WO2023246171A1 - 图像处理方法、装置及相关设备 - Google Patents

图像处理方法、装置及相关设备 Download PDF

Info

Publication number
WO2023246171A1
WO2023246171A1 PCT/CN2023/079807 CN2023079807W WO2023246171A1 WO 2023246171 A1 WO2023246171 A1 WO 2023246171A1 CN 2023079807 W CN2023079807 W CN 2023079807W WO 2023246171 A1 WO2023246171 A1 WO 2023246171A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
target object
pixel value
pixel
specified background
Prior art date
Application number
PCT/CN2023/079807
Other languages
English (en)
French (fr)
Inventor
解松
陈逸飞
陈大鹏
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2023246171A1 publication Critical patent/WO2023246171A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/50Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10016Video; Image sequence

Definitions

  • the present application relates to the field of artificial intelligence technology, and in particular, to an image processing method, device and related equipment.
  • AI artificial intelligence
  • video conferencing video conferencing
  • image design image design
  • poster design it is common to compare part of the image content in one image with part of the image content in another image. Synthesize and generate new images.
  • users can replace other image contents in the video image except for the person image with a specified designated background image.
  • the virtual background be used to hide the surrounding environment of the person and protect user privacy, but also by setting different Specify a background image to improve user video conferencing experience.
  • This application provides an image processing method to improve the quality of synthesized images and enhance the image synthesis effect.
  • this application also provides corresponding devices, computing equipment, computer-readable storage media, and computer program products.
  • the present application provides an image processing method, which can be executed by an image processing device.
  • the image processing device obtains an image to be processed and a specified background image.
  • the image to be processed includes an object of a target image, and the target The object can be a person, an object or other types of objects; then, the image processing device identifies the image of the target object in the image to be processed, for example, through an image segmentation algorithm to identify the image of the target object, etc., so that the image processing device recognizes the image of the target object according to the image and Perform an image harmonization operation on the specified background image to obtain a composite image, where the image harmonization operation is used to indicate a way to adjust the image of the target object according to the pixel value of the specified background image, so that the target object in the obtained composite image The image is more harmonious with the specified background image; finally, the image processing device outputs the composite image.
  • the image processing device performs an image harmonization operation based on the image of the target object and the specified background image during the process of synthesizing the image, the difference between the pixel values of the image of the target object and the specified background image is reduced, that is, the difference in pixel values during synthesis is reduced.
  • the brightness difference between the image of the target object in the image and the specified background image makes the image of the target object and the specified background image in the composite image more harmonious, thereby improving the quality of the generated composite image and improving the image synthesis effect.
  • the pixel values of each pixel in the image of the target object can be independently adjusted. Since the adjustment range of pixel values of different pixels may be different, adjusting the pixel values of all pixels in the image using a unified adjustment method will cause the adjusted image of the target object to have some pixels that are too bright or too dark. , and the independent adjustment method for pixels can be adjusted according to the differences in pixels. In this way, the quality of the synthesized image can be improved and the effect of image synthesis can be improved.
  • the image processing device when it performs an image harmonization operation, it may specifically adjust the pixel values of the image of the target object according to the pixel values of the specified background image to obtain an intermediate image, wherein the adjusted The difference between the pixel value of the intermediate image and the pixel value of the specified background image is smaller than the difference between the pixel value of the target object's image and the pixel value of the specified background image, thereby reducing the difference between the target object's image and the specified background image.
  • the image processing device synthesizes the intermediate image and the specified background image based on the difference in pixel values between them to obtain a composite image. In this way, the brightness difference between the image of the target object and the specified background image can be reduced, thereby making the generated composite image more harmonious. Harmony to improve the quality of synthesized images.
  • the image harmonization operation performed by the image processing device is used to instruct the pixel values of each pixel point in the image of the target object to be individually adjusted according to the pixel value of the specified background image. There are differences in how much pixel values are adjusted. Differential adjustment for each pixel can avoid the problem that some pixels in the adjusted target object image are too bright or too dark, which can effectively improve the quality of the synthesized image and improve the effect of image synthesis.
  • the image harmonization operation performed by the image processing device is used to instruct the pixel values of each pixel point in the first area in the image of the target object to be individually processed based on the pixel values of the specified background image. Adjust, and uniformly adjust the pixel values of multiple pixels in the second area in the image of the target object. In this way, the amount of calculation required to adjust the pixel value of the image of the target object can be reduced, thereby reducing resource consumption while generating a higher-quality synthetic image.
  • the image processing device can also acquire a new image to be processed, so as to autonomously adjust the image of the target object in the new image to be processed according to the pixel value of the image of the target object in the composite image.
  • the image of the target object can maintain similar brightness, eliminating the need for the user to make manual adjustments.
  • the image processing device may first calculate the difference between the pixel value of the image of the target object and the pixel value of the specified background image, so that when the difference is greater than the predetermined
  • the pixel value of the image of the target object is iteratively adjusted based on the difference to obtain an intermediate image.
  • the difference between the pixel value of the image of the target object and the pixel value of the specified background image can be gradually reduced, and the image of the target object can achieve a better brightness adjustment effect.
  • the image processing device when iteratively adjusts the pixel value of the image of the target object, specifically, it may determine the iteration direction and the number of iterations based on the difference between the image of the target object and the specified background image, so as to determine the iteration direction and the number of iterations according to the difference between the image of the target object and the specified background image.
  • the iteration direction and the number of iterations are used to iteratively adjust the pixel value of the image of the target object. In this way, iterative adjustment of the image of the target object can be achieved.
  • the image processing device when iteratively adjusts the pixel value of the image of the target object according to the iteration direction and the number of iterations, specifically, it may determine the corresponding pixel point of each pixel in the image of the target object according to the iteration direction.
  • the iterative calculation curve is used to iteratively adjust the pixel value of each pixel according to the number of iterations and the iterative calculation curve corresponding to each pixel.
  • the iterative calculation curve used by the image processing device to adjust the pixel value of each pixel point includes iteration parameters. Then, when the image processing device determines the iterative calculation curve corresponding to each pixel point, the target can be The image of the object is input to the artificial intelligence model, and the iteration parameters corresponding to each pixel point in the image of the target object output by the artificial intelligence model are obtained, and the corresponding iterative calculation curve is determined based on the iteration parameters corresponding to each pixel point.
  • the artificial intelligence model used to determine the iteration parameters is constructed based on a convolutional neural network with a preset number of layers, such as a 5-layer convolutional neural network model, so that it can be constructed through a smaller size
  • the artificial intelligence model determines the iteration parameters corresponding to each pixel, reducing the computational overhead of the image processing device.
  • the image processing device can also present a configuration interface including a plurality of different preview images, wherein each preview image is generated based on a specified background image and an image of the target object, and different preview images are generated
  • the adopted images of the target object have different pixel values, so that the image processing device determines the target pixel values of the image of the target object used to generate the user-selected preview image in response to the user's selection operations on multiple different preview images.
  • the image processing device when the image processing device generates the intermediate image, the pixel value of the image of the target object can be adjusted to the target pixel value to obtain the intermediate image.
  • the brightness of the image of the target object is the brightness selected by the user. In this way, the user's degree of freedom in selecting the brightness of the image of the target object can be improved and the user experience can be improved.
  • the plurality of preview images presented by the image processing device are generated based on images of target objects with multiple brightness levels, and the pixel values of images of target objects with different brightness levels are different. In this way, it is convenient for users to select images of target objects with corresponding brightness levels according to their own needs to generate composite images, thereby making the generated composite images more likely to meet user expectations and improve user experience.
  • the image processing device can also record the pixel values of the pixels in the image of the target object during each round of iterative adjustment of the pixel values of the image of the target object, and adjust the pixel values of the pixels in the image of the target object according to the Click on the pixel values during multiple rounds of iterative adjustments to generate multiple different preview images.
  • multiple preview images that are convenient for users to refer to for brightness adjustment can be determined based on the pixel values in multiple rounds of iterative adjustment processes, thereby improving the user's degree of freedom in brightness selection.
  • the target object is specifically a participant participating in the video conference. Therefore, when the image processing device obtains the image to be processed, it may specifically use a shooting device to photograph the participant. , obtain images to be processed including images of participants.
  • the image processing device may also be applied in other scenes, and the user generates composite images in other scenes.
  • the present application provides an image processing device, which includes various modules for executing the image processing method in the first aspect or any possible implementation of the first aspect.
  • the present application provides a computing device including a processor, a memory, and a display.
  • the processor and the memory communicate with each other.
  • the processor is configured to execute instructions stored in the memory, so that the computing device executes the image processing method as in the first aspect or any implementation of the first aspect.
  • the memory can be integrated into the processor or independent of the processor.
  • the computing device may also include a bus. Among them, the processor is connected to the memory through a bus.
  • the memory may include readable memory and random access memory.
  • the present application provides a computer-readable storage medium in which instructions are stored, which when run on a computing device, cause the computing device to execute the above-mentioned first aspect or any of the first aspects. Operation steps of the image processing method described in one implementation manner.
  • the present application provides a computer program product containing instructions that, when run on a computing device, causes the computing device to execute the image processing method described in the first aspect or any implementation of the first aspect. Steps.
  • Figure 1 is a schematic diagram of an exemplary application scenario provided by this application.
  • Figure 2 is a schematic diagram showing that the brightness of the image of user 101 in the composite image is much higher than the brightness of the specified background image;
  • Figure 3 is a schematic diagram of another exemplary application scenario provided by this application.
  • Figure 4 is a schematic flow chart of an image processing method provided by this application.
  • Figure 5 is a schematic diagram of different iterative calculation curves corresponding to different pixel points provided by this application;
  • Figure 6 is a schematic diagram of preview images of four target object images provided by this application with gradually increasing brightness
  • Figure 7 is a schematic diagram of a preview image in which the image brightness of the target object provided by this application gradually increases and gradually decreases;
  • FIG. 8 is a schematic structural diagram of an image processing device provided by this application.
  • Figure 9 is a schematic diagram of the hardware structure of a computing device provided by this application.
  • FIG 1 is a schematic diagram of an exemplary application scenario provided by an embodiment of the present application.
  • the terminal 201 is equipped with a camera, and the camera is used to continuously collect images of the user 101 to obtain multi-frame video images including the user 101; at the same time, the terminal 201 is equipped with a microphone and a speaker, where the microphone is used for recording.
  • the speaker is used to play the voice generated by user 101, such as the voice generated by other users (such as user 102, etc.).
  • the terminal 201 sends the multi-frame video image and the voice generated by the user 101 to the terminal 202 through the network 300, so that the terminal 202 presents the multi-frame video image on the display interface and plays the user 101's voice through the speaker of the terminal 202.
  • the terminal 201 can present the continuous multiple frames of video images captured by the camera on the display interface, so that the user 101 can view his own images in the video conference.
  • the terminal 201 will also receive the multi-frame video image including the user 102 and the voice of the user 102 sent by the terminal 200 through the network 300, and present the multi-frame video image including the user 102 on the display interface of the terminal 201.
  • the voice of user 102 is played through the speaker on terminal 201, thereby realizing an online video conference between user 101 and user 102.
  • the user 101 can set the designated background image of the user 101 in the video conference screen.
  • the background of the characters in the video image viewed by the user 102 on the display interface of the terminal 202 is the designated background set by the user 101. image.
  • the user 101 can set the specified background image in the video conference screen as the virtual background shown in Figure 2, so that the virtual background can not only be used to hide the environmental information around the user 101 and protect the privacy of the user 101, but also the user 101 can follow the Set different designated background images according to your own preferences to improve the experience of users 101 participating in video conferences.
  • the terminal 201 directly synthesizes the image of the user 101 in the video image with the designated background image, it will easily lead to disharmony between the image of the user 101 in the synthesized image and the designated background image.
  • the synthesized image shown in Figure 2 the user The brightness of the image of 101 is much higher than the brightness of the specified background image, which results in lower quality of the synthesized image and poor image synthesis effect.
  • embodiments of the present application provide an image processing method for improving the quality of synthesized images and improving the image synthesis effect.
  • the method can be executed by an image processing device, which acquires an image to be processed and a specified background image.
  • the image to be processed includes an image of the user 101, and the specified background image can be an image specified by the user 101, so that the image
  • the processing device first identifies the image of the user 101 from the image to be processed, and performs an image harmonization operation based on the image of the user 101 and the specified background image to obtain a composite image.
  • the image harmonization operation is used to indicate adjustment of pixel values according to the specified background image.
  • the image of the target object is modified in such a way that the difference between the pixel values of the specified background image and the pixel value of the image of the target object is reduced (that is, the image of the target object is more harmonious with the specified background image).
  • the terminal 201 outputs the composite image.
  • the composite image can be presented on the display interface of the terminal 201, and the composite image can be sent to the terminal 202 for presentation.
  • the image processing device performs an image harmonization operation based on the image of the user 101 and the specified background image during the process of synthesizing the image, specifically the pixel value of the image of the user 101 is adjusted to reduce the pixel value of the image of the user 101 and the specified background image.
  • the difference between them that is, the brightness difference between the image of the user 101 and the specified background image in the composite image is reduced, which makes the difference between the image of the user 101 and the specified background image in the composite image generated by the image processing device closer. Harmony, which can improve the quality of the generated synthetic images and improve the image synthesis effect.
  • the pixel values of each pixel point in the image of the user 101 can be adjusted independently. Since the adjustment range of pixel values of different pixels may be different, adjusting the pixel values of all pixels in the image using a unified adjustment method will cause the adjusted image of user 101 to have partial images. The problem of pixels being too bright or too dark, and the independent adjustment method for pixels can be adjusted according to the differences in pixels. In this way, the quality of the synthesized image can be improved and the effect of image synthesis can be improved.
  • the image processing apparatus can be implemented through a virtual device, for example, through at least one of a virtual machine, a container, and a computing engine.
  • the image processing device may be implemented by a physical device including a processor, where the processor may be a CPU, an application-specific integrated circuit (ASIC), a programmable logic device (PLD), a complex Programmable logic device (complex programmable logical device, CPLD), field-programmable gate array (FPGA), general array logic (GAL), system on chip (SoC), software definition Any kind of processor such as software-defined infrastructure (SDI) chip, artificial intelligence (AI) chip, or any combination thereof.
  • the number of processors included in the image processing device may be one or multiple. Specifically, the number of processors may be set according to the business requirements of the actual application, which is not limited in this embodiment.
  • the image processing method provided by the embodiment of the present application can also be applied to other applicable scenarios.
  • one or more computing devices 301 deployed in the cloud can be used to synthesize the image to be processed and the specified background image.
  • the image processing device can be Computing device 301.
  • the terminal 201 is responsible for sending the collected multi-frame video images including the user 101 and the specified background image specified by the user 101 to the computing device 301 in the cloud; after the computing device 301 generates the composite image, it will be able to send the composite image.
  • the terminal 201 and the computing device 301 can cooperate to execute the above image processing method; or the terminal 201 is not limited to providing video conferencing services to the user 101, but can also provide other services to the user 101.
  • the user 101 can generate The method of synthesizing images can realize image design, poster design, etc. This application is not limited to this.
  • Figure 4 is a schematic flow chart of an image processing method provided by an embodiment of the present application.
  • the image processing method shown in Figure 4 can be executed by a corresponding image processing device, which can be deployed on the user side.
  • the image processing device can be a terminal on the user side. 201;
  • the image processing device can be deployed in the cloud.
  • the image processing device can be the computing device 301.
  • the application scenario shown in Figure 1 is used as an example for illustrative description.
  • the image processing method shown in Figure 4 may specifically include:
  • the terminal 201 obtains the image to be processed and the specified background image.
  • the image to be processed includes an image of the target object.
  • the target object may be a person.
  • the target object may be the user 101, etc.; or the target object may be an object, such as in the poster design scene. It can be goods produced by manufacturers, etc., in order to advertise them.
  • the terminal 201 when the user 101 has a need for image synthesis, the terminal 201 obtains the image to be processed and the designated background image required to participate in the image synthesis.
  • the terminal 201 may be equipped with a camera, and the terminal 201 may use the camera to photograph the user 101 (ie, the target object) to obtain one or more frames of images to be processed including the user 201 image.
  • the terminal 201 can also present an interactive interface to the user 101.
  • the interactive interface includes a variety of background images, so that the user 101 can select the multiple background images.
  • the various designated background images may be downloaded locally from the network in advance by the terminal 201, or may be configured on the terminal 201 by technicians.
  • the background image selected by the user is called a designated background image.
  • the terminal 201 can use the specified background image and camera selected by the user 101 Image synthesis is performed on one of the frames of images to be processed collected by the imaging device.
  • the terminal 201 can present an interactive interface to the user 101, so that the user 101 can import the image to be processed and the specified background image into the terminal 201 on the interactive interface.
  • the terminal 201 can obtain the image to be processed.
  • the specific implementation of processing images and specifying background images is not limited.
  • the terminal 201 identifies the image of the target object in the image to be processed.
  • the terminal 201 can Extract the image of the target object from the image to be processed.
  • the terminal 201 can use a target segmentation algorithm to identify the target object in the image to be processed, and further determine the image area of the target object in the image.
  • the target segmentation algorithm may be, for example, at least one of the mask regions with convolution neural networks features (mask R-CNN) algorithm.
  • the terminal 201 can input the image to be processed into a target segmentation model constructed based on the mask R-CNN algorithm to obtain the target object output by the target segmentation model.
  • the mask corresponding to the image in the image to be processed is used to indicate the image area of the target object in the image, so that the image of the target object can be extracted from the image to be processed by using the mask.
  • the mask is a template of an image filter.
  • the image can be filtered by pixels through a matrix, so as to segment the part of the image information from the image.
  • the matrix is for the mask.
  • the specific implementation manner in which the terminal 201 determines the image of the target object is not limited.
  • the terminal 201 performs an image harmonization operation based on the image of the target object and the specified background image to obtain a composite image.
  • the image harmonization operation is used to indicate a manner of adjusting the image of the target object according to the pixel value of the specified background image.
  • the difference between the pixel value (or brightness) of the image to be processed and the pixel value of the specified background image is The difference exceeds the preset range, that is, the difference between the pixel value of the target object's image and the pixel value of the specified background image exceeds the preset range.
  • the terminal 201 directly synthesizes the image of the target object and the designated background image, the brightness of the image of the target object and the brightness of the designated background image will be too different, causing the final generated synthesized image to be disharmonious, affecting image synthesis. Effect.
  • the terminal 201 can perform an image harmonization operation on the image of the target object according to the pixel value of the specified background image, so that the image of the target object and the specified background image are more harmonious in brightness.
  • the terminal 201 can adjust the pixel value of the image of the target object.
  • the image of the target object after the pixel value has been adjusted is called the intermediate image.
  • the difference between the pixel value of the image and the pixel value of the specified background image is smaller than the difference between the pixel value of the image of the target object before adjustment and the pixel value of the specified background image. That is, the pixel value of the target object is reduced by adjusting the pixel value.
  • the difference between the brightness of the image and the brightness of the specified background image In this way, subsequent synthesis of the adjusted target object image and the specified background image can make the final synthesized image more harmonious because the brightness difference between different parts of the image content is smaller, thereby improving the effect of image synthesis. .
  • the terminal 201 can gradually reduce the brightness difference between the image of the target object and the specified background image by iteratively adjusting the pixel value of the image of the target object. Specifically, since usually both the image of the target object and the specified background image include multiple pixels, the terminal 201 can calculate the average (or median) of the pixel values of all pixels in the image of the target object, And the average value (or median) is used as the pixel value of the image of the target object.
  • the pixel value of a pixel refers to the position of the pixel in red (red, R), green (green, G), blue (blue, B)
  • the values of the three color channels; correspondingly, the pixel value of the image (such as the pixel value of the aforementioned target object image, the pixel value of the specified background image, etc.) is the value of the three RGB color channels, and the image is in
  • the value of each color channel is the average (or median) of the values of all pixels in the image in that color channel.
  • the terminal 201 may use the average (or median) of the pixel values of all pixels in the specified background image as the pixel value of the specified background image.
  • the terminal 201 can first calculate the difference between the pixel value of the image of the target object and the pixel value of the specified background image based on the image of the target object, and iteratively adjust the pixel value of the image of the target object based on the difference.
  • An intermediate image is obtained until the number of iterations to adjust the pixel value reaches the preset number, or the difference between the pixel value of the image of the target object and the pixel value of the specified background image is less than the preset range.
  • the pixel value of each channel is represented by a percentage system (the value range is 0 ⁇ 1)
  • the pixel value of the target object image is (0.1, 0.2, 0.1)
  • the pixel value of the specified background image is (0.5, 0.6, 0.6 )
  • the difference of the pixel values of the R channel is 0.4
  • the difference of the pixel values of the G channel is 0.4
  • the difference of the pixel values of the B channel is 0.5.
  • the pixel value of the R channel of the target object's image can be adjusted to 0.3, 0.4, 0.45, 0.48, so that the difference between the pixel value of the R channel of the intermediate image (0.48) obtained after 4 iterative adjustments and the pixel value of the R channel of the specified background image (0.5) is smaller.
  • the adjustment of the pixel values of the G channel and B channel is similar to it, and will not be described again here.
  • the terminal 201 may first determine the iteration direction and the number of iterations based on the difference between the pixel value of the image of the target object and the pixel value of the specified background image.
  • the iteration direction is used to indicate the direction of adjusting the pixel value size of the image of the target object, including the direction of increasing the pixel value and the direction of decreasing the pixel value, which can be determined based on the sign of the difference.
  • the terminal 201 can determine to iteratively adjust the pixel value of the image of the target object along the direction of increasing the pixel value;
  • the terminal 201 can determine iteratively adjust the pixel value of the image of the target object along the direction of decreasing the pixel value.
  • the number of iterations is used to indicate the number of adjustments to the pixel value size of the image of the target object. The number of iterations can be determined based on the absolute value of the difference.
  • the number of iterations is 1; the absolute value of the difference is between [0.2, 0.3) between, the number of iterations is 2; the absolute value of the difference is between [0.3, 0.4), the number of iterations is 3; the absolute value of the difference is above 0.4, the number of iterations is 4, etc.
  • technicians can configure the corresponding relationship between the absolute value of the difference and the number of iterations in the terminal 201 in advance, so that the terminal 201 calculates the difference between the pixel value of the image of the target object and the pixel value of the specified background image.
  • the corresponding relationship can be found based on the absolute value of the difference to determine the number of iterations.
  • the user 101 may also specify the iteration direction and the number of iterations on the interactive interface, which is not limited in this embodiment.
  • the terminal 201 may detect when the difference between the pixel value of the target object's image and the pixel value of the specified background image is greater than the preset range (for example, the difference between the two pixel values on the R channel exceeds -0.1 ⁇ 0.1), the terminal 201 performs one or more iterative adjustments to the pixel value of the image of the target object; and when the difference between the pixel value of the image of the target object and the pixel value of the specified background image does not exceed the
  • the range is preset (for example, the difference between the two pixel values on the R channel is in the range of -0.1 to 0.1)
  • the terminal 201 may not adjust the pixel value of the image of the target object, that is, the terminal 201 may directly adjust the pixel value of the target object.
  • the image of the target object is composited with the specified background image.
  • the image harmonization operation performed by the terminal 201 on the image of the target object may include the following two implementation methods.
  • the image harmonization operation is used to instruct the pixel values of each pixel point in the image of the target object to be individually adjusted according to the pixel value of the specified background image.
  • the adjustment amplitudes of the pixel values of different pixel points are differences in the adjustment amplitudes of the pixel values of different pixel points. No There can also be differences in the iteration directions of the same pixel.
  • the adjustment range of the pixel value of pixel point A is 10%, while the adjustment range of the pixel value of pixel point A is 25%, etc. In this way, the problem of excessively bright or dark pixels caused by excessive adjustment of some pixels can be effectively avoided.
  • the image harmonization operation is used to instruct the pixel values of each pixel point in the first area in the image of the target object to be individually adjusted according to the pixel value of the specified background image, and the image of the target object
  • the pixel values of multiple pixels in the second area are uniformly adjusted, where the first area and the second area are non-overlapping image areas.
  • the terminal 201 can uniformly adjust the pixel values of multiple pixels in the area, such as uniformly increasing the pixel values of the multiple pixels in the area by 20%, etc.
  • the terminal 201 can perform a calculation on the pixel values of each pixel in the area. Adjust individually.
  • this embodiment provides the following two implementation methods for adjusting the pixel value of a pixel point.
  • the terminal 201 can iteratively adjust the pixel value of each pixel based on the iterative calculation curve corresponding to each pixel, and during multiple iterations, the iterative calculation curve corresponding to each pixel remains unchanged.
  • the terminal 201 can determine the iterative calculation curve corresponding to each pixel point in the image of the target object according to the iterative direction of adjusting the pixel value. There may be differences in the iterative calculation curves of different pixel points. Then, the terminal 201 can use the iterative calculation curve corresponding to each pixel to perform one or more iterative calculations on the pixel value of the pixel according to the number of iterations determined above, so that the terminal 201 can calculate the pixel value of the pixel. Adjust to the pixel value calculated in the last iteration to adjust the pixel value of each pixel.
  • the terminal 201 can determine the iterative calculation curve for each pixel according to the iteration direction as shown in the following formula (1):
  • x i is the pixel value of the i-th pixel in the image of the target object before iterative adjustment
  • y i is the pixel value of the i-th pixel in the image of the target object after iterative adjustment
  • is, where, when the iteration direction When the direction of increasing the pixel value is, the value of ⁇ is greater than 0, and when the iteration direction is the direction of decreasing the pixel value, the value of ⁇ is less than 0, and the value of ⁇ in the iterative calculation curve corresponding to different pixel points can be are different, so there may be differences in the iterative calculation curves corresponding to different pixels.
  • pixel point 1 to pixel point 5 respectively correspond to different iterative calculation curves.
  • the terminal 201 determines that the number of iterations is 2, the pixel value of the first pixel in the image of the target object is m, and the value of the iteration parameter determined by the terminal 201 for this pixel is ⁇ 1 , Then the pixel value of the pixel after the first iteration is The pixel value after the second iteration Similarly, the pixel values of the remaining pixels in the image of the target object after two iterations can also be calculated in a similar manner as described above.
  • the pixel value is adjusted individually for each pixel in the image of the target object (that is, the iteration parameters corresponding to different pixels can be different), rather than uniformly adjusting all pixels (such as all pixels
  • the pixel values are increased by 0.1, etc.), which can effectively avoid the problem of some pixels being too bright or too dark.
  • the pixel values of the R channel before iteration are 0.5 and 0.1 respectively.
  • the pixel value of pixel A in the R channel can be adjusted to 0.55
  • the pixel value of pixel B in the R channel can be adjusted to 0.45.
  • the terminal 201 can separately adjust the pixel values of each pixel point on the G channel and the B channel based on the above-mentioned similar method.
  • the iteration parameters in the iterative calculation curve corresponding to each pixel point can be obtained by inference using an artificial intelligence (artificial intelligence, AI) model that has been trained in advance.
  • AI artificial intelligence
  • the terminal 201 can input the image of the target object into the AI model, and the AI model outputs a parameter matrix.
  • Each element in the parameter matrix is the iteration parameter corresponding to each pixel in the image of the target object.
  • the size (and sign) of the iteration parameters corresponding to different pixels may be different. In this way, based on the above formula (1), the iterative calculation curve corresponding to each pixel point can be further determined according to the iteration parameters corresponding to each pixel point.
  • the AI model can output two parameter matrices, namely the parameter matrix a used to increase the brightness of pixels and the parameter matrix b used to reduce the brightness of pixels.
  • each element in the parameter matrix a is the iteration parameter corresponding to each pixel in the image of the target object, and based on the iteration parameters in the parameter matrix a, the pixel value of each pixel can be increased (i.e., the image of the target object Brighten);
  • each element in the parameter matrix b is the iteration parameter corresponding to each pixel in the image of the target object, and based on the iteration parameters in the parameter matrix b, the pixel value of each pixel can be reduced (that is, the target object the image is darkened).
  • the terminal 201 can choose to use parameter matrix a or parameter matrix b to iteratively adjust the pixel value of the image of the target object based on the pixel value of the image of the target object and the pixel value of the specified background image.
  • the terminal 201 may be configured with two AI models, where the AI model A is used to output the above parameter matrix a, and the AI model B is used to output the above parameter matrix b, so that the terminal 201 can be based on the image of the target object.
  • the difference between the pixel value and the pixel value of the specified background image determines the iteration direction, and then selects the corresponding AI model according to the iteration direction to determine the parameter matrix that conforms to the iteration direction for the image of the target object, so that the parameter matrix can be used to target the target.
  • the pixel values of the object's image are iteratively adjusted.
  • the above AI model can be constructed based on a convolutional neural network with a preset number of layers.
  • the above-mentioned AI model can be built based on the visual pleasing lighting enhancement network (VPLE-Net) including a five-layer convolutional neural network.
  • VPLE-Net visual pleasing lighting enhancement network
  • Implementation method 2 The terminal 201 can iteratively adjust the pixel value of each pixel based on the iterative calculation curve corresponding to the pixel, and in different rounds of iterative processes, the iterative calculation curve used for each pixel is different.
  • the terminal 201 re-determines the iteration parameters in this round of iteration process for each pixel point in the image of the target object, that is, in different rounds of iteration processes, each The iteration parameters of pixels can be different.
  • the terminal 201 inputs the image of the target object into the AI model and obtains the matrix parameter I output by the AI model.
  • Each element in the matrix parameter I is the target.
  • the corresponding iteration parameters of each pixel point in the object's image in the first round of iteration process so based on the above formula (1), the iterative calculation curve I corresponding to each pixel point can be further determined based on the iteration parameters corresponding to each pixel point, In order to calculate the pixel value of each pixel after the first round of iteration according to the iterative calculation curve I corresponding to each pixel, thereby generating a temporary image based on the updated pixel value of each pixel.
  • the terminal 201 can input the temporary image into the AI model to obtain the matrix parameter II output by the AI model.
  • Each element in the matrix parameter II is the image of the target object.
  • the corresponding iteration parameters of each pixel in the second round of iteration process so based on the above formula (1), the iteration calculation curve II corresponding to each pixel can be further determined according to the iteration parameters corresponding to each pixel, so that according to each pixel
  • the iterative calculation curve II corresponding to the point calculates the pixel value of each pixel point after the second round of iteration, thereby generating the above-mentioned intermediate image based on the pixel value of each pixel point after the second round of iteration.
  • Implementation method three for a part of the area in the image of the target object, the terminal 201 can The iterative calculation curve corresponding to the prime point is iteratively adjusted to the pixel value of the pixel, and in different rounds of iteration processes, the iterative calculation curve used for each pixel can be the same or different; while in the image of the target object For the remaining area, the terminal 201 can make unified iterative adjustments to the pixel values of all pixels in the area based on the iterative calculation curve corresponding to one of the pixels in the area, and in different rounds of iteration processes, The iterative calculation curves used by all pixels in this area can be the same or different.
  • the terminal 201 may adjust the pixel value of each pixel in the image of the target object based on the difference between the pixel value of the image of the target object and the pixel value of the specified background image. (Multiple iterative adjustments are not needed), and the adjustment range of the pixel values of different pixels can be different, so that an intermediate image can be directly generated, etc.
  • the terminal 201 can automatically generate an intermediate image through the above iterative process.
  • the user 101 can also manually intervene in the adjustment of pixel values, so that the terminal 201 can generate an intermediate image based on the operation of the user 101 .
  • the terminal 201 in the process of iteratively adjusting the pixel values of the image of the target object, can record the pixel values of the pixels in the image of the target object after each round of iteration, and based on the pixel values after the round of iteration Based on the pixel value of the point, a temporary image can be obtained, and multiple temporary images can be obtained based on multiple rounds of iterative processes. Then, the terminal 201 can fuse multiple temporary images with the specified background image respectively. In this embodiment, the fused image is called a preview image, so that multiple previews generated based on the multiple temporary images can be obtained. image, and the pixel values of the image of the target object in different preview images are different.
  • the terminal 201 can present the generated plurality of preview images to the user 101 so that the user 101 can select the plurality of preview images.
  • the preview image selected by the user 101 is usually the preview image that the user 101 feels has the most harmonious synthesis effect among the plurality of preview images.
  • the terminal 201 can present four preview images to the user 101. There are differences in the pixel values of the image of the target object in the four preview images, as shown in Figure 6. It is shown that the image of the target object in the four preview images gradually becomes brighter, so that the user 101 can select a preview image that he thinks is the most harmonious from the four preview images.
  • the terminal 201 can iteratively adjust the pixel value of the image of the target object along two directions of increasing brightness and decreasing brightness respectively, and each iteration adjusts a brightness level corresponding to the image of the target object,
  • the pixel values of target object images with different brightness levels are different, so that multiple preview images can be generated during the process of increasing the brightness, and multiple preview images can be generated during the process of reducing the brightness.
  • the terminal 201 can present these preview images to the user 101 for the user 101 to select.
  • the terminal 201 can generate 4 preview images based on increasing the image brightness of the target object and reducing the image brightness of the target object, thereby obtaining 8 preview images of different brightness levels for the user 101 to perform. choose. In this way, the terminal 201 can support the user 101's two-way adjustment of brightening or darkening the image of the target object.
  • the terminal 201 may, in response to the selection operation of the user 101 on the plurality of different preview images, determine the pixel value of the image of the target object used to generate the preview image selected by the user 101, which is hereinafter referred to as is the target pixel value. In this way, the terminal 201 can adjust the pixel value of the image of the target object to the target pixel value to obtain an intermediate image.
  • the terminal 201 may fuse the intermediate image with the specified background image according to the mask corresponding to the imaging of the target object in the image to be processed, to generate a composite image. For example, the terminal 201 can determine the image area of the intermediate image in the composite image based on the mask, thereby replacing the image within the mask area in the specified background image with the intermediate image, thereby generating a composite image, etc.
  • the difference between the pixel values of the intermediate image and the pixel values of the specified background image is small, the difference in image brightness in different areas in the generated composite image is small, which makes the image content in the composite image more harmonious and image-like. quality higher.
  • the terminal 201 may be configured with a display device (such as a display screen, etc.), so that the terminal 201 can present the composite image to the user 101 through the display device.
  • a display device such as a display screen, etc.
  • the terminal 201 can continue to perform image synthesis for the user 101.
  • the terminal 201 can use the process of the embodiment shown in FIG. 4 to synthesize the image of the target object in the image to be processed and the specified background image.
  • the terminal 201 can autonomously adjust the image of the target object in the acquired new image to be processed based on the pixel value of the image of the target object used when generating the composite image (such as the above-mentioned target pixel value, etc.).
  • the terminal 201 can continuously collect images to be processed.
  • the terminal 201 can, based on the process of the embodiment shown in FIG. 4 above, collect the image to be processed based on the first image to be processed. And after the background image generates the composite image, the pixel value of the image of the target object in the second image to be processed can be adjusted according to the pixel value of the image of the target object in the composite image, so that the adjusted second image to be processed The brightness of the image of the target object in the image and the brightness of the image of the target object in the composite image are at the same brightness level, so the terminal 201 can generate a new composite image based on the adjusted image of the target object and the specified background image.
  • the terminal 201 can use the above-mentioned similar process to make corresponding adjustments. In this way, for the image of the target object in each image to be processed, the terminal 201 does not need to iteratively adjust its pixel value each time, thereby improving image synthesis efficiency and reducing resource consumption required for image synthesis.
  • the process of image synthesis performed by the terminal 201 is taken as an example for illustrative description.
  • the above method may also be executed by the computing device 301 deployed in the cloud, or the above method may be executed by the terminal 201 and the computing device 301 collaboratively.
  • the terminal 201 may first identify the image to be processed.
  • the mask corresponding to the image of the target object in the image to be processed is sent to the computing device 301 in the cloud for image synthesis and the like.
  • the pixel values of the pixels in the image of the target object on the R, G, and B channels are respectively adjusted as an example.
  • the target object can also be adjusted based on the above similar method. Adjust the grayscale value of each pixel in the image.
  • the image processing method provided by the embodiment of the present application is introduced above with reference to FIGS. 1 to 7 .
  • the functions of the image processing device provided by the embodiment of the present application and the computing device that implements the image processing device are introduced with reference to the accompanying drawings.
  • the image processing device 800 includes:
  • Acquisition module 801 used to obtain the image to be processed and the specified background image
  • Recognition module 802 used to identify the image of the target object in the image to be processed
  • the harmonization module 803 is configured to perform an image harmonization operation according to the image of the target object and the specified background image to obtain a composite image.
  • the image harmonization operation is used to instruct the adjustment of the pixel values according to the specified background image. The way in which the image of the target object is described;
  • Output module 804 is used to output the composite image.
  • the harmonization module 803 is used to:
  • the pixel value of the image of the target object is adjusted to obtain an intermediate image.
  • the difference between the pixel value of the intermediate image and the pixel value of the specified background image is less than the pixel value of the specified background image.
  • the intermediate image and the specified background image are synthesized to obtain a composite image.
  • the harmonization module 803 is used to:
  • the pixel values of the image of the target object are iteratively adjusted according to the difference to obtain the intermediate image.
  • the harmonization module 803 is used to:
  • the pixel value of the image of the target object is iteratively adjusted.
  • the harmonization module 803 is used to:
  • the pixel value of each pixel point is iteratively adjusted.
  • the iterative calculation curve includes iteration parameters, and the harmonization module 803 is also used to:
  • the image of the target object is input to the artificial intelligence model, and the iteration parameters corresponding to each pixel point in the image of the target object output by the artificial intelligence model are obtained.
  • the artificial intelligence model is constructed based on a convolutional neural network with a preset number of layers.
  • the image harmonization operation is used to instruct the pixel values of each pixel point in the image of the target object to be individually adjusted according to the pixel value of the specified background image.
  • the pixels of different pixel points There is a difference in how much the values are adjusted.
  • the image harmonization operation is used to instruct the pixel values of each pixel point in the first area in the image of the target object to be individually adjusted according to the pixel values of the specified background image, And uniformly adjust the pixel values of multiple pixels in the second area in the image of the target object.
  • the acquisition module is also used to acquire new images to be processed
  • the harmonization module is also configured to autonomously adjust the pixel values of the image of the target object in the new image to be processed according to the pixel values of the image of the target object in the composite image.
  • the image processing device 800 further includes:
  • Interaction module 805 used to present a configuration interface, the configuration interface includes the plurality of different preview images, each preview image is generated based on the specified background image and the image of the target object, and the method used to generate different preview images is The pixel values of the images of the target object are different, and in response to the user's selection operation on the plurality of different preview images, determining the target pixel value of the image of the target object used to generate the preview image selected by the user;
  • the harmonization module 803 is used to:
  • the device 800 further includes:
  • the generation module 806 is configured to record the pixel values of the pixels in the image of the target object during each round of iterative adjustment of the pixel values of the image of the target object, and adjust the pixel values of the pixels in the image of the target object in multiple rounds according to the iterative adjustment process. Iteratively adjust the pixel values in the process to generate the plurality of different preview images.
  • the target object is a participant participating in a video conference
  • the acquisition module 801 is used to:
  • the plurality of preview images are generated based on images of target objects with multiple brightness levels, and the pixel values of images of target objects with different brightness levels are different.
  • the image processing device 800 shown in Fig. 8 corresponds to the method shown in Fig. 4, the specific implementation manner of the image processing device 800 shown in Fig. 8 and its technical effects can be referred to the relevant information in the foregoing embodiments. Described above, no further details will be given here.
  • FIG. 9 is a schematic diagram of a computing device 900 provided by this application.
  • the computing device 900 may be, for example, the terminal 201 in the embodiment shown in FIG. 4 , etc.
  • the computing device 900 includes a processor 901 , a memory 902 , and a communication interface 903 .
  • the processor 901, the memory 902, and the communication interface 903 communicate through the bus 904, and the communication can also be achieved through other means such as wireless transmission.
  • the memory 902 is used to store instructions
  • the processor 901 is used to execute the instructions stored in the memory 902.
  • the computing device 900 may also include a memory unit 905, and the memory unit 905 may be connected to the processor 901, the storage medium 902, and the communication interface 903 through the bus 904.
  • the memory 902 stores program code, and the processor 901 can call the program code stored in the memory 902 to perform the following operations:
  • An image harmonization operation is performed according to the image of the target object and the specified background image to obtain a composite image.
  • the image harmonization operation is used to indicate a way to adjust the image of the target object according to the pixel value of the specified background image. ;
  • the processor 901 may be a CPU, and the processor 901 may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), or field programmable gate arrays. (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete device components, etc.
  • DSPs digital signal processors
  • ASICs application-specific integrated circuits
  • FPGA field programmable gate arrays.
  • a general-purpose processor can be a microprocessor or any conventional processor, etc.
  • the memory 902 may include read-only memory and random access memory, and provides instructions and data to the processor 901. Memory 902 may also include non-volatile random access memory. For example, memory 902 may also store device type information.
  • the memory 902 may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.
  • non-volatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable programmable read-only memory (erasable PROM, EPROM), electrically removable memory. Erase electrically programmable read-only memory (EPROM, EEPROM) or flash memory.
  • Volatile memory can be random access memory (RAM), which is used as an external cache.
  • RAM static random access memory
  • DRAM dynamic random access memory
  • SDRAM synchronous dynamic random access memory
  • Double data rate synchronous dynamic random access memory double data date SDRAM, DDR SDRAM
  • enhanced synchronous dynamic random access memory enhanced SDRAM, ESDRAM
  • synchronous link dynamic random access memory direct rambus RAM, DR RAM
  • the communication interface 903 is used to communicate with other devices connected to the computing device 900 .
  • the bus 904 may also include a power bus, a control bus, a status signal bus, etc.
  • the various buses are labeled bus 904 in the figure.
  • the computing device 900 may correspond to the image processing device 800 in the embodiment of the present application, and may correspond to the terminal 201 executing the method shown in FIG. 4 according to the embodiment of the present application, and the computing device
  • the above and other operations and/or functions implemented by 900 are respectively to implement the corresponding processes of each method in Figure 4. For the sake of simplicity, they will not be described again here.
  • An embodiment of the present application also provides a computer-readable storage medium.
  • the computer-readable storage medium may be any available medium that a computing device can store or a data storage device such as a data center that contains one or more available media.
  • the available media may be magnetic media (eg, floppy disk, hard disk, tape), optical media (eg, DVD), or semiconductor media (eg, solid state drive), etc.
  • the computer-readable storage medium includes instructions that instruct the computing device to perform the above-described image processing method.
  • An embodiment of the present application also provides a computer program product.
  • the computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computing device, the processes or functions described in accordance with the embodiments of the present application are generated in whole or in part.
  • the computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another, e.g., the computer instructions may be transmitted over a wired connection from a website, computer, or data center. (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) to another website, computer or data center.
  • a wired connection such as coaxial cable, optical fiber, digital subscriber line (DSL)
  • wireless such as infrared, wireless, microwave, etc.
  • the computer program product may be a software installation package. If it is necessary to use any of the foregoing image processing methods, the computer program product may be downloaded and executed on the computing device.
  • the above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination.
  • the above-described embodiments may be implemented in whole or in part in the form of a computer program product.
  • the computer program product includes one or more computer instructions.
  • the processes or functions described in the embodiments of the present application are generated in whole or in part.
  • the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.
  • the computer instructions may be stored in or transmitted from one computer-readable storage medium to another, e.g., the computer instructions may be transferred from a website, computer, server, or data center Transmission to another website, computer, server or data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) means.
  • the computer-readable storage medium may be any available medium that a computer can access, or a data storage device such as a server or a data center that contains one or more sets of available media.
  • the usable media may be magnetic media (eg, floppy disk, hard disk, tape), optical media (eg, DVD), or semiconductor media.
  • the semiconductor medium may be a solid state drive.

Landscapes

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

Abstract

一种图像处理方法,包括:图像处理装置获取待处理图像以及指定背景图像,并在待处理图像中识别目标对象的图像,从而图像处理装置根据目标对象的图像和指定背景图像执行图像和谐化操作,得到合成图像并输出该合成图像,其中,图像和谐化操作用于指示根据该指定背景图像的像素值调整目标对象的图像的方式。如此,根据目标对象的图像和指定背景图像执行图像和谐化操作,可以使得合成图像中目标对象的图像与指定背景图像之间更加和谐,从而提高生成的合成图像的质量、提升图像合成效果。并且,针对像素点的独立调整方式则可以根据像素点的差异进行调整,可以避免目标对象的图像中存在部分像素点过亮或者过暗的问题。

Description

图像处理方法、装置及相关设备 技术领域
本申请涉及人工智能技术领域,尤其涉及一种图像处理方法、装置及相关设备。
背景技术
在人工智能领域(artificial intelligent,AI)领域中,尤其是在视频会议、形象设计、海报设计等场景中,通常会将一张图像中的部分图像内容与另一张图像中的部分图像内容进行合成,生成新的图像。比如,在视频会议场景中,用户可以将视频图像中除人物图像之外的其它图像内容替换成指定的指定背景图像,不仅可以利用虚拟背景隐藏人物周围环境,保护用户隐私,还可以通过设置不同指定背景图像提高用户视频会议体验。
但是,实际应用时,基于虚拟背景和人物图像所合成得到的新图像,容易出现图像质量较差、图像合成效果较低的问题。因此,如何提高合成图像的质量,提升图像合成效果,成为亟需解决的重要问题。
发明内容
本申请提供了一种图像处理法,以提高合成图像的质量、提升图像合成效果。此外,本申请还提供了对应的装置、计算设备、计算机可读存储介质以及计算机程序产品。
第一方面,本申请提供一种图像处理方法,该方法可以由图像处理装置执行,具体的,图像处理装置获取待处理图像以及指定背景图像,该待处理图像中包括目标图像的对象,该目标对象可以人物、物体或者其他类型的对象等;然后,图像处理装置在待处理图像中识别目标对象的图像,例如通过图像分割算法识别目标对象的图像等,从而图像处理装置根据目标对象的图像和指定背景图像执行图像和谐化操作,得到合成图像,其中,图像和谐化操作用于指示根据该指定背景图像的像素值调整目标对象的图像的方式,这样,所得到的合成图像中的目标对象的图像与指定背景图像更加和谐;最后,图像处理装置输出该合成图像。
由于图像处理装置在合成图像的过程中,根据目标对象的图像和指定背景图像执行图像和谐化操作,减少了目标对象的图像与指定背景图像的像素值之间的差异,也即减少了在合成图像中目标对象的图像与指定背景图像之间的亮度差异,这使得在合成图像中目标对象的图像与指定背景图像之间更加和谐,从而可以提高生成的合成图像的质量、提升图像合成效果。
进一步地,在执行图像和谐化操作的过程中,可以对目标对象的图像中的各个像素点的像素值进行独立调整。由于不同像素点的像素值的调整幅度可以存在差异,统一调整方式对该图像中的所有像素点的像素值进行调整会导致调整后的目标对象的图像存在部分像素点过亮或者过暗的问题,而针对像素点的独立调整方式则可以根据像素点的差异进行调整,如此,可提高合成图像的质量,提升图像合成的效果。
在一种可能的实施方式中,图像处理装置在执行图像和谐化操作时,具体可以是根据指定背景图像的像素值,调整目标对象的图像的像素值,得到中间图像,其中,所调整得到的中间图像的像素值与指定背景图像的像素值之间的差值,小于目标对象的图像的像素值与指定背景图像的像素值之间的差值,以此减少目标对象的图像与指定背景图像之间的像素值差异,从而图像处理装置合成中间图像以及指定背景图像,得到合成图像。如此,可以实现减小目标对象的图像与指定背景图像之间的亮度差异,从而可以使得所生成的合成图像更加和 谐,提高合成图像的质量。
在一种可能的实施方式中,图像处理装置所执行的图像和谐化操作用于指示根据该指定背景图像的像素值对目标对象的图像中各个像素点的像素值进行单独调整,不同像素点的像素值的调整幅度存在差异。针对各个像素点进行差异化调整,可以避免调整后的目标对象的图像存在部分像素点过亮或者过暗的问题,以此可以有效提高合成图像的质量,提升图像合成的效果。
在一种可能的实施方式中,图像处理装置所执行的图像和谐化操作用于指示根据该述指定背景图像的像素值对目标对象的图像中第一区域内的各个像素点的像素值进行单独调整,并对目标对象的图像中第二区域内的多个像素点的像素值进行统一调整。如此,可以减少调整目标对象的图像的像素值所需的计算量,实现在生成较高质量的合成图像的同时,降低资源消耗。
在一种可能的实施方式中,图像处理装置还可以获取新的待处理图像,从而根据合成图像中目标对象的图像的像素值,对新的待处理图像中目标对象的图像进行自主调节。如此,在后续生成的合成图像与之前所生成的合同图像中,目标对象的图像均可以保持相近的亮度,从而无需用户再进行手动调整。
在一种可能的实施方式中,图像处理装置在生成中间图像时,具体可以是先计算目标对象的图像的像素值与指定背景图像的像素值之间的差值,从而当该差值大于预设范围时,根据该差值迭代调整目标对象的图像的像素值,得到中间图像。如此,通过对目标对象的图像进行迭代调整像素值,可以实现逐渐减小目标对象的图像的像素值与指定背景图像的像素值的差异,并且目标对象的图像能够达到较好的亮度调节效果。
在一种可能的实施方式中,图像处理装置在迭代调整目标对象的图像的像素值时,具体可以是根据目标对象的图像与指定背景图像之间的差值确定迭代方向以及迭代次数,从而根据该迭代方向以及该迭代次数,对目标对象的图像的像素值进行迭代调整。如此,可以实现对目标对象的图像的迭代调整。
在一种可能的实施方式中,图像处理装置在根据迭代方向以及迭代次数,对目标对象的图像的像素值进行迭代调整时,具体可以是根据迭代方向,确定目标对象的图像中各个像素点对应的迭代计算曲线,从而根据迭代次数以及各个像素点对应的迭代计算曲线,对各个像素点的像素值进行迭代调整。如此,针对各个像素点进行差异化调整,可以避免调整后的目标对象的图像存在部分像素点过亮或者过暗的问题,以此可以有效提高合成图像的质量,提升图像合成的效果。
在一种可能的实施方式中,图像处理装置调整各个像素点的像素值所采用的迭代计算曲线包括迭代参数,则,图像处理装置在确定各个像素点所对应的迭代计算曲线时,可以将目标对象的图像输入至人工智能模型,得到人工智能模型输出的目标对象的图像中各个像素点对应的迭代参数,并根据该各个像素点对应的迭代参数确定相应的迭代计算曲线。
在一种可能的实施方式中,用于确定迭代参数的人工智能模型基于预设层数的卷积神经网络完成构建,如基于5层卷积神经网络模型进行构建,如此可以通过尺寸较小的人工智能模型确定各个像素点对应的迭代参数,减小图像处理装置的计算开销。
在一种可能的实施方式中,图像处理装置还可以呈现包括多个不同的预览图像的配置界面,其中,每个预览图像基于指定背景图像以及目标对象的图像生成的,并且,生成不同预览图像所采用的目标对象的图像的像素值不同,从而图像处理装置响应于用户针对多个不同的预览图像的选择操作,确定用于生成用户选择的预览图像的目标对象的图像的目标像素值。 则,图像处理装置在生成中间图像时,可以调整目标对象的图像的像素值为该目标像素值,得到中间图像。这样,最终所生成的合成图像中,目标对象的图像的亮度即为该用户所选择的亮度,如此,可以提高用户对于目标对象的图像亮度进行选择的自由度,提高用户体验。
在一种可能的实施方式中,图像处理装置所呈现的多个预览图像基于多个亮度等级的目标对象的图像生成,不同亮度等级的目标对象的图像之间像素值不同。如此,可以方便用户根据自己的需要选择相应亮度等级的目标对象的图像来生成合成图像,从而使得所生成的合成图像更容易符合用户的预期,提高用户体验。
在一种可能的实施方式中,图像处理装置还可以记录每轮迭代调整目标对象的图像的像素值的过程中该目标对象的图像中像素点的像素值,并根据该目标对象的图像中像素点在多轮迭代调整过程中的像素值,生成多个不同的预览图像。如此,可以基于多轮迭代调整过程中的像素值,确定多张方便用户参考亮度调节的预览图像,以此提高用户对于亮度选择的自由度。
在一种可能的实施方式中,在视频会议场景中,目标对象具体为参加该视频会议的与会人员,从而图像处理装置在获取待处理图像时,具体可以是利用拍摄装置对该与会人员进行拍摄,获得包括与会人员图像的待处理图像。
实际应用时,图像处理装置也可以是适用于其他场景中,用户生成其他场景中的合成图像。
第二方面,本申请提供一种图像处理装置,所述图像处理装置包括用于执行第一方面或第一方面任一种可能实现方式中的图像处理方法的各个模块。
第三方面,本申请提供一种计算设备,所述计算设备包括处理器、存储器和显示器。所述处理器、所述存储器进行相互的通信。所述处理器用于执行存储器中存储的指令,以使得计算设备执行如第一方面或第一方面的任一种实现方式中的图像处理方法。需要说明的是,该存储器可以集成于处理器中,也可以是独立于处理器之外。计算设备还可以包括总线。其中,处理器通过总线连接存储器。其中,存储器可以包括可读存储器以及随机存取存储器。
第四方面,本申请提供一种计算机可读存储介质,所述计算机可读存储介质中存储有指令,当其在计算设备上运行时,使得计算设备执行上述第一方面或第一方面的任一种实现方式所述图像处理方法的操作步骤。
第五方面,本申请提供了一种包含指令的计算机程序产品,当其在计算设备上运行时,使得计算设备执行上述第一方面或第一方面的任一种实现方式所述图像处理方法的操作步骤。
本申请在上述各方面提供的实现方式的基础上,还可以进行进一步组合以提供更多实现方式。
附图说明
图1为本申请提供的一示例性应用场景示意图;
图2为合成图像中用户101的图像亮度远远高于指定背景图像的亮度的示意图;
图3为本申请提供的另一示例性应用场景示意图;
图4为本申请提供的一种图像处理方法的流程示意图;
图5为本申请提供的不同像素点对应不同迭代计算曲线的示意图;
图6为本申请提供的4张目标对象的图像亮度逐渐增大的预览图像示意图;
图7为本申请提供的目标对象的图像亮度逐渐增大、以及逐减小的预览图像示意图;
图8为本申请提供的一种图像处理装置的结构示意图;
图9为本申请提供的一种计算设备的硬件结构示意图。
具体实施方式
下面将结合本申请实施例中的附图,对本申请中的技术方案进行描述。
参见图1,为本申请实施例提供的一示例性应用场景示意图。如图1所示,不同用户之间可以通过各自使用的终端进行在线视频会议。具体地,终端201上配置有摄像头,并且,该摄像头用于连续采集用户101的图像,获得包括用户101的多帧视频图像;同时,终端201上配置有麦克风以及扬声器,其中,麦克风用于录入用户101产生的语音,该扬声器用于播放其它用户(如用户102等产生的语音)。然后,终端201将多帧视频图像与用户101产生的语音通过网络300发送给终端202,以便由终端202在显示界面呈现该多帧视频图像,并通过终端202的扬声器播放用户101的语音。同时,终端201可以在显示界面上呈现该摄像头所拍摄到的连续多帧视频图像,以便用户101查看自身在视频会议中的画面。在此过程中,终端201也会通过网络300接收到终端200所发送的包括用户102的多帧视频图像以及用户102的语音,并在终端201的显示界面呈现包括用户102的多帧视频图像、通过终端201上的扬声器播放用户102的语音,以此实现用户101与用户102的在线视频会议。
实际应用时,用户101可以设置用户101在视频会议画面中的指定背景图像,相应的,用户102在终端202的显示界面所查看到的视频图像中的人物背景即为用户101所设置的指定背景图像。比如,用户101可以将视频会议画面中的指定背景图像设置为图2所示的虚拟背景,从而不仅可以利用该虚拟背景隐藏用户101周围的环境信息,保护用户101隐私,而且,用户101可以按照自身喜好设置不同指定背景图像,以此提高用户101参加视频会议的体验。但是,如果终端201直接将视频图像中用户101的图像与指定背景图像进行合成,则容易导致合成图像中的用户101的图像与指定背景图像不和谐,如图2所示的合成图像中,用户101的图像亮度远远高于指定背景图像的亮度,这就导致合成图像的质量较低、图像合成效果较差。
基于此,本申请实施例提供了一种图像处理方法,用于提高合成的图像的质量、提升图像合成效果。具体地,该方法可以由图像处理装置执行,该图像处理装置获取待处理图像以及指定背景图像,该待处理图像中包括用户101的图像,而指定背景图像可以是用户101指定的图像,从而图像处理装置先从待处理图像中识别用户101的图像,并根据用户101的图像和指定背景图像执行图像和谐化操作,得到合成图像,该图像和谐化操作用于指示根据指定背景图像的像素值调整目标对象的图像的方式,以降低指定背景图像的像素值与目标对象的图像的像素值之间的差异(也即使得目标对象的图像与指定背景图像更加和谐)。最后,终端201输出合成图像,例如可以在终端201的显示界面呈现该合成图像,以及将该合成图像发送给终端202进行呈现等。
由于图像处理装置在合成图像的过程中,根据用户101的图像和指定背景图像执行图像和谐化操作,具体是对用户101的图像的像素值进行了调整,减少了其与指定背景图像的像素值之间的差异,也即减少了在合成图像中用户101的图像与指定背景图像之间的亮度差异,这使得图像处理装置所生成的合成图像中,用户101的图像与指定背景图像之间更加和谐,从而可以提高生成的合成图像的质量、提升图像合成效果。
进一步地,在调整用户101的图像的像素值的过程中,可以对用户101的图像中的各个像素点的像素值进行独立调整。由于不同像素点的像素值的调整幅度可以存在差异,统一调整方式对该图像中的所有像素点的像素值进行调整会导致调整后的用户101的图像存在部分像 素点过亮或者过暗的问题,而针对像素点的独立调整方式则可以根据像素点的差异进行调整,如此,可提高合成图像的质量,提升图像合成的效果。
示例性地,图像处理装置可以通过虚拟设备实现,例如可以是通过虚拟机、容器、计算引擎中的至少一种实现等。或者,图像处理装置可以通过包括处理器的物理设备实现,其中,处理器可以是CPU,以及专用集成电路(application-specific integrated circuit,ASIC)、可编程逻辑器件(programmable logic device,PLD)、复杂程序逻辑器件(complex programmable logical device,CPLD)、现场可编程门阵列(field-programmable gate array,FPGA)、通用阵列逻辑(generic array logic,GAL)、片上系统(system on chip,SoC)、软件定义架构(software-defined infrastructure,SDI)芯片、人工智能(artificial intelligence,AI)芯片等任意一种处理器或其任意组合。并且,图像处理装置中所包括的处理器的数量可以是一个,也可以是多个,具体可以根据实际应用的业务需求设定处理器数量,本实施例对此并不进行限定。
值得注意的是,上述图1所示的应用场景仅作为一种示例性说明,实际应用时,本申请实施例提供的图像处理方法也可以应用于其他可适用的场景。比如,在图3所示的应用场景中,可以由部署于云端的一个或者多个计算设备301(如服务器等)实现将待处理图像与指定背景图像进行合成,此时,图像处理装置可以是计算设备301。相应地,终端201负责将采集到的包括用户101的多帧视频图像以及用户101指定的指定背景图像发送给云端的计算设备301;计算设备301在生成合成图像后,将可以将该合成图像发送给终端201以及终端202,以便将该合成图像分别呈现给用户201以及用户202。或者,终端201与计算设备301可以协同执行上述图像处理方法等;又或者,终端201不局限于为用户101提供视频会议服务,还可以为用户101提供其它服务,如用户101可以通过终端201生成合成图像的方式实现形象设计、海报设计等,本申请对此并不进行限定。
为便于理解,下面结合附图,对本申请提供的图像处理方法的实施例进行描述。
参见图4,图4为本申请实施例提供的一种图像处理方法的流程示意图。其中,图4所示的图像处理方法可以由相应的图像处理装置执行,该图像处理装置可以部署于用户侧,例如,在图1所示的应用场景中,图像处理装置可以是用户侧的终端201;或者,图像处理装置可以部署于云端,例如,图3所示的应用场景中,图像处理装置可以是计算设备301。为便于说明,本实施例中以应用于图1所示的应用场景为例进行示例性说明。
基于图1所示的应用场景,图4所示的图像处理方法具体可以包括:
S401:终端201获取待处理图像以及指定背景图像。
其中,该待处理图像中包括目标对象的图像。本实施例中,目标对象,可以是人物,如在图1所示的视频会议场景中,目标对象可以是用户101等;或者,目标对象,可以是物体,如在海报设计场景中,目标对象可以是生产厂商生产的商品等,以便对其进行广告宣传等。
本实施例中,当用户101存在图像合成的需求时,终端201获取参与图像合成所需的待处理图像以及指定背景图像。
作为一种实现示例,终端201上可以配置有摄像装置,并且,终端201可以利用该摄像装置对用户101(即目标对象)进行拍摄,得到包括用户201图像的一帧或者多帧待处理图像。同时,终端201还可以向用户101呈现交互界面,该交互界面包括多种背景图像,以便用户101对该多种背景图像进行选择。其中,该多种指定背景图像可以预先由终端201从网络中下载至本地,或者由技术人员将其配置于终端201中等。为便于区分,本实施例中将用户所选择的背景图像称之为指定背景图像。相应的,终端201可以利用用户101所选择的指定背景图像与摄 像装置所采集到的其中一帧待处理图像进行图像合成。
而在另一种实现示例中,终端201可以向用户101呈现交互界面,从而用户101在该交互界面上将待处理图像与指定背景图像导入终端201等,本实施例中,对于终端201获取待处理图像以及指定背景图像的具体实现方式并不进行限定。
S402:终端201在待处理图像中识别目标对象的图像。
由于待处理图像中关于目标对象的图像需要参与后续的图像合成,而该待处理图像中的其余图像内容并不会与指定背景图像进行合成,因此,终端201在获取到待处理图像后,可以从待处理图像中提取出目标对象的图像。
在一种可能的实施方式中,终端201可以利用目标分割算法识别出待处理图像中的目标对象,并进一步确定出该目标对象在图像中的图像区域。其中,目标分割算法,例如可以是具有卷积神经网络特征的掩码区域(mask regions with convolution neural networks features,mask R-CNN)算法中的至少一种。
举例来说,当基于mask R-CNN算法对待处理图像进行目标检测时,终端201可以将待处理图像输入至基于mask R-CNN算法构建得到的目标分割模型,得到该目标分割模型输出的目标对象在待处理图像中的图像所对应的掩码(mask),该掩码用于指示目标对象在图像中的图像区域,从而利用该掩码能够从待处理图像中提取出目标对象的图像。其中,掩码,是一种图像滤镜的模板,当从图像中分割出部分图像信息时,可以通过矩阵对图像进行像素过滤,以此实现从图像中分割出该部分图像信息,该矩阵即为掩码。
本实施例中,对于终端201确定目标对象的图像的具体实现方式并不进行限定。
S403:终端201根据目标对象的图像和指定背景图像执行图像和谐化操作,得到合成图像,该图像和谐化操作用于指示根据指定背景图像的像素值调整目标对象的图像的方式。
实际应用场景中,由于待处理图像与指定背景图像通常是在不同的拍摄条件下采集得到的,因此,待处理图像的像素值(或者称之为亮度)与指定背景图像的像素值之间的差值超出预设范围,也即目标对象的图像的像素值与指定背景图像的像素值之间的差值超出预设范围。此时,如果终端201直接将目标对象的图像与指定背景图像进行合成,则会因为目标对象的图像的亮度与指定背景图像的亮度差异过大而导致最终生成的合成图像不和谐,影响图像合成的效果。
因此,本实施例中,终端201可以根据指定背景图像的像素值对目标对象的图像进行图像和谐化操作,以使得目标对象的图像与指定背景图像在亮度上更加和谐。
具体地,终端201在获得目标对象的图像后,可以对目标对象的图像的像素值进行调整,为便于区分与描述,以下将经过像素值调整后的目标对象的图像称之为中间图像,中间图像的像素值与指定背景图像的像素值之间差值,小于调整之前目标对象的图像的像素值与指定背景图像的像素值之间的差值,即通过调整像素值的方式降低目标对象的图像的亮度与指定背景图像的亮度之间的差异。这样,后续再对调整后的目标对象的图像与指定背景图像进行合成,可以使得最终生成的合成图像因为不同部分的图像内容之间的亮度差异较小而更加和谐,以此提高图像合成的效果。
本实施例中,终端201可以通过迭代调整目标对象的图像的像素值的方式,逐渐减小目标对象的图像与指定背景图像之间的亮度差异。具体地,由于通常情况下,目标对象的图像与指定背景图像均包括多个像素点,因此,终端201可以计算目标对象的图像中所有像素点的像素值的平均值(或者中位数),并将该平均值(或者中位数)作为目标对象的图像的像素值。本实施例中,像素点的像素值,是指该像素点在红(red,R)、绿(green,G)、蓝(blue, B)三个颜色通道的值;相应的,图像的像素值(如前述目标对象的图像的像素值、指定背景图像的像素值等)即为RGB三个颜色通道的值,并且,该图像在每个颜色通道的值,为该图像中的所有像素点在该颜色通道的值的平均值(或者中位数)。类似地,终端201可以将指定背景图像中的所有像素点的像素值的平均值(或者中位数)作为指定背景图像的像素值。然后,终端201可以先根据目标对象的图像,计算出目标对象的图像的像素值与指定背景图像的像素值之间的差值,并根据该差值,迭代调整目标对象的图像的像素值,得到中间图像,直至迭代调整像素值的次数达到预设次数,或者目标对象的图像的像素值与指定背景图像的像素值之间的差值小于预设范围。比如,假设采用百分制表示各个通道的像素值(取值范围为0~1),目标对象的图像的像素值为(0.1,0.2,0.1),指定背景图像的像素值为(0.5,0.6,0.6),其R通道的像素值之差为0.4、G通道的像素值之差为0.4、B通道的像素值之差为0.5,以调整目标对象的图像在R通道的像素值为例,终端201根据该差值,可以对目标对象的图像在R通道的像素值进行4次迭代调整,比如,在4次迭代过程中目标对象的图像在R通道的像素值可以依次被调整为0.3、0.4、0.45、0.48,从而经过4次迭代调整后所得到的中间图像在R通道的像素值(0.48)与指定背景图像在R通道的像素值(0.5)之间的差值更小。其中,G通道、B通道的像素值的调整与其类似,在此不做赘述。
示例性地,在迭代调整目标对象的图像的像素值的过程中,终端201可以先根据目标对象的图像的像素值与指定背景图像的像素值之间的差值,确定迭代方向以及迭代次数。其中,迭代方向用于指示对目标对象的图像的像素值大小的调整方向,包括增大像素值的方向、减小像素值的方向,其可以根据差值的正负进行确定。比如,当该差值小于0时,表征目标对象的图像的像素值小于指定背景图像的像素值,从而终端201可以确定沿着增大像素值的方向,迭代调整目标对象的图像的像素值;而当该差值大于0时,表征目标对象的图像的像素值大于指定背景图像的像素值,从而终端201可以确定沿着减小像素值的方向,迭代调整目标对象的图像的像素值。迭代次数,用于指示对目标对象的图像的像素值大小的调整次数,迭代次数可以根据差值的绝对值大小进行确定。比如,当目标对象的图像与指定背景图像在R通道的像素值之间的差值的绝对值在[0.1,0.2)之间,迭代次数为1;差值的绝对值在[0.2,0.3)之间,迭代次数为2;差值的绝对值在[0.3,0.4)之间,迭代次数为3;差值的绝对值在0.4以上,迭代次数为4等。实际应用时,可以由技术人员预先在终端201中配置差值的绝对值与迭代次数之间的对应关系,从而终端201在计算得到目标对象的图像的像素值与指定背景图像的像素值之间的差值后,可以根据该差值的绝对值大小查找该对应关系,以确定迭代次数。在其它实现方式中,也可以是由用户101在交互界面上指定迭代方向以及迭代次数等,本实施例对此并不进行限定。
实际应用时,终端201可以是在目标对象的图像的像素值与指定背景图像的像素值之间的差值大于预设范围时(比如在R通道上的两个像素值之间的差值超出-0.1~0.1的范围),终端201对目标对象的图像的像素值进行一次或者多次迭代调整;而当目标对象的图像的像素值与指定背景图像的像素值之间的差值不超过该预设范围时(比如在R通道上的两个像素值之间的差值在出-0.1~0.1范围内),终端201可以不对目标对象的图像的像素值进行调整,即终端201可以直接将目标对象的图像与指定背景图像进行合成。
本实施例中,终端201对目标对象的图像所执行的图像和谐化操作,可以包括以下两种实现方式。
在第一种实现方式中,图像和谐化操作用于指示根据指定背景图像的像素值对目标对象的图像中各个像素点的像素值进行单独调整,不同像素点的像素值的调整幅度存在差异、不 同像素点的迭代方向也可以存在差异。比如,像素点A的像素值的调整幅度为10%,而像素点A的像素值的调整幅度为25%等。为如此,可以有效避免部分像素点调整幅度过大而导致该像素点过亮或者过暗的问题。
在第二种实现方式中,图像和谐化操作用于指示根据指定背景图像的像素值对目标对象的图像中第一区域内的各个像素点的像素值进行单独调整,并对该目标对象的图像中第二区域内的多个像素点的像素值进行统一调整,其中,第一区域与第二区域为不重叠的图像区域。以目标对象为与会人员为例,针对图像中与会人员对应的的头发图像区域、衣服图像区域等(如上述第二区域),该区域图像内的各个像素点之间的像素值差异通常较小,因此,终端201可以对该区域内的多个像素点的像素值进行统一调整,如将该区域内的多个像素点的像素值统一增大20%等。而对于该目标对象的图像中的其它区域(如上述第一区域),该区域内的不同像素点的像素值差异较大,因此,终端201可以对该区域内的各个像素点的像素值进行单独调整。
为便于理解,本实施例提供了以下两种调整像素点的像素值的实现方式。
实现方式一,终端201可以基于各个像素点所对应的迭代计算曲线,对该像素点的像素值进行迭代调整,并且多轮迭代过程中,各个像素点分别对应的迭代计算曲线不变。
在每次迭代过程中,终端201可以根据调整像素值的迭代方向,确定目标对象的图像中的各个像素点分别对应的迭代计算曲线,不同像素点的迭代计算曲线可以存在差异。然后,终端201可以根据上述确定出的迭代次数,利用每个像素点对应的迭代计算曲线对该像素点的像素值进行一次或者多次的迭代计算,从而终端201可以将该像素点的像素值调整为将最后一次迭代计算所得到的像素值,以此实现对各个像素点的像素值的调整。
比如,针对像素点在各个通道上的像素值,终端201可以根据迭代方向为每个像素点所确定的迭代计算曲线如下述公式(1)所示:
其中,xi为迭代调整之前目标对象的图像中第i个像素点的像素值,yi为迭代调整之后目标对象的图像中第i个像素点的像素值;α为,其中,当迭代方向为增大像素值的方向时,α取值大于0,而当迭代方向为减小像素值的方向时,α取值小于0,并且,不同像素点对应的迭代计算曲线中α的取值可以不同,从而不同像素点对应的迭代计算曲线可以存在差异。如图5所示,像素点1至像素点5分别对应于不同的迭代计算曲线。
举例来说,假设终端201确定迭代次数为2,针对目标对象的图像中第1个像素点,其像素值为m,并且,终端201为该像素点确定出的迭代参数的值为α1,则该像素点经过第一次迭代后的像素值为在经过第二次迭代后的像素值类似地,目标对象的图像中的其余像素点经过两次迭代后的像素值,也可以通过上述类似方式进行计算得到。
值得注意的是,通过对目标对象的图像中各个像素点单独调整像素值(即不同像素点所对应的迭代参数可以不同),而并非是对所有像素点进行统一的调整(如将所有像素点的像素值均增加0.1等),这可以有效避免部分像素点过亮或者过暗的问题。比如,对于目标对象的图像中的像素点A以及像素点B,其在迭代之前R通道的像素值分别为0.5、0.1,则,在经过两次迭代后,像素点A在R通道的像素值可以被调整为0.55、像素点B在R通道的像素值可以被调整为0.45,以此可以避免像素点A在R通道的像素值过大而导致目标对象的图像中像素点A 过亮。其中,终端201可以基于上述类似方式,对各个像素点在G通道、B通道上的像素值分别进行调整。
其中,每个像素点分别对应的迭代计算曲线中的迭代参数,可以利用预先完成训练的人工智能(artificial intelligence,AI)模型进行推理得到。具体实现时,终端201可以将目标对象的图像输入至AI模型中,并由该AI模型输出参数矩阵,该参数矩阵中的每个元素即为目标对象的图像中各个像素点对应的迭代参数,并且,不同像素点所对应的迭代参数的大小(以及正负)可以存在差异。如此,可以基于上述公式(1),根据各个像素点对应的迭代参数进一步确定出各个像素点对应的迭代计算曲线。
实际应用时,AI模型可以输出两个参数矩阵,分别为用于增大像素点亮度的参数矩阵a以及用于减小像素点亮度的参数矩阵b。其中,参数矩阵a中的每个元素即为目标对象的图像中各个像素点对应的迭代参数,并且,基于参数矩阵a中的迭代参数可以增大各个像素点的像素值(即将目标对象的图像调亮);参数矩阵b中的每个元素即为目标对象的图像中各个像素点对应的迭代参数,并且,基于参数矩阵b中的迭代参数可以减小各个像素点的像素值(即将目标对象的图像调暗)。这样,终端201可以根据目标对象的图像的像素值与指定背景图像的像素值,选择利用参数矩阵a还是利用参数矩阵b对目标对象的图像的像素值进行迭代调整。
或者,终端201中可以配置有两个AI模型,其中,AI模型A用于输出上述参数矩阵a,AI模型B用于输出上述参数矩阵b,从而终端201可以终端201可以根据目标对象的图像的像素值与指定背景图像的像素值之间的差值,确定迭代方向,进而根据该迭代方向选择相应的AI模型为目标对象的图像确定符合该迭代方向的参数矩阵,以便利用该参数矩阵对目标对象的图像的像素值进行迭代调整。
作为一种实现示例,上述AI模型,可以基于预设层数的卷积神经网络完成构建。例如,可以基于包括五层卷积神经网络的视觉愉悦的照明增强网络(visual pleasing lighting enhancement-net,VPLE-Net)构建出上述AI模型。
实现方式二,终端201可以基于各个像素点所对应的迭代计算曲线,对该像素点的像素值进行迭代调整,并且不同轮的迭代过程中,每个像素点所采用的迭代计算曲线不同。
与上述实现方式一不同的是,终端201在每次迭代过程中,均为目标对象的图像中各个像素点重新确定在本轮迭代过程中的迭代参数,即在不同轮的迭代过程中,每个像素点的迭代参数可以存在差异。
以两次迭代为例,在第一轮迭代过程中,终端201将目标对象的图像输入至AI模型中,得到该AI模型输出的矩阵参数I,该矩阵参数I中的每个元素即为目标对象的图像中各个像素点在第一轮迭代过程中各自对应的迭代参数,从而基于上述公式(1),可以根据各个像素点对应的迭代参数进一步确定出各个像素点对应的迭代计算曲线I,以便根据各个像素点对应的迭代计算曲线I计算出各个像素点经过第一轮迭代后的像素值,从而基于更新后的各个像素点的像素值,生成临时图像。
而在第二轮迭代过程中,终端201可以将该临时图像输入至该AI模型中,该得到该AI模型输出的矩阵参数II,该矩阵参数II中的每个元素即为目标对象的图像中各个像素点在第二轮迭代过程中各自对应的迭代参数,从而基于上述公式(1),可以根据各个像素点对应的迭代参数进一步确定出各个像素点对应的迭代计算曲线II,以便根据各个像素点对应的迭代计算曲线II计算出各个像素点经过第二轮迭代后的像素值,从而基于各个像素点在第二轮迭代后的像素值,生成上述中间图像。
实现方式三,针对目标对象的图像中的一部分区域,终端201可以基于该区域内的各个像 素点所对应的迭代计算曲线,对该像素点的像素值进行迭代调整,并且不同轮的迭代过程中,每个像素点所采用的迭代计算曲线可以相同或者不同;而针对目标对象的图像中的剩余部分区域,,终端201可以基于该区域内的其中一个像素点所对应的迭代计算曲线,对该区域内的所有像素点的像素值进行统一的迭代调整,并且不同轮的迭代过程中,该区域内的所有像素点所采用的迭代计算曲线可以相同或者不同。
可以理解,上述各实现方式仅作为一些示例性说明,并不用于限定终端201生成中间图像的具体过程。比如,在其它可能的实现方式中,终端201可以根据目标对象的图像的像素值与指定背景图像的像素值之间的差值,对目标对象的图像中的各个像素点的像素值进行一次调整(可以不用多次迭代调整),并且,不同像素点的像素值的调整幅度可以存在差异,以此可以直接生成中间图像等。
本实施例中,终端201可以通过上述迭代过程自动生成中间图像,而在其它可能的实现方式中,用户101也可以人工干预像素值的调整,从而终端201可以基于用户101的操作,生成中间图像。
作为一种实现示例,终端201在迭代调整目标对象的图像的像素值的过程中,可以记录经过每一轮迭代后目标对象的图像中像素点的像素值,并且,基于该轮迭代后的像素点的像素值,可以得到一张临时图像,从而基于多轮迭代过程可以得到多张临时图像。然后,终端201可以将多张临时图像分别与指定背景图像进行融合,本实施例中,将融合得到的图像称之为预览图像,以此可以得到基于多张临时图像所分别生成的多张预览图像,并且,不同预览图像中目标对象的图像的像素值不同。然后,终端201可以将生成的多张预览图像呈现给用户101,以便由用户101对多张预览图像进行选择。实际应用时,用户101所选择的预览图像,通常为用户101觉得该多张预览图像中合成效果最和谐的预览图像。以增大目标对象的图像亮度为例,如图6所示,终端201可以将4张预览图像呈现给用户101,该4张预览图像中目标对象的图像的像素值存在差异,如图6所示,4张预览图像中目标对象的图像逐渐变亮,从而用户101可以从这4张预览图像中选择一张其认为最和谐的预览图像。
在进一步的实施方式中,终端201可以沿着增大亮度以及减小亮度两个方向分别对目标对象的图像的像素值进行迭代调整,每次迭代调整对应于目标对象的图像的一个亮度等级,不同亮度等级的目标对象图像之间像素值不同,从而可以在增大亮度过程中生成多张预览图像,并在减小亮度的过程中生成多张预览图像。然后,终端201可以将这些预览图像均呈现给用户101,以供用户101进行选择。如图7所示,终端201可以基于增大目标对象的图像亮度以及减小目标对象的图像亮度,分别生成4张预览图像,从而可以得到8个不同亮度等级的预览图像,以供用户101进行选择。如此,终端201可以支持用户101对于目标对象的图像的变亮或者变暗的双向调节。
然后,终端201可以响应于用户101针对该多个不同的预览图像的选择操作,确定用于生成用户101所选择的预览图像的目标对象的图像的像素值,为便于区分,以下将其称之为目标像素值。这样,终端201可以目标对象的图像的像素值调整为该目标像素值,得到中间图像。
在生成中间图像后,终端201可以将根据目标对象在待处理图像中的成像所对应的掩码,将中间图像与指定背景图像进行融合,生成合成图像。比如,终端201可以根据该掩码,确定中间图像在合成图像中的图像区域,从而利用中间图像替换指定背景图像中该掩码区域内的图像,以此生成合成图像等。
由于中间图像的像素值与指定背景图像的像素值之间差异较小,因此,所生成的合成图像中,不同区域内的图像亮度差异较小,这使得合成图像中的图像内容更加和谐、图像质量 更高。
S404:终端201输出合成图像。
作为一种实现示例,终端201上可以配置有显示装置(如显示屏等),从而终端201可以通过该显示装置将合成图像呈现给用户101。
在部分应用场景中,如视频会议场景中,终端201可以持续为用户101进行图像合成。在此过程中,针对每个待处理图像,终端201均可以采用上述图4所示实施例的过程,将待处理图像中的目标对象的图像与指定背景图像进行合成。或者,终端201可以根据上述生成合成图像时所采用的目标对象的图像的像素值(如上述目标像素值等),对获取的新的待处理图像中的目标对象的图像进行自主调节。具体地,终端201可以持续采集待处理图像,以连续采集到第一待处理图像以及第二待处理图像为例,终端201可以基于上述图4所示实施例的过程,基于第一待处理图像以及背景图像生成合成图像后,可以根据该合成图像中的目标对象的图像的像素值,对第二待处理图像中的目标对象的图像的像素值进行调整,以使得调整后的第二待处理图像中的目标对象的图像的亮度与合成图像中的目标对象的图像亮度为同一亮度等级,从而终端201可以基于调整后的目标对象的图像与指定背景图像进行合成,生成新的合成图像。类似地,针对后续采集的其它待处理图像,终端201均可以采用上述类似过程进行相应的调整。如此,针对每个待处理图像中的目标对象的图像,终端201可以不用每次重新迭代调整其像素值,以此可以提高图像合成效率、减小图像合成所需的资源消耗。
值得注意的是,上述实施例中,是以终端201执行图像合成的过程为例进行示例性说明。在其它实施例中,也可以是由部署于云端的计算设备301执行上述方法,或者由终端201以及计算设备301协同执行上述方法,如终端201在接收到待处理图像后,可以先识别出该待处理图像中的目标对象的图像所对应的掩码,并将待处理图像、掩码以及指定背景图像发送给云端的计算设备301进行图像合成等。另外,本实施例中,是以分别调整目标对象的图像中的像素点在R、G、B通道上的像素值为例,在其它实施例中,也可以是基于上述类似方式,对目标对象的图像中各个像素点的灰度值进行调整。
值得注意的是,本领域的技术人员根据以上描述的内容,能够想到的其他合理的步骤组合,也属于本申请的保护范围内。其次,本领域技术人员也应该熟悉,说明书中所描述的实施例均属于优选实施例,所涉及的动作并不一定是本申请所必须的。
以上结合图1至图7对本申请实施例提供的图像处理方法进行介绍,接下来结合附图对本申请实施例提供的图像处理装置的功能以及实现该图像处理装置的计算设备进行介绍。
参见图8,示出了一种图像处理装置的结构示意图,该图像处理装置800包括:
获取模块801,用于获取待处理图像以及指定背景图像;
识别模块802,用于在所述待处理图像中识别所述目标对象的图像;
和谐化模块803,用于根据所述目标对象的图像和所述指定背景图像执行图像和谐化操作,得到合成图像,所述图像和谐化操作用于指示根据所述指定背景图像的像素值调整所述目标对象的图像的方式;
输出模块804,用于输出所述合成图像。
在一种可能的实施方式中,所述和谐化模块803,用于:
根据所述指定背景图像的像素值,调整所述目标对象的图像的像素值,得到中间图像,所述中间图像的像素值与所述指定背景图像的像素值之间的差值,小于所述目标对象的图像的像素值与所述指定背景图像的像素值之间的差值;
合成所述中间图像以及所述指定背景图像,得到合成图像。
在一种可能的实施方式中,所述和谐化模块803,用于:
计算所述目标对象的图像的像素值与所述指定背景图像的像素值之间的差值;
当所述差值大于预设范围时,根据所述差值迭代调整所述目标对象的图像的像素值,得到所述中间图像。
在一种可能的实施方式中,所述和谐化模块803,用于:
根据所述差值确定迭代方向以及迭代次数;
根据所述迭代方向以及所述迭代次数,对所述目标对象的图像的像素值进行迭代调整。
在一种可能的实施方式中,所述和谐化模块803,用于:
根据所述迭代方向,确定所述目标对象的图像中各个像素点对应的迭代计算曲线;
根据所述迭代次数以及所述各个像素点对应的迭代计算曲线,对所述各个像素点的像素值进行迭代调整。
在一种可能的实施方式中,所述迭代计算曲线包括迭代参数,所述和谐化模块803,还用于:
将所述目标对象的图像输入至人工智能模型,得到所述人工智能模型输出的所述目标对象的图像中各个像素点对应的迭代参数。
在一种可能的实施方式中,所述人工智能模型基于预设层数的卷积神经网络完成构建。
在一种可能的实施方式中,所述图像和谐化操作用于指示根据所述指定背景图像的像素值对所述目标对象的图像中各个像素点的像素值进行单独调整,不同像素点的像素值的调整幅度存在差异。
在一种可能的实施方式中,所述图像和谐化操作用于指示根据所述指定背景图像的像素值对所述目标对象的图像中第一区域内的各个像素点的像素值进行单独调整,并对所述目标对象的图像中第二区域内的多个像素点的像素值进行统一调整。
在一种可能的实施方式中,所述获取模块,还用于获取新的待处理图像;
所述和谐化模块,还用于根据所述合成图像中所述目标对象的图像的像素值,对所述新的待处理图像中所述目标对象的图像的像素值进行自主调节。
在一种可能的实施方式中,所述图像处理装置800还包括:
交互模块805,用于呈现配置界面,所述配置界面包括所述多个不同的预览图像,每个预览图像基于所述指定背景图像以及所述目标对象的图像生成的,生成不同预览图像所采用的目标对象的图像的像素值不同,并响应于用户针对所述多个不同的预览图像的选择操作,确定用于生成所述用户选择的预览图像的所述目标对象的图像的目标像素值;
则,所述和谐化模块803,用于:
调整所述目标对象的图像的像素值为所述目标像素值,得到所述中间图像。
在一种可能的实施方式中,所述装置800还包括:
生成模块806,用于记录每轮迭代调整所述目标对象的图像的像素值的过程中所述目标对象的图像中像素点的像素值,并根据所述目标对象的图像中像素点在多轮迭代调整过程中的像素值,生成所述多个不同的预览图像。
在一种可能的实施方式中,所述目标对象为参加视频会议的与会人员,所述获取模块801,用于:
利用拍摄装置对所述与会人员进行拍摄,获得所述待处理图像。
在一种可能的实施方式中,所述多个预览图像基于多个亮度等级的目标对象的图像生成,不同亮度等级的目标对象的图像之间像素值不同。
由于图8所示的图像处理装置800对应于图4所示的方法,故图8所示的图像处理装置800的具体实现方式及其所具有的技术效果,可以参见前述实施例中的相关之处描述,在此不做赘述。
图9为本申请提供的一种计算设备900的示意图,该计算设备900例如可以是上述图4所示实施例中的终端201等。
如图9所示,所述计算设备900包括处理器901、存储器902、通信接口903。其中,处理器901、存储器902、通信接口903通过总线904进行通信,也可以通过无线传输等其他手段实现通信。该存储器902用于存储指令,该处理器901用于执行该存储器902存储的指令。进一步的,计算设备900还可以包括内存单元905,还内存单元905可以通过总线904与处理器901、存储介质902以及通信接口903连接。其中,该存储器902存储程序代码,且处理器901可以调用存储器902中存储的程序代码执行以下操作:
获取待处理图像以及指定背景图像;
在所述待处理图像中识别所述目标对象的图像;
根据所述目标对象的图像和所述指定背景图像执行图像和谐化操作,得到合成图像,所述图像和谐化操作用于指示根据所述指定背景图像的像素值调整所述目标对象的图像的方式;
输出所述合成图像。
应理解,在本申请实施例中,该处理器901可以是CPU,该处理器901还可以是其他通用处理器、数字信号处理器(DSP)、专用集成电路(ASIC)、现场可编程门阵列(FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立器件组件等。通用处理器可以是微处理器或者是任何常规的处理器等。
该存储器902可以包括只读存储器和随机存取存储器,并向处理器901提供指令和数据。存储器902还可以包括非易失性随机存取存储器。例如,存储器902还可以存储设备类型的信息。
该存储器902可以是易失性存储器或非易失性存储器,或可包括易失性和非易失性存储器两者。其中,非易失性存储器可以是只读存储器(read-only memory,ROM)、可编程只读存储器(programmable ROM,PROM)、可擦除可编程只读存储器(erasable PROM,EPROM)、电可擦除可编程只读存储器(electrically EPROM,EEPROM)或闪存。易失性存储器可以是随机存取存储器(random access memory,RAM),其用作外部高速缓存。通过示例性但不是限制性说明,许多形式的RAM可用,例如静态随机存取存储器(static RAM,SRAM)、动态随机存取存储器(DRAM)、同步动态随机存取存储器(synchronous DRAM,SDRAM)、双倍数据速率同步动态随机存取存储器(double data date SDRAM,DDR SDRAM)、增强型同步动态随机存取存储器(enhanced SDRAM,ESDRAM)、同步连接动态随机存取存储器(synchlink DRAM,SLDRAM)和直接内存总线随机存取存储器(direct rambus RAM,DR RAM)。
该通信接口903用于与计算设备900连接的其它设备进行通信。该总线904除包括数据总线之外,还可以包括电源总线、控制总线和状态信号总线等。但是为了清楚说明起见,在图中将各种总线都标为总线904。
应理解,根据本申请实施例的计算设备900可对应于本申请实施例中的图像处理装置800,并可以对应于执行根据本申请实施例中图4所示方法中的终端201,并且计算设备900所实现的上述和其它操作和/或功能分别为了实现图4中的各个方法的相应流程,为了简洁,在此不再赘述。
本申请实施例还提供了一种计算机可读存储介质。所述计算机可读存储介质可以是计算设备能够存储的任何可用介质或者是包含一个或多个可用介质的数据中心等数据存储设备。所述可用介质可以是磁性介质,(例如,软盘、硬盘、磁带)、光介质(例如,DVD)、或者半导体介质(例如固态硬盘)等。该计算机可读存储介质包括指令,所述指令指示计算设备执行上述图像处理方法。
本申请实施例还提供了一种计算机程序产品。所述计算机程序产品包括一个或多个计算机指令。在计算设备上加载和执行所述计算机指令时,全部或部分地产生按照本申请实施例所述的流程或功能。
所述计算机指令可以存储在计算机可读存储介质中,或者从一个计算机可读存储介质向另一计算机可读存储介质传输,例如,所述计算机指令可以从一个网站站点、计算机或数据中心通过有线(例如同轴电缆、光纤、数字用户线(DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机或数据中心进行传输。
所述计算机程序产品可以为一个软件安装包,在需要使用前述图像处理方法的任一方法的情况下,可以下载该计算机程序产品并在计算设备上执行该计算机程序产品。
上述实施例,可以全部或部分地通过软件、硬件、固件或其他任意组合来实现。当使用软件实现时,上述实施例可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载或执行所述计算机程序指令时,全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以为通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中,或者从一个计算机可读存储介质向另一个计算机可读存储介质传输,例如,所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集合的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质(例如,软盘、硬盘、磁带)、光介质(例如,DVD)、或者半导体介质。半导体介质可以是固态硬盘。
以上所述,仅为本申请的具体实施方式,但本申请的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本申请揭露的技术范围内,可轻易想到各种等效的修改或替换,这些修改或替换都应涵盖在本申请的保护范围之内。因此,本申请的保护范围应以权利要求的保护范围为准。

Claims (16)

  1. 一种图像处理方法,其特征在于,所述方法包括:
    获取待处理图像以及指定背景图像;
    在所述待处理图像中识别目标对象的图像;
    根据所述目标对象的图像和所述指定背景图像执行图像和谐化操作,得到合成图像,所述图像和谐化操作用于指示根据所述指定背景图像的像素值调整所述目标对象的图像的方式;
    输出所述合成图像。
  2. 根据权利要求1所述的方法,其特征在于,所述根据所述目标对象的图像和所述指定背景图像执行图像和谐化操作,得到合成图像,包括:
    根据所述指定背景图像的像素值,调整所述目标对象的图像的像素值,得到中间图像,所述中间图像的像素值与所述指定背景图像的像素值之间的差值,小于所述目标对象的图像的像素值与所述指定背景图像的像素值之间的差值;
    合成所述中间图像以及所述指定背景图像,得到合成图像。
  3. 根据权利要求2所述的方法,其特征在于,所述图像和谐化操作用于指示根据所述指定背景图像的像素值对所述目标对象的图像中各个像素点的像素值进行单独调整,不同像素点的像素值的调整幅度存在差异。
  4. 根据权利要求2所述的方法,其特征在于,所述图像和谐化操作用于指示根据所述指定背景图像的像素值对所述目标对象的图像中第一区域内的各个像素点的像素值进行单独调整,并对所述目标对象的图像中第二区域内的多个像素点的像素值进行统一调整。
  5. 根据权利要求1至4任一项所述的方法,其特征在于,所述方法还包括:
    获取新的待处理图像;
    根据所述合成图像中所述目标对象的图像的像素值,对所述新的待处理图像中所述目标对象的图像的像素值进行自主调节。
  6. 根据权利要求1至5任一项所述的方法,其特征在于,所述方法还包括:
    呈现配置界面,所述配置界面包括所述多个不同的预览图像,每个预览图像基于所述指定背景图像以及所述目标对象的图像生成的,生成不同预览图像所采用的目标对象的图像的像素值不同;
    响应于用户针对所述多个不同的预览图像的选择操作,确定用于生成所述用户选择的预览图像的所述目标对象的图像的目标像素值;
    则,所述调整所述目标对象的图像的像素值,得到中间图像,包括:
    调整所述目标对象的图像的像素值为所述目标像素值,得到所述中间图像。
  7. 根据权利要求6所述的方法,其特征在于,所述多个预览图像基于多个亮度等级的目标对象的图像生成,不同亮度等级的目标对象的图像之间像素值不同。
  8. 一种图像处理装置,其特征在于,所述装置包括:
    获取模块,用于获取待处理图像以及指定背景图像;
    识别模块,用于在所述待处理图像中识别目标对象的图像;
    和谐化模块,用于根据所述目标对象的图像和所述指定背景图像执行图像和谐化操作,得到合成图像,所述图像和谐化操作用于指示根据所述指定背景图像的像素值调整所述目标对象的图像的方式;
    输出模块,用于输出所述合成图像。
  9. 根据权利要求8所述的装置,其特征在于,所述和谐化模块,用于:
    根据所述指定背景图像的像素值,调整所述目标对象的图像的像素值,得到中间图像,所述中间图像的像素值与所述指定背景图像的像素值之间的差值,小于所述目标对象的图像的像素值与所述指定背景图像的像素值之间的差值;
    合成所述中间图像以及所述指定背景图像,得到合成图像。
  10. 根据权利要求9所述的装置,其特征在于,所述图像和谐化操作用于指示根据所述指定背景图像的像素值对所述目标对象的图像中各个像素点的像素值进行单独调整,不同像素点的像素值的调整幅度存在差异。
  11. 根据权利要求9所述的装置,其特征在于,所述图像和谐化操作用于指示根据所述指定背景图像的像素值对所述目标对象的图像中第一区域内的各个像素点的像素值进行单独调整,并对所述目标对象的图像中第二区域内的多个像素点的像素值进行统一调整。
  12. 根据权利要求8至11任一项所述的装置,其特征在于,
    所述获取模块,还用于获取新的待处理图像;
    所述和谐化模块,还用于根据所述合成图像中所述目标对象的图像的像素值,对所述新的待处理图像中所述目标对象的图像的像素值进行自主调节。
  13. 根据权利要求8至12任一项所述的装置,其特征在于,所述装置还包括:
    交互模块,用于呈现配置界面,所述配置界面包括所述多个不同的预览图像,每个预览图像基于所述指定背景图像以及所述目标对象的图像生成的,生成不同预览图像所采用的目标对象的图像的像素值不同,并响应于用户针对所述多个不同的预览图像的选择操作,确定用于生成所述用户选择的预览图像的所述目标对象的图像的目标像素值;
    则,所述和谐化模块,用于:
    调整所述目标对象的图像的像素值为所述目标像素值,得到所述中间图像。
  14. 根据权利要求3所述的装置,其特征在于,所述多个预览图像基于多个亮度等级的目标对象的图像生成,不同亮度等级的目标对象的图像之间像素值不同。
  15. 一种计算设备,其特征在于,包括处理器、存储器;
    所述处理器用于执行所述存储器中存储的指令,以使所述计算设备执行如权利要求1至7任一项所述方法的步骤。
  16. 一种计算机可读存储介质,其特征在于,包括指令,当其在计算设备上运行时,使得所述计算设备执行如权利要求1至7中任一项所述方法的步骤。
PCT/CN2023/079807 2022-06-20 2023-03-06 图像处理方法、装置及相关设备 WO2023246171A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210698109.1 2022-06-20
CN202210698109.1A CN117314803A (zh) 2022-06-20 2022-06-20 图像处理方法、装置及相关设备

Publications (1)

Publication Number Publication Date
WO2023246171A1 true WO2023246171A1 (zh) 2023-12-28

Family

ID=89283525

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/079807 WO2023246171A1 (zh) 2022-06-20 2023-03-06 图像处理方法、装置及相关设备

Country Status (2)

Country Link
CN (1) CN117314803A (zh)
WO (1) WO2023246171A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006042229A (ja) * 2004-07-30 2006-02-09 Nippon Hoso Kyokai <Nhk> 映像オブジェクト抽出装置、映像オブジェクト軌跡合成装置、その方法及びそのプログラム
CN102158644A (zh) * 2009-12-25 2011-08-17 卡西欧计算机株式会社 图像合成装置和图像合成方法
CN106157273A (zh) * 2015-03-30 2016-11-23 阿里巴巴集团控股有限公司 生成合成图片的方法及装置
CN115239844A (zh) * 2022-05-31 2022-10-25 北京都视科技有限公司 图像生成方法、装置以及存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006042229A (ja) * 2004-07-30 2006-02-09 Nippon Hoso Kyokai <Nhk> 映像オブジェクト抽出装置、映像オブジェクト軌跡合成装置、その方法及びそのプログラム
CN102158644A (zh) * 2009-12-25 2011-08-17 卡西欧计算机株式会社 图像合成装置和图像合成方法
CN106157273A (zh) * 2015-03-30 2016-11-23 阿里巴巴集团控股有限公司 生成合成图片的方法及装置
CN115239844A (zh) * 2022-05-31 2022-10-25 北京都视科技有限公司 图像生成方法、装置以及存储介质

Also Published As

Publication number Publication date
CN117314803A (zh) 2023-12-29

Similar Documents

Publication Publication Date Title
Ma et al. Deep guided learning for fast multi-exposure image fusion
CN107516319B (zh) 一种高精度简易交互式抠图方法、存储设备及终端
US8606042B2 (en) Blending of exposure-bracketed images using weight distribution functions
CN111986129B (zh) 基于多摄图像融合的hdr图像生成方法、设备及存储介质
US10074165B2 (en) Image composition device, image composition method, and recording medium
CN111445564B (zh) 人脸纹理图像生成方法、装置、计算机设备和存储介质
CN107730444A (zh) 图像处理方法、装置、可读存储介质和计算机设备
US9256950B1 (en) Detecting and modifying facial features of persons in images
JP5696783B2 (ja) 画像処理装置
CN110611768B (zh) 多重曝光摄影方法及装置
DE112019007550T5 (de) Automatisches segmentieren und anpassen von bildern
WO2021128593A1 (zh) 人脸图像处理的方法、装置及系统
CN112258380A (zh) 图像处理方法、装置、设备及存储介质
WO2018166170A1 (zh) 一种图像处理的方法、装置及智能会议终端
CN115314617A (zh) 图像处理系统及方法、计算机可读介质和电子设备
CN111383166B (zh) 处理待显示图像的方法及装置、电子设备、可读存储介质
US9092889B2 (en) Image processing apparatus, image processing method, and program storage medium
CN111815729B (zh) 一种实时美肤方法、装置、设备和计算机存储介质
JP6967065B2 (ja) 情報処理装置、プログラム、及び、情報処理方法
WO2023246171A1 (zh) 图像处理方法、装置及相关设备
JP5896204B2 (ja) 画像処理装置及びプログラム
CN111462158A (zh) 一种图像处理方法、装置及智能设备、存储介质
CN112435173A (zh) 一种图像处理和直播方法、装置、设备和存储介质
US20230031897A1 (en) Dynamic low lighting adjustment within a video communication system
CN112561822B (zh) 美颜方法、装置、电子设备及存储介质

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: 23825829

Country of ref document: EP

Kind code of ref document: A1