The content of the invention
The main object of the present invention is provides a kind of image acquiring method, and the landscape that can choose needs is shot, and is lifted
Image effect.
The present invention proposes a kind of image acquiring method, including:
The landscape that camera lens is taken in is identified, and extracts the mark for having recognized landscape;
Selection according to photographer to mark, retains and identifies the landscape associated with selected;
After shooting is performed, generation retains the image of landscape.
Preferably, it is described that the landscape that camera lens is taken in is identified, and it is specific to extract the step of having recognized the mark of landscape
Including:
Under cleaning pattern of taking pictures, personage's overall profile that camera lens is taken in is identified, and extract character face's conduct
The mark of personage's overall profile.
Preferably, the selection according to photographer to mark, the step of retaining the landscape associated with selected mark, is specific
Including:
When there is multiple face identifications, being supplied to photographer to be selected the multiple face identification, will not be chosen
The personage's overall profile for selecting face identification association is rejected from image and allows to perform shooting again.
Preferably, also include after described image processing method:
The part rejected in image is replaced using background image.
The present invention also proposes a kind of image acquiring device, including:
Landscape identification module, for the landscape that camera lens is taken in be identified, and extracts the mark for having recognized landscape;
Selecting module is shot, for the selection according to photographer to mark, retains and identifies the landscape associated with selected;
Image generation module, for after shooting is performed, generation to retain the image of landscape.
Preferably, the landscape identification module specifically for:
Under cleaning pattern of taking pictures, personage's overall profile that camera lens is taken in is identified, and extract character face's conduct
The mark of personage's overall profile.
Preferably, it is described shooting selecting module specifically for:
When there is multiple face identifications, being supplied to photographer to be selected the multiple face identification, will not be chosen
The personage's overall profile for selecting face identification association is rejected from image and allows to perform shooting again.
Preferably, described image processing unit also includes:
Image complementary module, for the part rejected in image to be replaced using background image.
The present invention also proposes a kind of mobile terminal, including:
Landscape identification module, for the landscape that camera lens is taken in be identified, and extracts the mark for having recognized landscape;
Selecting module is shot, for the selection according to photographer to mark, retains and identifies the landscape associated with selected;
Image generation module, for after shooting is performed, generation to retain the image of landscape.
Preferably, the landscape identification module specifically for:
Under cleaning pattern of taking pictures, personage's overall profile that camera lens is taken in is identified, and extract character face's conduct
The mark of personage's overall profile.
Preferably, it is described shooting selecting module specifically for:
When there is multiple face identifications, being supplied to photographer to be selected the multiple face identification, will not be chosen
The personage's overall profile for selecting face identification association is rejected from image and allows to perform shooting again.
Preferably, the mobile terminal also includes:
Image complementary module, for the part rejected in image to be replaced using background image.
The present invention can by the identification to personage in taking lens, and provide identification character face's feature give photographer choosing
Select, retain the selected personage of photographer in generation image, so that the image that photographer shoots required for can easily obtaining.
Embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Reference picture 1, proposes a kind of embodiment of image acquiring method one of the invention.The image processing method may include:
Step S11, the landscape to camera lens intake are identified, and extract the mark for having recognized landscape;
Step S12, the selection according to photographer to mark, retain and identify the landscape associated with selected;
Step S13, perform shoot after, generation retain landscape image.
Above-mentioned steps S11 may particularly include:Under cleaning pattern of taking pictures, personage's overall profile that camera lens is taken in is known
Not, and mark of the character face as personage's overall profile is extracted.Before image taking, the selection of screening-mode can be carried out,
Than if desired for realizing that cleaning pattern of taking pictures may be selected in scheme in the present embodiment.
Above-mentioned character contour feature is used as a kind of one of inherent important clue that shape facility is human body behavior.Institute of the present invention
The contour feature used, which describes method, has star-like skeleton technique, Fourier to describe operator etc..Fourier describes operator coefficient
Low frequency part describes the general characteristics of profile, and the microscopic feature of profile is reflected with HFS.The present invention is in above two wheel
On the basis of wide description method, also by the profile described function of boundary point-radius, the local feature of profile is described with
Analysis, can to blocking, loss of data and robustness is stronger situations such as deformation.
The character contour of the present invention is extracted to be comprised the following steps with pretreatment:Image denoising, target detection and background segment,
Shadow removal and contours extract.Image denoising refer to remove obtain and transmit image when by various noise jammings, and this
A little noise jammings can bring very big inconvenience to subsequent treatment, and visual perception is also influenceed to a certain extent.In order to improve image
Quality and it is easy to various subsequent treatments, image denoising of the invention uses the method that gaussian filtering and neighborhood denoising are combined.
The target detection of the present invention comprises the following steps with background segment:Motion people in video sequence is detected using background subtraction
Body, initially sets up the background model of an adaptive change with time and spatial variations, then by present frame and background model
Difference is carried out, given threshold, binarization segmentation image obtains foreground and background.The shadow removal of the present invention comprises the following steps:
Remove when detecting moving target, the merging of the moving target caused, geometry deformation, and the data that target is lost.The present invention
Contours extract comprise the following steps:Using based on connective edge following algorithm, personage in whole image sequence is extracted
Profile.
Above-mentioned steps S12 may particularly include:When there is multiple face identifications, the multiple face identification is supplied to bat
The person of taking the photograph is selected, and personage's overall profile that not selected face identification is associated is rejected from image to be allowed to perform shooting again.
When camera lens take in landscape in identify there is multiple personages when, can extract the plurality of personage facial characteristics displaying
To photographer, it is to need to be retained in the image ultimately generated to select which personage for photographer, so as to retain
The landscape taken in from camera lens of personage in reject.So that after photographer presses shutter execution shooting, can in the image of the shooting
Retain the selected character image of photographer and shooting background image etc., and non-selected character image is then from landscape
Reject, will not be embodied in the image graph picture of the shooting.
Reference picture 2, in another embodiment of the invention, may also include after above-mentioned steps S13:
Step S14, the part rejected in image replaced using background image.
After above-mentioned non-selected character image is rejected from image, the image of the image will appear from the feelings of excalation
Condition, will so influence image effect.The background image that can have entered camera lens by interception in the present embodiment is filled to the lack part,
Or the lack part is filled up according to the background image as the background image of camera lens rebuilds the lack part so that image will not
Whole structure is influenceed because there is excalation.
Above-mentioned image acquiring method, can be by the identification to personage in taking lens, and it is special to provide the character face of identification
Levy and give photographer's selection, retain the selected personage of photographer in generation image, so that needed for photographer can easily obtain
The image to be shot.
Reference picture 3, proposes a kind of embodiment of image acquiring device 20 1 of the invention.The image processing apparatus 20 may include:
Landscape identification module 21, shooting selecting module 22 and image generation module 23;The landscape identification module 21, for being taken the photograph to camera lens
The landscape entered is identified, and extracts the mark for having recognized landscape;The shooting selecting module 22, for according to photographer to mark
Selection, retain and the selected landscape that associates of mark;The image generation module 23, for after shooting is performed, generation to retain scape
The image of sight.
Above-mentioned landscape identification module 21 can be specifically for:Under cleaning pattern of taking pictures, the personage's monolithic wheel taken in camera lens
Exterior feature is identified, and extracts mark of the character face as personage's overall profile.Before image taking, screening-mode can be carried out
Selection, than if desired for realizing that cleaning pattern of taking pictures may be selected in scheme in the present embodiment.
Above-mentioned character contour feature is used as a kind of one of inherent important clue that shape facility is human body behavior.Conventional wheel
Wide character description method has star-like skeleton technique, Fourier descriptors etc..Fourier describes operator and retouched with the low frequency part of coefficient
The general characteristics of profile are stated, the microscopic feature of profile is reflected with HFS.The basis of method is described in above two profile
On, a kind of profile described function for being referred to as boundary point-radius of definable is gone back, the local feature of profile can be described with dividing
Analysis, to blocking, loss of data and robustness is stronger situations such as deformation.
Character contour extracts as follows with pre-treatment step:Image denoising, target detection and background segment, shadow removal and
Contours extract.The image denoising:Video sequence often can be by various noise jammings in acquisition and transmission, and these noises can be given
Subsequent treatment brings very big inconvenience, and visual perception is also influenceed to a certain extent.In order to improve picture quality and be easy to each
Plant subsequent treatment, it is proposed that the method for many removal noises, algorithm uses the method that gaussian filtering and neighborhood denoising are combined.Should
Target detection and background segment:Movement human in video sequence is detected using background subtraction, one is initially set up with the time
With the background model of spatial variations and adaptive change, present frame and background model are then subjected to difference, given threshold, two-value
Change segmentation figure picture and obtain foreground and background.The shadow removal:When detecting moving target, the shade of moving target projection can also be detected
For a part for sport foreground, merging, the geometry deformation of moving target are caused, or even loses target.The contours extract:Using
Based on connective edge following algorithm, the profile for extracting pedestrian in whole image sequence is obtained.
Above-mentioned shooting selecting module 22 can be specifically for:When there is multiple face identifications, by the multiple face identification
It is supplied to photographer to be selected, personage's overall profile that not selected face identification is associated is rejected from image to be allowed to hold again
Row is shot.
When camera lens take in landscape in identify there is multiple personages when, can extract the plurality of personage facial characteristics displaying
To photographer, it is to need to be retained in the image ultimately generated to select which personage for photographer, so as to retain
The landscape taken in from camera lens of personage in reject.So that after photographer presses shutter execution shooting, can in the image of the shooting
Retain the selected character image of photographer and shooting background image etc., and non-selected character image is then from landscape
Reject, will not be embodied in the image graph picture of the shooting.
Reference picture 4, in an alternative embodiment of the invention, above-mentioned image processing apparatus 20 may also include:Image complementary module
24, for the part rejected in image to be replaced using background image.
After above-mentioned non-selected character image is rejected from image, the image of the image will appear from the feelings of excalation
Condition, will so influence image effect.The background image that can have entered camera lens by interception in the present embodiment is filled to the lack part,
Or the lack part is filled up according to the background image as the background image of camera lens rebuilds the lack part so that image will not
Whole structure is influenceed because there is excalation.
Above-mentioned image acquiring device 20, can be by the identification to personage in taking lens, and provides the character face of identification
Feature is selected to photographer, retains the selected personage of photographer in generation image, so that photographer can easily obtain institute
Need the image shot.
Reference picture 5, proposes an a kind of embodiment of mobile terminal 30 of the invention.The mobile terminal 30 may include:Landscape is known
Other module 21, shooting selecting module 22 and image generation module 23;The landscape identification module 21, for the scape taken in camera lens
Sight is identified, and extracts the mark for having recognized landscape;The shooting selecting module 22, for the choosing according to photographer to mark
Select, retain and identify the landscape associated with selected;The image generation module 23, for after shooting is performed, generation to retain landscape
Image.
Above-mentioned landscape identification module 21 can be specifically for:Under cleaning pattern of taking pictures, the personage's monolithic wheel taken in camera lens
Exterior feature is identified, and extracts mark of the character face as personage's overall profile.
Above-mentioned shooting selecting module 22 can be specifically for:When there is multiple face identifications, by the multiple face identification
It is supplied to photographer to be selected, personage's overall profile that not selected face identification is associated is rejected from image to be allowed to hold again
Row is shot.
Above-mentioned mobile terminal 30 may also include:Image complementary module 24, for the part rejected in image to be used into background
Image is replaced.
Above-mentioned landscape identification module 21, shooting selecting module 22, image generation module 23 and image complementary module 24 have
Body can be as described in above-mentioned embodiment.
Image acquiring method, image acquiring device in the present invention, and mobile terminal can apply still image and dynamic
The acquisition of state image.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the scope of the invention, it is every to utilize
Equivalent structure or equivalent flow conversion that description of the invention and accompanying drawing content are made, or directly or indirectly it is used in other correlations
Technical field, be included within the scope of the present invention.