WO2017173578A1 - 一种图像增强方法及装置 - Google Patents

一种图像增强方法及装置 Download PDF

Info

Publication number
WO2017173578A1
WO2017173578A1 PCT/CN2016/078492 CN2016078492W WO2017173578A1 WO 2017173578 A1 WO2017173578 A1 WO 2017173578A1 CN 2016078492 W CN2016078492 W CN 2016078492W WO 2017173578 A1 WO2017173578 A1 WO 2017173578A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
region
interest
enhancement
mask
Prior art date
Application number
PCT/CN2016/078492
Other languages
English (en)
French (fr)
Inventor
朱聪超
陈刚
罗巍
张星
杜成
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
Priority to PCT/CN2016/078492 priority Critical patent/WO2017173578A1/zh
Priority to CN201680013568.4A priority patent/CN107851309A/zh
Publication of WO2017173578A1 publication Critical patent/WO2017173578A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/50—Depth or shape recovery

Definitions

  • the present invention relates to the field of image processing, and in particular, to an image enhancement method and apparatus.
  • An image is a projection of a three-dimensional world on a two-dimensional plane. Since the amount of data contained in an image is large, people are particularly concerned with a part of the image. And the brain has selective attention, which determines which part of the visual data is most interesting to humans.
  • the image enhancement algorithms in the terminal with camera or camera function use the image as a two-dimensional data matrix to uniformly enhance the entire image, resulting in a large computational burden and neglecting the depth information and saliency information in the three-dimensional world. .
  • the present invention provides an image enhancement method and device, which can enhance different regions in an image, can improve the layering of the photo from the effect, highlight the subject of the shooting, and enhance the user sense.
  • the area of interest and prevents some areas from being over-enhanced; it can reduce the computational burden of the image-enhancing algorithm in the prior art to process the entire image.
  • a first aspect of the present invention provides an image enhancement method, which is applied to a terminal, the method comprising: constructing a depth map of the image to generate a depth map of the image; and dividing the image according to the depth map of the image, Segmenting different depth regions of the image; performing saliency detection on the image, and using the depth region with high significance as a region of interest of the image; generating the The mask value of the image; the image is processed according to the mask value of the image.
  • the method for acquiring a region of interest of an image further includes: constructing a depth map of the image to generate a depth map of the image; and according to the depth map of the image Image Performing segmentation to segment different depth regions of the image; acquiring the depth region where the focus point is located, and using the depth region where the focus point is located as the region of interest.
  • the method for acquiring a region of interest of an image further includes: performing face detection on the image, and using a face region in the image as the region of interest of the image .
  • the step of generating a mask value of the image according to the region of interest of the image Specifically, the image is binary-divided according to the region of interest of the image, wherein a mask value of the region of interest of the image is 1, and a mask value of the non-interest region of the image is 0; The region of interest of the image and the adjacent boundary of the non-interest region of the image are smoothed, and the mask value of the non-interest region is obtained as 0 ⁇ Mask ⁇ 1.
  • the enhancement processing the larger the Mask value, the stronger the brightness enhancement, edge sharpening, contrast enhancement and color saturation enhancement of the filter kernel and parameters for the region; the smaller the Mask, the weaker the enhancement.
  • a second aspect of the present invention provides an image enhancement apparatus including an image understanding unit and an image processing unit, wherein the image understanding unit is configured to extract a region of interest of an image, and according to the image The region of interest generates a mask value of the image; the image processing unit is configured to process the image according to a mask value of the image.
  • the image understanding unit includes a depth map construction module, a saliency detection module or a focus capture module, and an image mask generation module; wherein the depth map construction module is configured to The image acquired by the terminal is configured to perform a depth map to generate a depth map of the image.
  • the saliency detection module is configured to perform saliency detection on the image, and segment the image according to the depth map of the image.
  • the focus acquisition module Separating different depth regions of the image, using the depth region with high significance as a region of interest; or the focus acquisition module acquiring a depth region of the image where the focus is located, and the focus point Where The depth region is used as the region of interest; the image mask generating module is configured to generate an image mask value according to the region of interest of the image.
  • the image understanding unit includes a face detection module and an image mask generation module; wherein the face detection module is configured to The image is subjected to face detection, and a face region in the image is used as a region of interest of the image; the image mask generating module is configured to generate the image mask according to the region of interest of the image value.
  • the image processing unit includes a distributed filter core and a parameter generator, contrast enhancement and brightness adjustment, and edge sharpening module and color saturation a degree enhancement module; wherein the distributed filter kernel and the parameter generator are configured to assign different filter kernels and parameters to different regions of the image according to the image-sensing region mask value, and give the image Different regions perform different processing by the contrast enhancement and brightness adjustment and edge sharpening module and the color saturation enhancement module; the contrast enhancement and brightness adjustment and edge sharpening module for enhancing contrast of the image , brightness and edge sharpening processing; the color saturation enhancement module for enhancing color saturation of the image.
  • a third aspect of the present invention provides a terminal, where the terminal includes:
  • a memory for storing instructions
  • the image is processed according to the mask value of the image.
  • a fourth aspect of the present invention provides a non-transitory computer readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a terminal, perform the following events:
  • the image is processed according to the mask value of the image.
  • the invention can enhance different regions in the image differently, can improve the layering of the photos from the effect, highlight the subject of the shooting, enhance the region of interest of the user, and prevent some regions from being over-enhanced;
  • FIG. 1 is a schematic structural diagram of an image enhancement apparatus according to an embodiment of the present invention.
  • FIG. 2a is a schematic structural diagram of an image enhancement apparatus according to another embodiment of the present invention.
  • FIG. 2b is a schematic diagram of an image understanding unit according to another embodiment of the present invention.
  • FIG. 2c is a schematic diagram of still another image understanding unit according to another embodiment of the present invention.
  • FIG. 3 is a schematic structural diagram of a terminal of an image enhancement apparatus according to an embodiment of the present disclosure
  • FIG. 4 is a schematic flowchart of an image enhancement method according to an embodiment of the present invention.
  • FIG. 5 is a schematic flowchart of an image enhancement method according to another embodiment of the present invention.
  • 6a is a schematic diagram of an image taken by a dual camera according to an embodiment of the present invention.
  • 6b is a schematic diagram of a depth map generated according to an image captured by a dual camera according to an embodiment of the present invention
  • FIG. 7a is a schematic diagram of an image before saliency detection according to an embodiment of the present invention.
  • FIG. 7b is a schematic diagram of generating a saliency image after a saliency detection according to an embodiment of the present invention.
  • FIG. 8a is a schematic diagram of an image after generating an image mask according to an embodiment of the present invention.
  • FIG. 8b is a schematic diagram of an image after smoothing an image after generating an image mask according to an embodiment of the present invention.
  • FIG. 9a is a schematic diagram of an image without image enhancement according to an embodiment of the present invention.
  • FIG. 9b is a schematic diagram of an image after image enhancement according to an embodiment of the present invention.
  • Embodiments of the present invention provide an image enhancement apparatus and method, which are capable of differently enhancing different regions in an image, thereby improving the layering of the photo from the effect, highlighting the subject of the photograph, and enhancing the region of interest of the user. And to prevent some areas from being over-enhanced; and it can reduce the computational burden of the image-enhancing algorithm in the prior art to process the entire image.
  • An image enhancement device may specifically be a camera module on a terminal such as a camera, a video camera, a mobile phone, a palmtop computer or a tablet computer.
  • FIG. 1 is a schematic structural diagram of an image enhancement apparatus according to an embodiment of the present invention.
  • the image enhancement device 1 includes an image understanding unit 11 and an image processing unit 12.
  • the image understanding unit 11 is used for the image acquired by the terminal, the region of interest of the image is extracted, and an image mask value is generated according to the region of interest of the image; the image processing unit 12 according to the image The image mask value is enhanced differently for different areas of the image.
  • region of interest ROI refers to the area to be processed from the processed image in the form of a box, a circle, an ellipse, an irregular polygon, or the like in machine vision and image processing.
  • the image includes a region of interest and a region of non-interest.
  • the image processing apparatus 1 includes an image understanding unit 11 and an image processing unit 12.
  • the image understanding unit 11 includes a depth map construction module 111, a significance detection module 112, and an image mask generation module 113.
  • the depth map construction module 111 is configured to acquire depth image information of the image captured by the mobile phone, and the commonly used depth image information acquisition may be acquired by two camera stereo vision technologies, or may be acquired by using laser, infrared, and other ranging technologies.
  • FIG. 6a is a double embodiment of the present invention
  • FIG. 6b is a schematic diagram of an image taken by a camera.
  • FIG. 6b is a schematic diagram of a depth map generated according to an image taken by a dual camera according to an embodiment of the present invention. As shown in Figure 6b, the brighter area in Figure 6b indicates the closer the distance between the scene and the camera.
  • the method for acquiring depth image information in the embodiment of the invention is not limited.
  • the saliency detection module 112 is configured to identify an area of interest in the image collected by the mobile phone, and generate a saliency map.
  • FIG. 7a is a schematic diagram of an image before saliency detection according to an embodiment of the present invention. After significant detection, a saliency image is generated as shown in FIG. 7b, and FIG. 7b is a saliency detection according to an embodiment of the present invention. After that, a schematic diagram of the salient image is generated. As shown in Figure 7b, the brighter areas indicate higher saliency.
  • the saliency detection module 112a After the saliency detection module 112a performs the saliency detection on the image collected by the mobile phone and generates the saliency image, the image is segmented according to the depth map, and the different depth regions of the image are segmented, and then the saliency image is higher in the saliency image.
  • the layer extracts the region of interest. For example, the depth map construction module 111 divides the image into three regions, and the third region is highly significant, and then the third region is the region of interest of the image.
  • FIG. 2b is a schematic diagram of an image understanding unit according to another embodiment of the present invention. As shown in Figure 2b,
  • the Image Understanding unit 11' includes a depth map construction module 111', a focus focus acquisition module 112', and an image mask generation module 113'.
  • the focus acquisition module 112 ′ obtains the depth region where the focus is located, and uses the depth region where the focus is located as the region of interest; that is, the focus acquisition module 112 ′ divides the image according to the depth map, and divides the image.
  • the different depth regions are then extracted based on the depth region where the focus is located, wherein the focus acquisition module obtains the focus point according to the focus algorithm, and then obtains the depth region of the image according to the focus point.
  • FIG. 2c is a schematic diagram of still another image understanding unit according to another embodiment of the present invention. As shown in Figure 2c,
  • the image understanding unit 11" includes a face detection module 111" and an image mask generation module Block 112".
  • the face detection module 111" performs face detection on the image collected by the mobile phone. Face detection refers to the image captured by the mobile phone and is searched to determine whether it contains a human face, and if so, returns the position, size and posture of a face. If the face detection module 111 ′′ detects a human face in the image collected by the mobile phone, the face region is the region of interest at this time. In the embodiment of the invention, based on the result of the face detection, the degree of enhancement of the face region is limited. To prevent the saturation of the face area from being too high, or to enlarge the face of the face when sharpening.
  • the functions of other modules in Fig. 2c are the same as those in Fig. 2a, and will not be described here.
  • the image understanding device acquires the region of interest of the image, and the manner of obtaining may be obtained by the saliency detection module 112 or the focus acquisition module 112 ′ or the face detection module 111 ′′, and then extracted according to the method.
  • the image of the region of interest, the image mask value is generated; or any combination of the above three modules, and the masking value of the image is obtained by selecting a certain module or a weighted average of the three modules; .
  • the image mask generating module 113 generates an image mask Mask value according to the extracted region of interest.
  • the range of the Mask Mask is 0 ⁇ Mask ⁇ 1.
  • the mask value of the generated image based on the region of interest of the image will be described below.
  • a masking value of the region of interest of the image is 1, and a mask value of the non-interest region of the image is 0;
  • the boundary transition between the region of interest and the region of non-interest is natural when enhanced, so the mask Mask of the region of interest and the region of non-interest is smoothed.
  • the region of interest of the image and the adjacent boundary of the non-interest region of the image are smoothed, and the mask value of the non-interest region is obtained as 0 ⁇ Mask ⁇ 1.
  • FIG. 8a is a schematic diagram of an image after generating an image mask according to an embodiment of the present invention
  • FIG. 8a is a schematic diagram of a mask of interest generated by FIG. 6a.
  • the white part in the figure is a region of interest
  • the black portion in the figure is a non-interest region.
  • the image after smoothing at the boundary between the black portion and the white portion is as shown in Fig. 8b.
  • FIG. 8b is a schematic diagram of an image after smoothing an image after generating an image mask according to an embodiment of the present invention. As shown in FIG.
  • the image processing unit 12 includes a distributed filter kernel and parameter generator 121, contrast enhancement and brightness adjustment, and edge sharpening module 122, and color saturation enhancement. Module 123.
  • a distributed filter and parameter generator (Distributed Kneral and Parameter Generator) 121 is configured to allocate different filter kernels and parameters to different regions according to the image mask Mask value, and to enhance contrast, brightness, and edge sharpening.
  • Module 122 and color saturation enhancement module 123 perform different enhancement processes on different regions in the image. Contrast enhancement, brightness and edge sharpening module 122 for brightness enhancement, edge sharpening, and contrast enhancement of an image at a predetermined Mask value; color saturation enhancement module 123 for color saturation of an image at a predetermined Mask value Enhanced.
  • Mask 0 area, no processing, maintaining the initial effect, or blurring, suppressing non-interest areas, further highlighting the subject;
  • Mask 1 area for brightness enhancement, edge sharpening, contrast enhancement, and color saturation enhancement ;
  • FIG. 9a is a schematic diagram of an image without image enhancement according to an embodiment of the present invention
  • the “love flower” sign in the figure is a region of interest of the user, the picture does not highlight the subject; and the sign and image The background is fused together and the image has no layering
  • FIG. 9b is a schematic diagram of an image after image enhancement according to an embodiment of the present invention. As shown in Fig. 9b, the sign and the background have a clear layering, and the subject in the figure is prominent, and the effect of image enhancement in different regions of the entire image is significantly different.
  • the image enhancement device in the embodiment of the invention can enhance different regions in the image differently, can improve the layering of the photo from the effect, highlight the subject of the shooting, enhance the region of interest of the user, and prevent certain regions from passing through. Enhancement; and can reduce the computational burden of the image processing algorithm in the prior art to process the entire image.
  • each module in the image enhancement apparatus 1 can be implemented by the processor as a whole, that is, the functions performed by the units of the image enhancement apparatus in the embodiment of the present invention may have a function.
  • the processor executes as shown in Figure 3.
  • the terminal 2 includes: at least one processor 21, a memory 23, a lens 24, and a communication interface 25 connected by a bus 22 and completing communication with each other, wherein:
  • the bus 22 can be an Industry Standard Architecture (ISA) bus, external Peripheral Component (PCI) bus or Extended Industry Standard Architecture (EISA) bus.
  • ISA Industry Standard Architecture
  • PCI Peripheral Component
  • EISA Extended Industry Standard Architecture
  • the bus 32 can be divided into an address bus, a data bus, a control bus, etc., for convenience of representation, only one thick line is shown in FIG. 3, but it does not mean that there is only one bus or one type of bus.
  • the memory 23 is for storing executable program code and corresponding data, the program code including computer operating instructions.
  • the memory 23 may include a high speed RAM memory, and may also include a nonvolatile memory.
  • the memory is at least used to store a depth map construction algorithm, a significance detection algorithm, a face detection algorithm, and an image mask generation algorithm and image. Enhance the algorithm.
  • a lens 24 for capturing images is provided.
  • the communication interface 25 is configured to implement data exchange between the terminal 2 and the outside world.
  • the processor 21 is configured to image the image captured by the lens 24, recognize and understand the content of the image, extract the region of interest, and generate an image mask Mask value; and use the image processing unit to generate an image mask Mask value according to the image processing unit.
  • the regions with different Mask values are assigned different filter kernels and parameters, and the filter kernel and parameters are processed differently according to the Mask values of different regions.
  • the layering of the image is significantly improved, highlighting the subject of the shooting.
  • the image enhancement method provided in the embodiment of the present invention is applied to the image enhancement device in the embodiment of the present invention, and can be directly applied to a camera having a photographing function, such as a camera, a mobile phone, and a palmtop computer.
  • the image enhancement method includes steps S401-S405:
  • Step S401 Perform an image on the depth map to generate a depth map of the image
  • the terminal performs enhancement processing on the image, which may be that the terminal directly performs enhancement processing on the captured image when photographing or capturing, or may enhance processing on the image collected by the non-terminal.
  • an image is collected by the terminal, and the image is enhanced.
  • Step S402 segment the image according to the depth map of the image, and segment different depths of the image. Degree area
  • Step S403 performing saliency detection on the image, and using the depth region with high significance as the region of interest of the image;
  • Step S404 Generate a mask value of the image according to the region of interest of the image
  • the step of generating a mask value of the image according to the region of interest of the image is specifically: performing binary segmentation on the image according to the region of interest of the image, wherein the region of interest of the image
  • the mask value is 1, the mask value of the non-region of interest of the image is 0; the adjacent region of the image and the non-interest region of the image are smoothed, thereby obtaining non-interest
  • Step S405 Processing the image according to the mask value of the image.
  • the brightness enhancement, edge sharpening, contrast enhancement, and color saturation enhancement are stronger; the smaller the Mask, the weaker the enhancement.
  • different enhancement processing is performed on different regions in the image, so that the image is more capable of highlighting the subject of the photograph, thereby improving the layering of the photograph as a whole, and reducing the image enhancement algorithm of the prior art to the entire image.
  • the image enhancement method includes:
  • S501a The terminal collects an image, and performs the depth map construction on the image to generate a depth map of the image.
  • the depth map can be derived from dual Camera (multicolor Color+Color, or monochrome Mono+Mono), obtained through binocular stereo vision technology, and can also be derived from laser, infrared and other ranging techniques.
  • dual Camera multicolor Color+Color, or monochrome Mono+Mono
  • the manner of obtaining the depth map is not limited in the embodiment of the present invention.
  • S501b The terminal collects an image and performs face detection on the image.
  • the existing image enhancement algorithm does not utilize the high-level semantic information of the image, only the processing of the two-dimensional image data is easy to over-enhance the face region, and an unnatural phenomenon occurs.
  • Low-level visual features of the image such as color, Texture, shape, and human description of the image and the judgment of the similarity between the images are generally based on the object, scene and behavioral features described by the image.
  • the high-level semantics of the image are the objects and behaviors expressed by the image. , scene or emotional color.
  • the embodiment of the present invention is based on the detection result of the face, limiting the degree of enhancement of the face region, preventing the saturation of the face region from being too high, or enlarging the flaw of the face region when sharpening.
  • S502a segment the image according to a depth map of the image, and segment different depth layer regions of the image;
  • the depth map reflects the distance between the camera and the scene. Based on the depth image, the image can be segmented into image regions of different depth levels.
  • FIG. 6a is an image acquired by the dual camera in the embodiment of the present invention. After segmentation according to the depth map of the image, the image is as shown in FIG. 6b, and the brighter the image, the closer the image is to the camera.
  • S502b taking an area where a face is in the image as a region of interest, and generating a mask value of the image;
  • S504 using the depth region where the focus point is located as the region of interest of the image, and further generating an image mask value according to the region of interest of the image;
  • the region of interest of the extracted image may be step S504 or step S502b; or the masking value of the image may be obtained by weighted averaging in any combination of the two steps;
  • a non-transitory computer readable storage medium storing one or more programs, the one or more programs including instructions, when executed by a terminal, the terminal performs the following event:
  • the image is processed according to the mask value of the image.
  • An image enhancement method provided by an embodiment of the present invention can differently enhance different regions in an image, can improve the layering of the photo from the effect, highlight the subject of the shooting, enhance the region of interest of the user, and prevent a certain These areas are over-enhanced; and the computational burden of the image-enhancing algorithm in the prior art for processing the entire image can be reduced.
  • the steps of a method or algorithm described in connection with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both.
  • the software module can be placed in random access memory (RAM), memory, read only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or technical field. Any other form of storage medium known.

Landscapes

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

Abstract

本发明提供一种图像增强方法及装置，应用于终端中，该方法首先提取图像的感兴趣区域；然后根据该图像的感兴趣区域，生成该图像蒙版值；最后根据该图像蒙版值对该图像中不同区域进行不同的增强处理。本发明能够对图像中的不同区域进行不同的增强，既能从效果上提升照片的层次感，突出拍摄的主体，增强用户感兴趣的区域，并防止某些区域过增强；又能减少现有技术中图像增强算法对整幅图像进行处理的计算负担。

Description

一种图像增强方法及装置 技术领域
本发明涉及图像处理领域，尤其涉及一种图像增强方法及装置。
背景技术
图像是三维立体世界在二维平面上的投影，由于图像蕴含的数据量很大，人们特别关心地往往只是图像中的一部分区域。且大脑具有选择性注意力(selective attention)，决定着视觉数据中哪一部分是人类最感兴趣的。
目前具有照相或摄像功能的终端中的图像增强算法均把图像作为二维数据矩阵，对整幅图像进行统一增强处理，导致算法计算负担大，并且忽略了三维世界中的深度信息和显著性信息。
发明内容
本发明为解决现有技术的不足，提供一种图像增强方法及装置，能够对图像中的不同区域进行不同的增强，既能从效果上提升照片的层次感，突出拍摄的主体，增强用户感兴趣的区域，并防止某些区域过增强；又能减少现有技术中图像增强算法对整幅图像进行处理的计算负担。
本发明第一方面提供一种图像增强方法，应用于终端中，该方法包括：将图像进行深度图构建，生成所述图像的深度图；根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；根据所述图像的感兴趣区域，生成所述图像的蒙版值；根据所述图像的蒙版值对所述图像进行处理。
在第一方面的第一种可能的实现方式中，获取图像的感兴趣区域的方法还包括：将图像进行深度图构建，生成所述图像的深度图；根据所述图像的深度图对所述图像 进行分割，分割出所述图像的不同深度区域；获取对焦点所在的所述深度区域，并将所述对焦点所在的深度区域作为感兴趣区域。
在第一方面的第二种可能的实现方式中，获取图像的感兴趣区域的方法还包括：对图像进行人脸检测，并将所述图像中的人脸区域作为所述图像的感兴趣区域。
结合第一方面，或者第一方面的第一种至第二种可能的实现方式，在第三种可能的实现方式中，根据所述图像的感兴趣区域，生成所述图像的蒙版值步骤具体为：根据所述图像的感兴趣区域对所述图像进行二值分割，其中所述图像的感兴趣区域的蒙版值为1，所述图像非感兴趣区域的蒙版值为0；将所述图像的感兴趣区域与所述图像的非感兴趣区域的相邻边界进行平滑处理，进而得到非感兴趣区域的蒙版值范围为0≤Mask＜1。
结合第一方面的第三种可能实现方式，在第四种可能实现的方式中，根据所述图像的蒙版值对所述图像进行处理步骤具体为：Mask＝0的区域，不进行处理，或者进行模糊化处理，抑制非感兴趣区域；Mask＝1的区域进行亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的一种或多种；0＜Mask＜1的区域，进行不同程度的增强处理，Mask值越大，滤波器核和参数对该区域的亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的效果越强；Mask越小，增强的程度越弱。通过对不同的蒙版值进行不同的增强处理，进而提升照片的层次感，突出拍摄的主体。
本发明第二方面提供了一种图像增强装置，该图像增强装置包括图像理解单元以及图像处理单元；其中，所述图像理解单元，用于提取出图像的感兴趣区域，并根据所述图像的感兴趣区域生成所述图像的蒙版值；所述图像处理单元，用于根据所述图像的蒙版值，对所述图像进行处理。
在第二方面的第一种可能实现方式中，图像理解单元包括深度图构建模块、显著性检测模块或对焦点获取模块以及图像蒙版生成模块；其中，所述深度图构建模块，用于将终端采集的图像进行深度图构建，生成所述图像的深度图；所述显著性检测模块，用于对所述图像进行显著性检测，并根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域，将显著性高的所述深度区域作为感兴趣区域；或者所述对焦点获取模块，获取对焦点所在的所述图像的深度区域，并将所述对焦点所在 的深度区域作为感兴趣区域；所述图像蒙版生成模块，用于根据图像的感兴趣区域生成图像蒙版值。
结合第二方面的第一种可能实现方式，在第二种可能实现方式中，所述图像理解单元包括人脸检测模块以及图像蒙版生成模块；其中，所述人脸检测模块，用于对图像进行人脸检测，并将所述图像中的人脸区域作为所述图像的感兴趣区域；所述图像蒙版生成模块，用于根据所述图像的感兴趣区域，生成所述图像蒙版值。
结合第二方面的第一种可能实现方式，在第三种可能实现方式中，所述图像处理单元包括分布式滤波器核与参数发生器、对比度增强和亮度调节以及边缘锐化模块以及色彩饱和度增强模块；其中，所述分布式滤波器核与参数发生器，用于根据所述图像感兴区域蒙版值给图像的不同区域分配不同的滤波器核和参数，并给所述图像的不同区域通过所述对比度增强和亮度调节以及边缘锐化模块以及所述色彩饱和度增强模块，进行不同的处理；所述对比度增强和亮度调节以及边缘锐化模块，用于增强所述图像的对比度、亮度以及边缘锐化处理；所述色彩饱和度增强模块，用于增强所述图像的色彩饱和度。
本发明第三发明提供一种终端，该终端包括：
显示器，用于显示图像；
存储器，用于存储指令；
处理器，用于调用存储在所述存储器中的指令以实现：
将图像进行深度图构建，生成所述图像的深度图；
根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
根据所述图像的感兴趣区域，生成所述图像的蒙版值；
根据所述图像的蒙版值对所述图像进行处理。
本发明第四方面提供一种存储一个或多个程序的非易失性计算机可读存储介质，所述一个或多个程序包括指令，所述指令当被终端执行时，终端执行以下事件：
将图像进行深度图构建，生成所述图像的深度图；
根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
根据所述图像的感兴趣区域，生成所述图像的蒙版值；
根据所述图像的蒙版值对所述图像进行处理。
本发明能够对图像中的不同区域进行不同的增强，既能从效果上提升照片的层次感，突出拍摄的主体，增强用户感兴趣的区域，并防止某些区域过增强；又能减少现有技术中图像增强算法对整幅图像进行处理的计算负担。
附图说明
图1为本发明实施例提供的一种图像增强装置结构示意图；
图2a为本发明另一实施例提供的一种图像增强装置结构示意图；
图2b为本发明另一实施例提供的一种图像理解单元示意图；
图2c为本发明另一实施例提供的又一种图像理解单元示意图；
图3为本发明实施例提供的一种图像增强装置的终端结构示意图；
图4为本发明实施例提供的一种图像增强方法流程示意图；
图5为本发明另一实施例提供的一种图像增强方法流程示意图；
图6a为本发明实施例提供的双摄像头拍摄的图像示意图；
图6b为本发明实施例提供的根据双摄像头拍摄的图像生成的深度图示意图；
图7a为本发明实施例提供的一张显著性检测前的图像示意图；
图7b为本发明实施例提供的一张显著性检测后，生成显著性图像的示意图；
图8a为本发明实施例提供的一张生成图像蒙版后的图像示意图；
图8b为本发明实施例提供的一张对生成图像蒙版后的图像进行平滑处理后的图像示意图；
图9a为本发明实施例提供的一张未经图像增强的图像示意图；
图9b为本发明实施例提供的一张经过图像增强后的图像示意图。
具体实施方式
下面通过附图和实施例，对本发明的技术方案做进一步的详细描述。
本发明的实施例提供一种图像增强装置及方法，该装置能够对图像中的不同区域进行不同的增强，既能从效果上提升照片的层次感，突出拍摄的主体，增强用户感兴趣的区域，并防止某些区域过增强；又能减少现有技术中图像增强算法对整幅图像进行处理的计算负担。
本发明实施例的一种图像增强装置，该装置具体可以为照相机、摄像机、手机、掌上电脑或平板电脑等终端上的照相模块。
下面以图1为例，对本发明实施例中的图像增强装置进行说明。图1为本发明实施例提供的一种图像增强装置结构示意图。如图1所示，图像增强装置1包括：图像理解单元11和图像处理单元12。
具体地，在上述的装置中，图像理解单元11用于终端采集的图像后，提取该图像的感兴趣区域，并且根据该图像的感兴趣区域生成图像蒙版值；图像处理单元12会根据该图像蒙版值对该图像的不同区域进行不同的增强处理。
需要说明的是，感兴趣区域ROI(Region of Interest)是指机器视觉、图像处理中，从被处理的图像以方框、圆、椭圆、不规则多边形等方式勾勒出需要处理的区域。图像包括感兴趣区域与非感兴趣区域。
本发明实施例通过对蒙版值不同的区域进行不同的增强处理，能够从整体上提高图像的层次感，突出拍摄的主体。
下面以终端为手机为例，即以具备拍照功能的手机为例对本发明实施例的图像处理装置进行详细说明。如图2所示，该图像处理装置1包括图像理解单元11以及图像处理单元12。
具体地，在上述的装置中，图像理解(Image Understanding)单元11包括深度图构建模块111、显著性检测模块112以及图像蒙版生成模块113。
深度图构建模块111，用于获取手机采集的图像的深度图像信息，常用的深度图像信息获取可以通过两个摄像头立体视觉技术获取，也可以用激光、红外等测距技术来获取。
下面以图6为例，对深度图像信息的获取进行说明。图6a为本发明实施例提供的双 摄像头拍摄的图像示意图，获取的深度图像如图6b所示；图6b为本发明实施例提供的根据双摄像头拍摄的图像生成的深度图示意图。如图6b所示，图6b中越亮的区域表示场景和摄像头的之间距离越近。
对此，需要说明的是，发明实施例对于深度图像信息的获取方法并不限定。
显著性(Saliency)检测模块112，用于识别手机采集的图像中用户感兴趣的区域，生成显著性图像(Saliency Map)。
需要说明的是，本发明实施例对于显著性检测的算法并不限定。
下面以图7为例，对显著性检测进行说明。图7a为本发明实施例提供的一张显著性检测前的图像示意图，经过显著性检测后，生成了显著性图像如图7b所示，图7b为本发明实施例提供的一张显著性检测后，生成显著性图像的示意图。如图7b所示，越亮的区域表示显著性越高。
显著性检测模块112a在对手机采集的图像进行显著性检测并生成显著性图像后，会根据深度图对图像进行分割，分割出图像的不同深度区域，然后基于上述显著性图像中显著性较高的层，提取感兴趣区域。例如：深度图构建模块111将图像分割成了三个区域，显著性高的是第三区域，那么第三区域就是该图像的感兴趣区域。
作为提取感兴趣区域的另外一种方法，图2b为本发明另一实施例提供的一种图像理解单元示意图。如图2b所示，
图像理解(Image Understanding)单元11'包括深度图构建模块111'、对焦点获取模块112'以及图像蒙版生成模块113'。其中，对焦点获取模块112'获取对焦点所在的深度区域，并将对焦点所在的深度区域作为感兴趣区域；也就是说，对焦点获取模块112'根据深度图对图像进行分割，分割出图像的不同深度区域，然后基于对焦点所在的深度区域，提取出感兴趣区域；其中，对焦点获取模块是根据对焦算法获取对焦点，进而根据该对焦点获取其所在图像的深度区域。
需要说明的是，图2b中的其它模块的功能与图2a中一致，在此不再赘述。
作为提取感兴趣区域的另外一种方法，图2c为本发明另一实施例提供的又一种图像理解单元示意图。如图2c所示，
图像理解(Image Understanding)单元11”包括人脸检测模块111”以及图像蒙版生成模 块112”。其中，人脸检测模块111”对手机采集的图像进行人脸检测(Face Detection)。人脸检测是指对于手机采集的图像，并对其进行搜索以确定其中是否含有人脸，如果是则返回一脸的位置、大小和姿态。人脸检测模块111”若检测到手机采集到的图像中有人脸，此时人脸区域为感兴趣区域。本发明实施例中基于人脸检测的结果，对人脸区域的增强程度进行限制，防止人脸区域饱和度过高，或者锐化时放大人脸区域的瑕疵。图2c中的其它模块的功能与图2a中一致，在此不再赘述。
在此，需要说明的是，图像理解装置中获取图像的感兴趣区域，获取的方式可以是通过显著性检测模块112或者对焦点获取模块112'或者人脸检测模块111”获得，然后根据提取出的图像感兴趣区域，生成图像蒙版值；也可以是上述三个模块的任意组合，通过选择某一个模块或者三个模块加权平均来获取图像的蒙版值；本发明实施例对此不作限定。
图像蒙版生成模块113，根据上述提取的感兴趣区域，生成图像蒙版Mask值。蒙版Mask的值域范围为0≤Mask≤1。
下面对根据图像的感兴趣区域，生成图像的蒙版值进行说明。
根据所述图像的感兴趣区域对所述图像进行二值分割，其中所述图像的感兴趣区域的蒙版值为1，所述图像非感兴趣区域的蒙版值为0；为了使后续图像增强时对感兴趣区域和非感兴趣区域的边界过渡自然，因此对感兴趣区域和非感兴趣区域的蒙版Mask进行平滑处理。将所述图像的感兴趣区域与所述图像的非感兴趣区域的相邻边界进行平滑处理，进而得到非感兴趣区域的蒙版值范围为0≤Mask＜1。
下面以图8为例对根据图像的感兴趣区域生成蒙版值的过程进行说明。图8a为本发明实施例提供的一张生成图像蒙版后的图像示意图，图8a是图6a生成的感兴趣蒙版示意图，图中白色部分是感兴趣区域，图中黑色部分是非感兴趣区域，在黑色部分与白色部分的边界处进行平滑处理后的图像如图8b所示。图8b为本发明实施例提供的一张对生成图像蒙版后的图像进行平滑处理后的图像示意图，如图8b所示，感兴趣区域的白色部分与非感兴趣区域的黑色部分平滑处理后有介于黑色与白色中间的灰色，这样上述两个区域的过渡更加自然，因此得到非感兴趣区域的蒙版值范围为0≤Mask＜1。
具体地，在上述的装置中，图像处理(Image Processing)单元12包括分布式滤波器核与参数发生器121、对比度增强和亮度调节以及边缘锐化模块122以及色彩饱和度增强 模块123。
分布式滤波器核与参数发生器(Distributed Kneral and Parameter Generator)121，用于根据上述图像蒙版Mask值给不同的区域分配不同的滤波器核和参数，并通过对比度增强、亮度以及边缘锐化模块122以及色彩饱和度增强模块123对图像中的不同区域进行不同的增强处理。对比度增强、亮度以及边缘锐化模块122，用于将在预定Mask值的图像进行亮度增强、边缘锐化以及对比度增强；色彩饱和度增强模块123用于将在预定Mask值的图像的色彩饱和度增强。
下面对根据图像蒙版Mask值对图像中不同的区域进行不同的增强处理进行说明。Mask＝0的区域，不进行处理，保持初始效果，或者进行模糊化处理，抑制非感兴趣区域，进一步突出主体；Mask＝1的区域进行亮度增强、边缘锐化、对比度增强以及色彩饱和度增强；0＜Mask＜1的区域，进行不同程度的增强处理，Mask值越大，滤波器核和参数对该区域的亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的效果越强；Mask越小，增强的程度越弱。
下面以图9为例，对本发明实施例的图像增强方法进行说明。图9a为本发明实施例提供的一张未经图像增强的图像示意图；如图9a所示，图中的“爱护花草”标牌是用户的感兴趣区域，图片没有突出拍摄主体；并且标牌与图像的背景是融合在一起的，图像没有层次感。图9b为本发明实施例提供的一张经过图像增强后的图像示意图。如图9b所示，标牌与背景有清晰的层次感，图中的拍摄主体突出，对于整个图像的不同的区域图像增强的效果有明显的区别。
本发明实施例中的图像增强装置能够对图像中的不同区域进行不同的增强，既能从效果上提升照片的层次感，突出拍摄的主体，增强用户感兴趣的区域，并防止某些区域过增强；又能减少现有技术中图像增强算法对整幅图像进行处理的计算负担。
基于现有的终端功能，上述图像增强装置1中各个模块的功能实现可以通过处理器进行整体的控制实现，也就是说上述本发明实施例中的图像增强装置各个单元所执行的功能可有一个处理器执行，如图3所示。该终端2包括：至少一个处理器21、存储器23、镜头24以及通信接口25通过总线22连接并完成相互间的通信，其中：
该总线22可以是工业标准体系结构(Industry Standard Architecture,ISA)总线、外部 设备互联(Peripheral Component,PCI)总线或扩展工业标准体系结构(Extended Industry Standard Architecture,EISA)总线等。该总线32可以分为地址总线、数据总线、控制总线等，为了便于表示，图3中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
存储器23用于存储可执行程序代码及相应的数据，该程序代码包括计算机操作指令。存储器23可能包括高速RAM存储器，也可能包括非易失性存储器，在本发明实施例中存储器至少用于存储深度图构建算法、显著性检测算法、人脸检测算法以及图像蒙版生成算法和图像增强算法。
镜头24，用于采集图像。
通信接口25，用于实现终端2与外界的数据交换。
处理器21用于将镜头24采集的图像，通过图像理解，识别和理解图像的内容并提取出感兴趣区域，生成图像蒙版Mask值；并利用图像处理单元根据生成图像蒙版Mask值，给Mask值不同的区域分配不同的滤波器核和参数，滤波器核和参数根据不同区域的Mask值进行不同的处理。
处理器21对不同的区域的Mask值处理方式为：Mask＝0的区域，不进行处理，保持初始效果，或者进行模糊化处理，抑制非感兴趣区域，进一步突出主体；Mask＝1的区域进行亮度增强、边缘锐化、对比度增强以及色彩饱和度增强；0＜Mask＜1的区域，进行不同程度的增强处理，Mask值越大，滤波器核和参数对该区域的亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的效果越强；Mask越小，增强的程度越弱。
终端采集的图像经过处理器21处理后，图像的层次感明显提升，突出了拍摄的主体。
本发明实施例所提供的图像增强方法，应用于本发明实施例中的图像增强装置，可直接应用于照相机、手机、掌上电脑等具有拍摄功能的终端。具体地，如图4所示，该图像增强方法包括步骤S401-S405：
步骤S401：将图像进行深度图构建，生成所述图像的深度图；
对此，需要说明的是，本发明实施例中终端对图像进行增强处理，可以是终端在拍照或者摄像时，对采集的图像直接进行增强处理；也可以是对非终端采集的图像进行增强处理。在本发明实施例中，以终端采集一张图像，并对该图像进行增强处理为例进行说明。
步骤S402：根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深 度区域；
步骤S403：将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
步骤S404：根据所述图像的感兴趣区域，生成所述图像的蒙版值；
在该步骤中，根据图像的感兴趣区域，生成所述图像的蒙版值步骤具体为：根据所述图像的感兴趣区域对所述图像进行二值分割，其中所述图像的感兴趣区域的蒙版值为1，所述图像非感兴趣区域的蒙版值为0；将所述图像的感兴趣区域与所述图像的非感兴趣区域的相邻边界进行平滑处理，进而得到非感兴趣区域的蒙版值范围为0≤Mask＜1。
步骤S405：根据所述图像的蒙版值对所述图像进行处理。
在该步骤中，根据所述图像的蒙版值对所述图像进行处理步骤具体为：Mask＝0的区域，不进行处理，或者进行模糊化处理，抑制非感兴趣区域；Mask＝1的区域进行亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的一种或多种；0＜Mask＜1的区域，进行不同程度的增强处理，Mask值越大，滤波器核和参数对该区域的亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的效果越强；Mask越小，增强的程度越弱。
本发明实施例由于对图像中不同区域进行了不同的增强处理，使得图像更能够突出拍摄的主体，从整体上提升了照片的层次感，同时减少了现有技术中图像增强算法对整幅图像进行处理的负担。
下面以图5为例，对本发明实施例的图像增强方法进行详细说明。如图5所示，图像增强方法包括：
S501a：终端采集一张图像，将所述图像进行深度图构建，生成所述图像的深度图；
深度图可以来自双Camera(多色Color+Color，或者单色Mono+Mono)，通过双目立体视觉技术获取，也可以来自激光、红外等测距技术。本发明实施例对于深度图的获取方式不做限定。
S501b：终端采集一张图像，对图像进行人脸检测；
由于现有的图像增强算法中没有利用图像的高层语义信息，只是对二维图像数据进行处理，容易对人脸区域造成过增强，出现不自然的现象。图像的低层视觉特征，如颜色、 纹理、形状,而人对图像的描述以及对图像之间相似性的判断则一般是建立在图像所描述的对象、场景以及行为特征之上图像，图像的高层语义就是图像所表达的对象、行为、场景或感情色彩。
因此，本发明实施例是基于人脸的检测结果，对人脸区域的增强程度进行限制，防止人脸区域饱和度过高，或者锐化时放大人脸区域的瑕疵。
S502a：根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度层区域；
深度图所反映的是摄像头与景物之间的远近关系，基于深度图像可以将图像分割成若干不同深度层次的图像区域。例如：图6a是本发明实施例中双摄像头采集的图像，根据图像的深度图进行分割后，图像如图6b所示，图像越亮表示图像离摄像头的距离越近。
S502b：将图像中人脸所在的区域作为感兴趣区域，并生成图像的蒙版值；
需要说明的是，本发明实施例对于人脸检测(Face Detection)的算法也不做限定。
S503：获取对焦点所在的深度区域；
S504：将所述对焦点所在的深度区域作为图像的感兴趣区域，进而根据该图像的感兴趣区域生成图像蒙版值；
需要说明的是，提取图像的感兴趣区域可以是步骤S504或者步骤S502b；也可以这二个步骤的任意组合，通过加权平均的方式得到图像的蒙版值；本发明实施例对此不作限定。
S505：根据所述蒙版值对所述图像进行处理。
处理的方式为：Mask＝0的区域，不进行处理，保持初始效果，或者进行模糊化处理，抑制非感兴趣区域，进一步突出主体；Mask＝1的区域进行亮度增强、边缘锐化、对比度增强以及色彩饱和度增强；0＜Mask＜1的区域，进行不同程度的增强处理，Mask值越大，滤波器核和参数对该区域的亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的效果越强；Mask越小，增强的程度越弱。
本发明的另一个实施例中，一种存储一个或多个程序的非易失性计算机可读存储介质，所述一个或多个程序包括指令，所述指令当被终端执行时，终端执行以下事件：
将图像进行深度图构建，生成所述图像的深度图；
根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
根据所述图像的感兴趣区域，生成所述图像的蒙版值；
根据所述图像的蒙版值对所述图像进行处理。
本发明实施例提供的一种图像增强方法，能够对图像中的不同区域进行不同的增强，既能从效果上提升照片的层次感，突出拍摄的主体，增强用户感兴趣的区域，并防止某些区域过增强；又能减少现有技术中图像增强算法对整幅图像进行处理的计算负担。
专业人员应该还可以进一步意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、计算机软件或者二者的结合来实现，为了清楚地说明硬件和软件的可互换性，在上述说明中已经按照功能一般性地描述了各示例的组成及步骤。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本发明的范围。
结合本文中所公开的实施例描述的方法或算法的步骤可以用硬件、处理器执行的软件模块，或者二者的结合来实施。软件模块可以置于随机存储器(RAM)、内存、只读存储器(ROM)、电可编程ROM、电可擦除可编程ROM、寄存器、硬盘、可移动磁盘、CD-ROM、或技术领域内所公知的任意其它形式的存储介质中。
以上所述的具体实施方式，对本发明的目的、技术方案和有益效果进行了进一步详细说明，所应理解的是，以上所述仅为本发明的具体实施方式而已，并不用于限定本发明的保护范围，凡在本发明的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。

Claims (11)

  1. 一种图像增强方法，应用于终端中，其特征在于，包括：
    将图像进行深度图构建，生成所述图像的深度图；
    根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
    将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
    根据所述图像的感兴趣区域，生成所述图像的蒙版值；
    根据所述图像的蒙版值对所述图像进行处理。
  2. 根据权利要求1所述的方法，其特征在于，所述获取图像的感兴趣区域的方法还包括：
    将图像进行深度图构建，生成所述图像的深度图；
    根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
    获取对焦点所在的所述深度区域，并将所述对焦点所在的深度区域作为感兴趣区域。
  3. 根据权利要求1所述的方法，其特征在于，所述获取图像的感兴趣区域的方法还包括：
    对图像进行人脸检测，并将所述图像中的人脸区域作为所述图像的感兴趣区域。
  4. 根据权利要求1-3任一所述的方法，其特征在于，所述根据所述图像的感兴趣区域，生成所述图像的蒙版值步骤具体为：
    根据所述图像的感兴趣区域对所述图像进行二值分割，其中所述图像的感兴趣区域的蒙版值为1，所述图像非感兴趣区域的蒙版值为0；
    将所述图像的感兴趣区域与所述图像的非感兴趣区域的相邻边界进行平滑处理，进而得到非感兴趣区域的蒙版值范围为0≤Mask＜1。
  5. 根据权利要求4所述的方法，其特征在于，所述根据所述图像的蒙版值对所述图像进行处理步骤具体为：
    Mask＝0的区域，不进行处理，或者进行模糊化处理，抑制非感兴趣区域；
    Mask＝1的区域进行亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的一种或多 种；
    0＜Mask＜1的区域，进行不同程度的增强处理，Mask值越大，滤波器核和参数对该区域的亮度增强、边缘锐化、对比度增强以及色彩饱和度增强的效果越强；Mask越小，增强的程度越弱。
  6. 一种图像增强装置，其特征在于，所述图像增强装置包括图像理解单元以及图像处理单元；其中，
    所述图像理解单元，用于提取出图像的感兴趣区域，并根据所述图像的感兴趣区域生成所述图像的蒙版值；
    所述图像处理单元，用于根据所述图像的蒙版值，对所述图像进行处理。
  7. 根据权利要求6所述的装置，其特征在于，所述图像理解单元包括深度图构建模块、显著性检测模块或对焦点获取模块以及图像蒙版生成模块；其中，
    所述深度图构建模块，用于将终端采集的图像进行深度图构建，生成所述图像的深度图；
    所述显著性检测模块，用于对所述图像进行显著性检测，并根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域，将显著性高的所述深度区域作为感兴趣区域；或者所述对焦点获取模块，获取对焦点所在的所述图像的深度区域，并将所述对焦点所在的深度区域作为感兴趣区域；
    所述图像蒙版生成模块，用于根据图像的感兴趣区域生成图像蒙版值。
  8. 根据权利要求7所述的装置，其特征在于，所述图像理解单元包括人脸检测模块以及图像蒙版生成模块；其中，
    所述人脸检测模块，用于对图像进行人脸检测，并将所述图像中的人脸区域作为所述图像的感兴趣区域；
    所述图像蒙版生成模块，用于根据所述图像的感兴趣区域，生成所述图像蒙版值。
  9. 根据权利要求7所述的装置，其特征在于，所述图像处理单元包括分布式滤波器核与参数发生器、对比度增强和亮度调节以及边缘锐化模块以及色彩饱和度增强模块；其中，
    所述分布式滤波器核与参数发生器，用于根据所述图像感兴区域蒙版值给图像的 不同区域分配不同的滤波器核和参数，并给所述图像的不同区域通过所述对比度增强和亮度调节以及边缘锐化模块以及所述色彩饱和度增强模块，进行不同的处理；
    所述对比度增强和亮度调节以及边缘锐化模块，用于增强所述图像的对比度、亮度以及边缘锐化处理；
    所述色彩饱和度增强模块，用于增强所述图像的色彩饱和度。
  10. 一种终端，其特征在于，所述终端包括：
    显示器，用于显示图像；
    存储器，用于存储指令；
    处理器，用于调用存储在所述存储器中的指令以实现：
    将图像进行深度图构建，生成所述图像的深度图；
    根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
    将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
    根据所述图像的感兴趣区域，生成所述图像的蒙版值；
    根据所述图像的蒙版值对所述图像进行处理。
  11. 一种存储一个或多个程序的非易失性计算机可读存储介质，其特征在于，所述一个或多个程序包括指令，所述指令当被终端执行时，终端执行以下事件：
    将图像进行深度图构建，生成所述图像的深度图；
    根据所述图像的深度图对所述图像进行分割，分割出所述图像的不同深度区域；
    将所述图像进行显著性检测，并将显著性高的所述深度区域作为所述图像的感兴趣区域；
    根据所述图像的感兴趣区域，生成所述图像的蒙版值；
    根据所述图像的蒙版值对所述图像进行处理。
PCT/CN2016/078492 2016-04-05 2016-04-05 一种图像增强方法及装置 WO2017173578A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2016/078492 WO2017173578A1 (zh) 2016-04-05 2016-04-05 一种图像增强方法及装置
CN201680013568.4A CN107851309A (zh) 2016-04-05 2016-04-05 一种图像增强方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/078492 WO2017173578A1 (zh) 2016-04-05 2016-04-05 一种图像增强方法及装置

Publications (1)

Publication Number Publication Date
WO2017173578A1 true WO2017173578A1 (zh) 2017-10-12

Family

ID=60000794

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/078492 WO2017173578A1 (zh) 2016-04-05 2016-04-05 一种图像增强方法及装置

Country Status (2)

Country Link
CN (1) CN107851309A (zh)
WO (1) WO2017173578A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110378852A (zh) * 2019-07-11 2019-10-25 北京奇艺世纪科技有限公司 图像增强方法、装置、计算机设备和存储介质
CN112887605A (zh) * 2021-01-26 2021-06-01 维沃移动通信有限公司 图像防抖方法、装置及电子设备
CN114390197A (zh) * 2021-12-27 2022-04-22 维沃移动通信有限公司 拍摄方法和装置、电子设备及可读存储介质
US12014508B2 (en) 2021-10-18 2024-06-18 Ford Global Technologies, Llc Distance determination from image data

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113643186A (zh) * 2020-04-27 2021-11-12 华为技术有限公司 一种图像增强方法及电子设备
CN113808054B (zh) * 2021-11-19 2022-05-06 北京鹰瞳科技发展股份有限公司 用于对眼底图像的视盘区域进行修复的方法和相关产品

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101676951A (zh) * 2008-09-19 2010-03-24 索尼株式会社 图像处理设备和方法及其程序
CN101883291A (zh) * 2010-06-29 2010-11-10 上海大学 感兴趣区域增强的视点绘制方法
US20130101175A1 (en) * 2011-10-21 2013-04-25 James D. Lynch Reimaging Based on Depthmap Information
CN104094319A (zh) * 2012-01-19 2014-10-08 株式会社东芝 图像处理设备、立体图像显示设备和图像处理方法
CN104246822A (zh) * 2012-03-22 2014-12-24 高通股份有限公司 图像增强
CN104574366A (zh) * 2014-12-18 2015-04-29 华南理工大学 一种基于单目深度图的视觉显著性区域的提取方法
CN104835146A (zh) * 2015-04-14 2015-08-12 上海大学 基于深度信息和图切割的立体图像中显著对象分割方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130129195A1 (en) * 2011-11-17 2013-05-23 Industrial Technology Research Institute Image processing method and apparatus using the same
CN104123529B (zh) * 2013-04-25 2018-04-10 株式会社理光 人手检测方法及系统
CN105069808B (zh) * 2015-08-31 2017-09-26 四川虹微技术有限公司 基于图像分割的视频图像深度估计方法
CN105404888B (zh) * 2015-11-16 2019-02-05 浙江大学 结合颜色和深度信息的显著性目标检测方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101676951A (zh) * 2008-09-19 2010-03-24 索尼株式会社 图像处理设备和方法及其程序
CN101883291A (zh) * 2010-06-29 2010-11-10 上海大学 感兴趣区域增强的视点绘制方法
US20130101175A1 (en) * 2011-10-21 2013-04-25 James D. Lynch Reimaging Based on Depthmap Information
CN104094319A (zh) * 2012-01-19 2014-10-08 株式会社东芝 图像处理设备、立体图像显示设备和图像处理方法
CN104246822A (zh) * 2012-03-22 2014-12-24 高通股份有限公司 图像增强
CN104574366A (zh) * 2014-12-18 2015-04-29 华南理工大学 一种基于单目深度图的视觉显著性区域的提取方法
CN104835146A (zh) * 2015-04-14 2015-08-12 上海大学 基于深度信息和图切割的立体图像中显著对象分割方法

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110378852A (zh) * 2019-07-11 2019-10-25 北京奇艺世纪科技有限公司 图像增强方法、装置、计算机设备和存储介质
CN112887605A (zh) * 2021-01-26 2021-06-01 维沃移动通信有限公司 图像防抖方法、装置及电子设备
CN112887605B (zh) * 2021-01-26 2022-09-30 维沃移动通信有限公司 图像防抖方法、装置及电子设备
US12014508B2 (en) 2021-10-18 2024-06-18 Ford Global Technologies, Llc Distance determination from image data
CN114390197A (zh) * 2021-12-27 2022-04-22 维沃移动通信有限公司 拍摄方法和装置、电子设备及可读存储介质

Also Published As

Publication number Publication date
CN107851309A (zh) 2018-03-27

Similar Documents

Publication Publication Date Title
WO2017173578A1 (zh) 一种图像增强方法及装置
US11457138B2 (en) Method and device for image processing, method for training object detection model
US10872420B2 (en) Electronic device and method for automatic human segmentation in image
WO2021022983A1 (zh) 图像处理方法和装置、电子设备、计算机可读存储介质
WO2016180224A1 (zh) 一种人物图像处理方法及装置
WO2019101113A1 (zh) 一种图像融合方法及其设备、存储介质、终端
CN105681627B (zh) 拍摄图像方法及电子设备
KR101662846B1 (ko) 아웃 포커싱 촬영에서 빛망울 효과를 생성하기 위한 장치 및 방법
CN108846807B (zh) 光效处理方法、装置、终端及计算机可读存储介质
EP3644599B1 (en) Video processing method and apparatus, electronic device, and storage medium
WO2022121895A1 (zh) 双目活体检测方法、装置、设备和存储介质
CN107944420B (zh) 人脸图像的光照处理方法和装置
CN111008935B (zh) 一种人脸图像增强方法、装置、系统及存储介质
WO2018094653A1 (zh) 一种用户头发模型的重建方法、装置及终端
CN110796663B (zh) 图片剪裁方法、装置、设备和存储介质
CN110503704B (zh) 三分图的构造方法、装置和电子设备
CN113658197B (zh) 图像处理方法、装置、电子设备及计算机可读存储介质
CN111161136B (zh) 图像虚化方法、图像虚化装置、设备及存储装置
WO2022261828A1 (zh) 图像处理方法、装置、电子设备及计算机可读存储介质
WO2020038065A1 (zh) 一种图像处理方法、终端及计算机存储介质
CN107564085B (zh) 图像扭曲处理方法、装置、计算设备及计算机存储介质
CN107705279B (zh) 实现双重曝光的图像数据实时处理方法及装置、计算设备
CN113221767A (zh) 训练活体人脸识别模型、识别活体人脸的方法及相关装置
KR20160046399A (ko) 텍스쳐 맵 생성 방법 및 장치와 데이터 베이스 생성 방법
CN115761827A (zh) 一种化妆进度检测方法、装置、设备及存储介质

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

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

Ref document number: 16897513

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 16897513

Country of ref document: EP

Kind code of ref document: A1