述目标图像对应的评估指标信息,包括:计算所有暗通道值的中位数或平均值;根据所述中位数或所述平均值确定所述目标图像对应的亮度信息。这样设置,可以更加快速准确地计算目标图像对应的亮度信息,进而有利于对图像清晰度进行准确评估。
在一些实施例中,所述预设清晰度确定模型通过以下方式得到:获取训练样本集,其中,所述训练样本集中的每个训练样本包括训练样本数据和训练样本标签,所述训练样本数据包括训练样本图像和训练样本图像对应的评估指标信息,所述训练样本标签包括训练样本图像对应的清晰度评分;利用所述训练样本集对预设机器学习模型进行训练,得到相应的预设清晰度确定模型。这样设置,在进行训练样本标注时,采用分数对清晰度进行量化,通过对机器学习模型的训练,使得预设清晰度确定模型的输出结果更加接近于样本标注时对应的清晰度评分,从而能够更加准确地得到目标图像的清晰度评分结果。
在一些实施例中,在所述获取目标图像中每个像素对应的暗通道值之前,还包括:获取摄像头采集的目标图像。这样设置,可以对摄像头采集的图像的清晰度进行实时评估,有利于摄像头所在设备或用户根据清晰度评估结果来及时调整设备参数、摄像头参数或摄像头所处状态等,进而拍摄出更加清晰的图像。
随着携带摄像头的手机等移动终端的广泛普及,拍摄记录生活点滴已经成为了人们生活的一部分,人们也逐渐将摄像头画质好坏纳入到评价移动终端优劣的指标中。以手机为例,由于日常使用手机的频率极高,用户极易触摸到手机摄像头,将手指的油脂、灰尘等附着在摄像头表面,当人们使用相机程序时, 其表面污渍会降低镜片透光率并产生散射,使得拍摄出来的图像出现白雾,造成画质损失,但用户往往无法将画质损失与污渍联系到一起,而是将画质损失归咎于手机摄像头存在质量问题。因误触摄像头产生的灰尘、油脂等污渍,降低了用户的使用体验,为用户与手机厂商带来了困扰,因此本申请实施例中的方案可以用来提醒用户擦拭摄像头,改善镜片透光率,以此改善拍摄画质。
在一些实施例中,在根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度之后,还包括:若根据所述目标图像的清晰度确定所述摄像头当前处于脏污状态,则进行脏污提醒操作。这样设置,摄像头拍摄的图像的清晰度可以在一定程度上体现出镜头的清洁程度,若清晰度较差,则可能是镜头附着有手指油脂或灰尘等脏污,可以及时进行提醒,有利于设备或用户及时采取相应的措施来消除脏污,提高图像拍摄质量。其中,进行脏污提醒操作的形式不做限定。若提醒对象为摄像头所在设备,则可以输出脏污提醒指令;若提醒对象为用户,则可以控制摄像头所在设备采用预设提醒方式对用户进行提醒,如在显示屏上显示摄像头处于脏污状态的提醒文字或提醒图标等,又如进行语音提醒等。
在一些实施例中,所述根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度,包括:根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度评分;所述根据所述目标图像的清晰度确定所述摄像头当前处于脏污状态,包括:当所述目标图像的清晰度评分低于预设评分阈值时,确定所述摄像头当前处于脏污状态。这样设置,通过设置预设评分阈值,可以快速准确地判断出摄像头是否处于脏污状态。可选的,所述预设评分阈值根据所述预设清晰度确定模型对应的训练样本集中的训练样本标签确定。这样设置,可以合理地设定用于判定是否处于脏污状态的阈值。其中,预设评分阈值的确定方式不做限定,例如可以是训练样本标签的平均值或中位数等等。
图2为本申请实施例提供的又一种图像清晰度确定方法的流程示意图,可以适用于对因摄像头处于脏污状态导致所拍摄图像的清晰度不足进行识别的场景,如图2所示,该方法可包括如下操作。
步骤201、获取摄像头拍摄的目标图像。
步骤202、获取目标图像中每个像素对应的暗通道值。
示例性的,以I
C表示RGB数据格式下的目标图像,C表示RGB三通道,I
dark表示对应的暗通道,则可以存在如下对应关系:
即暗通道为目标图像的RGB三通道中最小值的灰度图。可以假设暗通道中数值接近0(可认为是黑色)的区域没有因脏污产生的白雾,而数值接近255(可认为是白色)的区域含白雾。对于一个像素来说,对应的暗通道值可以记为d。
步骤203、针对至少两个第一预设暗通道值区间中的每个第一预设暗通道值区间,统计当前第一预设暗通道值区间内包含的像素点的数量,得到当前第一预设暗通道值区间对应的暗通道积分值。
示例性的,暗通道积分值也可称为暗通道积分图,计算方法如下所示:
可选的,可以设置4个暗通道积分图,也即4个第一预设暗通道值区间,分别为[0,60)、[60,120)、[120,180)和[180,255],分别代表暗部区域、正常曝光区域、高光区域和白雾区域。
图3为本申请实施例提供的一种图像对比示意图,如图3所示,最左侧图像表示目标图像,中间图像表示暗通道对应的灰度图,最右侧图像表示暗通道积分图对应的图像。在最右侧图像中,针对暗部区域、正常曝光区域、高光区域和白雾区域分别采用不同的颜色进行填充,可以看出目标图像的不同明暗程度的区域的分布情况,其中,方框301中圈出了一部分的白雾区域。
步骤204、计算第二预设暗通道值区间内包含的像素点对应的暗通道值的平均值,并对所述平均值进行归一化处理,得到目标图像对应的曝光程度信息。
其中,第二预设暗通道值区间用于表示曝光过度,可以根据实际情况设置,例如可以是[250,255)。
示例性的,这里的曝光程度信息也可称为过曝率,可以描述图像的曝光程度,值域在[0,1]之间。曝光率可以采用如下表达式计算:
其中,∑
x∈[s,t)d表示在灰度图I
dark中的暗通道值属于值域[s,t)间的像素x的暗通道值的和,N表示暗通道值属于值域[s,t)间的像素x的个数,[s,t)可以是 [250,255)。
步骤205、计算所有暗通道值的中位数,得到目标图像对应的亮度信息。
示例性的,可以将暗通道值中位数视为目标图像的平均亮度,也即目标图像对应的亮度信息。
步骤206、将所有暗通道积分值、曝光程度信息、以及亮度信息输入至预设清晰度确定模型,并根据预设清晰度确定模型的输出结果确定目标图像的清晰度评分。
示例性的,可以收集训练样本图像,并对训练样本图像进行人工标注,也即为训练样本图像进行清晰度打分,分数值域为[0,10]。针对训练样本图像进行暗通道积分图、过曝率和平均亮度的计算,将{暗通道积分图,过曝率,平均亮度,清晰度分数}作为训练样本集,输入机器学习模型中进行训练,得到预设清晰度确定模型。
步骤207、若根据目标图像的清晰度评分确定所述摄像头当前处于脏污状态,则进行脏污提醒操作。
本申请实施例提供的图像清晰度确定方法,获取摄像头拍摄的目标图像,根据目标图像中每个像素的暗通道值计算相应的暗通道积分图、过曝率和平均亮度,并输入至预设清晰度确定模型中,可以快速准确地得出目标图像的清晰度评分,进而根据清晰度评分可以判断摄像头是否处于脏污状态,当处于脏污状态时可以及时进行脏污提醒操作,有利于提高拍摄质量。
图4为本申请实施例提供的另一种图像清晰度确定方法的流程示意图,可以适用于对摄像头的脏污状态进行识别并提醒的多种场景,如拍照场景、摄像场景以及直播场景等。如图4所示,该方法可包括如下操作。
步骤401、判断移动终端的摄像头是否打开,若是,则执行步骤402;否则,重复执行步骤401。
步骤402、判断移动终端的屏幕是否已出现脏污提醒,若是,则结束流程;否则,执行步骤403。
示例性的,若屏幕已经出现脏污提醒,则可不需要进行后续的判定,节省移动终端的运算资源。
步骤403、获取摄像头采集的目标图像,确定目标图像中每个像素对应的暗通道值。
示例性的,这里的目标图像例如可以是拍照场景中的预览图像,也可以是 摄像场景以及直播场景中的实时画面等。
步骤404、根据暗通道值计算预设数量的暗通道积分值、过曝率和平均亮度。
步骤405、将所有暗通道积分值、过曝率和平均亮度输入至预设清晰度确定模型,并根据预设清晰度确定模型的输出结果确定目标图像的清晰度评分。
步骤406、判断移动终端是否处于拍摄状态,若是,则执行步骤407;否则,返回执行步骤403。
示例性的,若移动终端未处于拍摄状态,说明此时提醒的意义不大,用户也可能不能及时查看屏幕内容,因此可以不需要进行判定,并返回步骤403获取新的图像进行判定。可以间隔预设时长后再返回执行步骤403,以降低移动终端的功耗。
步骤407、判断清晰度评分是否小于预设评分阈值,若是,则执行步骤408;否则,返回执行步骤403。
示例性的,预设评分阈值可以根据预设清晰度确定模型对应的训练样本集中的训练样本标签的平均值确定。
步骤408、在屏幕上显示脏污提醒。
示例性的,可以以弹窗的形式显示脏污提醒相关的文字或图标。图5为本申请实施例提供的一种屏幕界面示意图,如图5所示,提示文字例如可以是“您的镜头存在脏污,请您擦拭后继续拍摄”。
本申请实施例提供的方案,能够检测到摄像头脏污,并以弹窗的形式通知用户擦拭摄像头,解决因摄像头脏污造成的画质损失问题。相关方案中仅使用画面亮度作为评判指标,且极大依赖硬件设备,仅能应用在特定移动终端上,需要较多运算资源。而本申请实施例所采用的方案,使用暗通道积分图、过曝率、画面亮度水平作为评价指标,可以提升脏污提醒的准确性,减少提醒的误检率与漏检率,且能够适用于多种移动终端,对设备硬件要求较低,不需要与硬件进行交互,不需要较多运算资源,应用范围更加广泛。
图6为本申请实施例提供的一种图像清晰度确定装置的结构框图,该装置可由软件和/或硬件实现,一般可集成在计算机设备中,可通过执行图像清晰度确定方法来确定图像清晰度。如图6所示,该装置包括如下模块。
暗通道值获取模块601,设置为获取目标图像中每个像素对应的暗通道值;
评估指标信息确定模块602,设置为基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种;
清晰度确定模块603,设置为将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
本申请实施例提供的图像清晰度确定装置,获取目标图像中每个像素对应的暗通道值,基于暗通道值确定目标图像对应的评估指标信息,其中,评估指标信息包括暗通道积分信息、曝光程度信息和亮度信息中的至少一种,将评估指标信息输入至预设清晰度确定模型,并根据预设清晰度确定模型的输出结果确定所述目标图像的清晰度。通过采用上述技术方案,针对需要进行清晰度评估的目标图像,可以根据其中每个像素对应的暗通道值来合理准确地确定评估指标信息,在将评估指标信息输入到相应的预设清晰度确定模型后,可以得出准确的清晰度评估结果,从而提升图像清晰度确定方案的准确性。
本申请实施例提供了一种计算机设备,该计算机设备中可集成本申请实施例提供的图像清晰度确定装置。图7为本申请实施例提供的一种计算机设备的结构框图。计算机设备700包括存储器701、处理器702及存储在存储器701上并可在处理器702上运行的计算机程序,所述处理器702执行所述计算机程序时实现本申请实施例提供的图像清晰度确定方法。
本申请实施例还提供一种包含计算机可执行指令的存储介质,所述计算机可执行指令在由计算机处理器执行时用于执行本申请实施例提供的图像清晰度确定方法。存储介质可以是非暂态(non-transitory)存储介质。
上述实施例中提供的图像清晰度确定装置、设备以及存储介质可执行本申请任意实施例所提供的图像清晰度确定方法,具备执行该方法相应的功能模块和效果。未在上述实施例中详尽描述的技术细节,可参见本申请任意实施例所提供的图像清晰度确定方法。
Claims (10)
- 一种图像清晰度确定方法,包括:获取目标图像中每个像素对应的暗通道值;基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括以下至少之一:暗通道积分信息、曝光程度信息和亮度信息;将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
- 根据权利要求1所述的方法,其中,在所述评估指标信息包括所述暗通道积分信息的情况下,所述基于所述暗通道值确定所述目标图像对应的评估指标信息,包括:针对至少两个第一预设暗通道值区间中的每个第一预设暗通道值区间,统计当前第一预设暗通道值区间内包含的像素点的数量,得到所述当前第一预设暗通道值区间对应的暗通道积分值;根据至少两个暗通道积分值确定所述目标图像对应的暗通道积分信息。
- 根据权利要求1所述的方法,其中,在所述评估指标信息包括所述曝光程度信息的情况下,所述基于所述暗通道值确定所述目标图像对应的评估指标信息,包括:计算第二预设暗通道值区间内包含的像素点对应的暗通道值的总和,其中,所述第二预设暗通道值区间用于表示曝光过度;统计所述第二预设暗通道值区间内包含的像素点的总数;根据所述总和与所述总数的商确定所述目标图像对应的曝光程度信息。
- 根据权利要求1所述的方法,其中,在所述评估指标信息包括所述亮度信息的情况下,所述基于所述暗通道值确定所述目标图像对应的评估指标信息,包括:计算所有暗通道值的中位数或平均值;根据所述中位数或所述平均值确定所述目标图像对应的亮度信息。
- 根据权利要求1所述的方法,其中,所述预设清晰度确定模型通过以下方式得到:获取训练样本集,其中,所述训练样本集中的每个训练样本包括训练样本数据和训练样本标签,所述训练样本数据包括训练样本图像和所述训练样本图像对应的评估指标信息,所述训练样本标签包括所述训练样本图像对应的清晰度评分;利用所述训练样本集对预设机器学习模型进行训练,得到所述预设清晰度确定模型。
- 根据权利要求1-5中任一项所述的方法,其中,在所述获取目标图像中每个像素对应的暗通道值之前,还包括:获取摄像头采集的目标图像;在所述根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度之后,还包括:在根据所述目标图像的清晰度确定所述摄像头当前处于脏污状态的情况下,进行脏污提醒操作。
- 根据权利要求6所述的方法,其中,所述根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度,包括:根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度评分;所述根据所述目标图像的清晰度确定所述摄像头当前处于脏污状态,包括:在所述目标图像的清晰度评分低于预设评分阈值的情况下,确定所述摄像头当前处于脏污状态,其中,所述预设评分阈值根据所述预设清晰度确定模型对应的训练样本集中的训练样本标签确定。
- 一种图像清晰度确定装置,包括:暗通道值获取模块,设置为获取目标图像中每个像素对应的暗通道值;评估指标信息确定模块,设置为基于所述暗通道值确定所述目标图像对应的评估指标信息,其中,所述评估指标信息包括以下至少之一:暗通道积分信息、曝光程度信息和亮度信息;清晰度确定模块,设置为将所述评估指标信息输入至预设清晰度确定模型,并根据所述预设清晰度确定模型的输出结果确定所述目标图像的清晰度。
- 一种计算机设备,包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序,其中,所述处理器执行所述计算机程序时实现如权利要求1-7中任一项所述的图像清晰度确定方法。
- 一种计算机可读存储介质,存储有计算机程序,其中,所述计算机程序被处理器执行时实现如权利要求1-7中任一项所述的图像清晰度确定方法。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110231007.4 | 2021-03-02 | ||
CN202110231007.4A CN112767392B (zh) | 2021-03-02 | 2021-03-02 | 图像清晰度确定方法、装置、设备及存储介质 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022183902A1 true WO2022183902A1 (zh) | 2022-09-09 |
Family
ID=75690765
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2022/076274 WO2022183902A1 (zh) | 2021-03-02 | 2022-02-15 | 图像清晰度确定方法、装置、设备及存储介质 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN112767392B (zh) |
WO (1) | WO2022183902A1 (zh) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117253195A (zh) * | 2023-11-13 | 2023-12-19 | 广东申立信息工程股份有限公司 | 一种ipc安全监测方法、监测系统、计算机设备和可读存储介质 |
CN117395378A (zh) * | 2023-12-07 | 2024-01-12 | 北京道仪数慧科技有限公司 | 路产采集方法及采集系统 |
CN117676093A (zh) * | 2023-12-19 | 2024-03-08 | 苏州伟卓奥科三维科技有限公司 | 一种基于云服务的远程无线视频监控系统 |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112767392B (zh) * | 2021-03-02 | 2024-04-09 | 百果园技术(新加坡)有限公司 | 图像清晰度确定方法、装置、设备及存储介质 |
CN113505700A (zh) * | 2021-07-12 | 2021-10-15 | 北京字跳网络技术有限公司 | 一种图像处理方法、装置、设备及存储介质 |
CN115314633B (zh) * | 2022-06-27 | 2023-06-16 | 中国科学院合肥物质科学研究院 | 相机对焦方法、装置、计算机设备和存储介质 |
CN114820614A (zh) * | 2022-06-29 | 2022-07-29 | 上海闪马智能科技有限公司 | 一种图像类型的确定方法、装置、存储介质及电子装置 |
CN117135306A (zh) * | 2022-09-15 | 2023-11-28 | 深圳Tcl新技术有限公司 | 电视清晰度的调试方法及装置 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107958455A (zh) * | 2017-12-06 | 2018-04-24 | 百度在线网络技术(北京)有限公司 | 图像清晰度评估方法、装置、计算机设备及存储介质 |
CN108230288A (zh) * | 2016-12-21 | 2018-06-29 | 杭州海康威视数字技术股份有限公司 | 一种确定雾性状况的方法和装置 |
CN111739110A (zh) * | 2020-08-07 | 2020-10-02 | 北京美摄网络科技有限公司 | 一种图像过暗或过曝检测方法及装置 |
CN112102309A (zh) * | 2020-09-27 | 2020-12-18 | 中国建设银行股份有限公司 | 一种确定图像质量评估结果的方法、装置及设备 |
CN112188190A (zh) * | 2020-09-30 | 2021-01-05 | 广东美的厨房电器制造有限公司 | 污渍检测方法、烹饪器具、服务器和存储介质 |
CN112767392A (zh) * | 2021-03-02 | 2021-05-07 | 百果园技术(新加坡)有限公司 | 图像清晰度确定方法、装置、设备及存储介质 |
Family Cites Families (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101535630B1 (ko) * | 2014-01-21 | 2015-07-13 | 전남대학교산학협력단 | 밝기변환 모델을 이용한 야간 영상의 밝기 개선 장치 |
CN104063853B (zh) * | 2014-07-07 | 2017-02-08 | 南京通用电器有限公司 | 一种基于暗通道技术的提高交通视频图像清晰度方法 |
CN106339988B (zh) * | 2016-01-07 | 2019-06-14 | 深圳开阳电子股份有限公司 | 一种图像去雾方法及装置 |
CN107133927B (zh) * | 2017-04-21 | 2020-03-17 | 汪云飞 | 超像素框架下基于均值-均方差暗通道的单幅图像去雾方法 |
KR102096356B1 (ko) * | 2018-09-07 | 2020-04-02 | 고려대학교 산학협력단 | 다크 채널 기반의 단일 영상 안개 제거 방법, 장치 및 상기 방법을 수행하기 위한 기록 매체 |
CN109948625A (zh) * | 2019-03-07 | 2019-06-28 | 上海汽车集团股份有限公司 | 文本图像清晰度评估方法及系统、计算机可读存储介质 |
CN109978797B (zh) * | 2019-04-08 | 2020-06-23 | 自然资源部国土卫星遥感应用中心 | 基于暗通道噪声模板的主动式像素传感器星图去噪方法 |
CN112258380A (zh) * | 2019-07-02 | 2021-01-22 | 北京小米移动软件有限公司 | 图像处理方法、装置、设备及存储介质 |
CN111553880A (zh) * | 2020-03-26 | 2020-08-18 | 北京中科虹霸科技有限公司 | 模型生成方法、标签标注方法、虹膜图像质量评价方法及装置 |
CN112241452B (zh) * | 2020-10-16 | 2024-01-05 | 百度(中国)有限公司 | 一种模型训练方法、装置、电子设备及存储介质 |
-
2021
- 2021-03-02 CN CN202110231007.4A patent/CN112767392B/zh active Active
-
2022
- 2022-02-15 WO PCT/CN2022/076274 patent/WO2022183902A1/zh active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108230288A (zh) * | 2016-12-21 | 2018-06-29 | 杭州海康威视数字技术股份有限公司 | 一种确定雾性状况的方法和装置 |
CN107958455A (zh) * | 2017-12-06 | 2018-04-24 | 百度在线网络技术(北京)有限公司 | 图像清晰度评估方法、装置、计算机设备及存储介质 |
CN111739110A (zh) * | 2020-08-07 | 2020-10-02 | 北京美摄网络科技有限公司 | 一种图像过暗或过曝检测方法及装置 |
CN112102309A (zh) * | 2020-09-27 | 2020-12-18 | 中国建设银行股份有限公司 | 一种确定图像质量评估结果的方法、装置及设备 |
CN112188190A (zh) * | 2020-09-30 | 2021-01-05 | 广东美的厨房电器制造有限公司 | 污渍检测方法、烹饪器具、服务器和存储介质 |
CN112767392A (zh) * | 2021-03-02 | 2021-05-07 | 百果园技术(新加坡)有限公司 | 图像清晰度确定方法、装置、设备及存储介质 |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117253195A (zh) * | 2023-11-13 | 2023-12-19 | 广东申立信息工程股份有限公司 | 一种ipc安全监测方法、监测系统、计算机设备和可读存储介质 |
CN117253195B (zh) * | 2023-11-13 | 2024-02-27 | 广东申立信息工程股份有限公司 | 一种ipc安全监测方法、监测系统、计算机设备和可读存储介质 |
CN117395378A (zh) * | 2023-12-07 | 2024-01-12 | 北京道仪数慧科技有限公司 | 路产采集方法及采集系统 |
CN117395378B (zh) * | 2023-12-07 | 2024-04-09 | 北京道仪数慧科技有限公司 | 路产采集方法及采集系统 |
CN117676093A (zh) * | 2023-12-19 | 2024-03-08 | 苏州伟卓奥科三维科技有限公司 | 一种基于云服务的远程无线视频监控系统 |
CN117676093B (zh) * | 2023-12-19 | 2024-07-09 | 苏州伟卓奥科三维科技有限公司 | 一种基于云服务的远程无线视频监控系统 |
Also Published As
Publication number | Publication date |
---|---|
CN112767392A (zh) | 2021-05-07 |
CN112767392B (zh) | 2024-04-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2022183902A1 (zh) | 图像清晰度确定方法、装置、设备及存储介质 | |
WO2017215501A1 (zh) | 图像降噪处理方法、装置及计算机存储介质 | |
WO2018072271A1 (zh) | 一种图像显示优化方法及装置 | |
CN110100251A (zh) | 用于处理文档的设备、方法和图形用户界面 | |
CN105450932B (zh) | 逆光拍照方法和装置 | |
WO2021135945A1 (zh) | 图像处理方法、装置、存储介质以及电子设备 | |
CN109887044B (zh) | 一种生殖数据测评方法及系统 | |
WO2020108010A1 (zh) | 视频处理方法、装置、电子设备以及存储介质 | |
CN111881862B (zh) | 手势识别方法及相关装置 | |
WO2023005827A1 (zh) | 曝光补偿方法、装置和电子设备 | |
WO2023137956A1 (zh) | 图像处理方法、装置、电子设备及存储介质 | |
CN106773453B (zh) | 一种照相机曝光的方法、装置及移动终端 | |
CN114038370B (zh) | 显示参数调整方法、装置、存储介质及显示设备 | |
CN111429537A (zh) | 电影银幕的光学检测方法、装置、设备及智能网络传感器 | |
CN109040729B (zh) | 图像白平衡校正方法、装置、存储介质及终端 | |
CN111291778A (zh) | 深度分类模型的训练方法、曝光异常检测方法以及装置 | |
WO2020224459A1 (zh) | 图像处理方法、装置、终端及存储介质 | |
CN111447360A (zh) | 应用程序控制方法及装置、存储介质、电子设备 | |
CN110782392A (zh) | 图像处理方法、装置、电子设备及存储介质 | |
WO2022262848A1 (zh) | 图像处理方法、装置和电子设备 | |
WO2023001110A1 (zh) | 神经网络训练方法、装置及电子设备 | |
WO2021078276A1 (zh) | 连拍照片获取方法、智能终端及存储介质 | |
CN108881636B (zh) | 逆光模式自动切换方法、装置、电子设备及计算机可读存储介质 | |
WO2019090591A1 (zh) | 一种基于智能终端的拍照方法及拍照系统 | |
WO2023236209A1 (zh) | 图像处理方法、装置、电子设备和存储介质 |
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: 22762368 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: 22762368 Country of ref document: EP Kind code of ref document: A1 |