CN107292892A - The dividing method and device of video frame images - Google Patents
The dividing method and device of video frame images Download PDFInfo
- Publication number
- CN107292892A CN107292892A CN201710486311.7A CN201710486311A CN107292892A CN 107292892 A CN107292892 A CN 107292892A CN 201710486311 A CN201710486311 A CN 201710486311A CN 107292892 A CN107292892 A CN 107292892A
- Authority
- CN
- China
- Prior art keywords
- fuzziness
- vertical
- histogram
- video frame
- frame images
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
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
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of dividing method of video frame images and device, dividing method comprises the following steps：Obtain the horizontal edge information and vertical edge information of the video frame images；According to the vertical edge infomation detection image level region and the fuzziness of vertical area, coordinate is split according to the horizontal zone of testing result output image and vertical area splits coordinate；Receive the vertical area segmentation coordinate and horizontal zone splits the input of coordinate, artwork is split to obtain effective coverage.The present invention can automatically remove the inactive area of frame of video, and the poster figure of video frame content can more preferably be expressed by contributing to poster figure to generate system generation, improve poster plot quality.
Description
Technical field
The present invention relates to the dividing method and device of technical field of image processing, more particularly to a kind of video frame images.
Background technology
With the popularization of video capture equipment, people can make the video of various contents whenever and wherever possible and be shared.
But because the device category for shooting video is various, there are smart mobile phone, digital camera, DV etc., cause to shoot the quality of video, divide
Resolution, aspect ratio is different.When these videos, which are uploaded to network, to be shared, video website and video playback apparatus can bases
Equipment of itself situation, is processed to video, to adapt to demand of the equipment of itself for resolution ratio.It is most popular at present to be
Video capture is carried out using mobile phone, the video that mobile phone is shot is often vertical shooting, and most of websites can be entered to this kind of video
The processing of row addition frosted glass effect, so that this kind of video vertically shot, the horizontal widescreen for being changed into the convenient display of computer is regarded
Frequently.
Video website is daily all to receive the video data that a large number of users is uploaded, and video website needs to carry out these data
Analysis, generation represents the poster figure of these video contents.The requirement of poster figure intercepts a certain proportion of image in original video frame,
Certain size is narrowed down to, is shown in webpage and APP, in order to briefly describe the content of frame of video.Due to data volume
Huge, if including this kind of frame of video with frosted glass effect in the video that user uploads, and poster figure production algorithm is not
Have if being treated with a certain discrimination to this kind of data, the image-region with frosted glass effect is usually contained in the image of generation, together
When can also sacrifice effective resolution ratio, the poster figure so generated can largely effect on website and APP Consumer's Experience.So, urgently
Need a kind of algorithm can automatic discrimination frame of video whether have the frosted glass border of both sides, and frosted glass border is split, with
The effect of image generation algorithm is put forward, Consumer's Experience is lifted.
The content of the invention
It is a primary object of the present invention to provide a kind of dividing method of video frame images, it is intended to automatically remove frame of video
Inactive area, the poster figure of video frame content can more preferably be expressed by contributing to poster figure to generate system generation, improve poster plot quality.
To achieve the above object, a kind of dividing method for video frame images that the present invention is provided, comprises the following steps：
Obtain the horizontal edge information and vertical edge information of the video frame images；
Longitudinal fuzziness of the video frame images vertical area is calculated according to the vertical edge information, and, according to
The horizontal edge information calculates the landscape blur degree of the video frame images horizontal zone；
According to the vertical edge information and longitudinal fuzziness, the vertical area segmentation of the video frame images is determined
Coordinate, and, according to the horizontal edge information and the landscape blur degree, determine the horizontal zone point of the video frame images
Cut coordinate；
Coordinate and horizontal zone segmentation coordinate are split using the vertical area, have been partitioned into from the video frame images
Imitate region.
Preferably, the step of horizontal edge information and vertical edge information of the acquisition video frame images includes：
Obtain the color space of the video frame images；
The color space is converted into gray space or any brightness and color separated space；
Calculated level direction edge gradient operator and vertical direction edge gradient operator；
The gradient operator and the gray space or any brightness and color separated space are subjected to convolution, level is obtained
Edge graph Eh and vertical edge figure Ev；
The projection of horizontal direction is carried out to each row of the horizontal direction edge graph Eh, the histogram of horizontal direction is obtained
Hh[i]；
The projection of vertical direction is carried out to each row of the vertical direction edge graph Ev, the histogram of vertical direction is obtained
Hv[i]；The horizontal edge information of the video frame images is determined using the histogram Hh [i] of the horizontal direction；And, use
The histogram Hv [i] of the vertical direction determines the vertical edge information of the video frame images.
Preferably, longitudinal fuzziness that the video frame images vertical area is calculated according to vertical edge information, with
And, the step of landscape blur for calculating the video frame images horizontal zone according to horizontal edge information is spent includes：
The histogram Hv [i] of the vertical direction is normalized, histogram Hv [i] ` is obtained；
Histogram Hv [i] ` is begun stepping through from the first starting point, the value for obtaining first Hv [i] ` is more than the first Nogata
The position of figure threshold value is first position, and histogram Hv [i] ` is begun stepping through from the second starting point, obtains first Hv [i] `
Value be more than the first histogram thresholding position be the second place；Wherein, first starting point and second starting point are
It is located at the end points of histogram Hv [i] the ` both sides respectively；
Calculate the first longitudinal direction fuzziness and second longitudinal direction fuzziness of vertical direction two side areas；Wherein, described first indulge
To fuzziness of the fuzziness for region between first starting point and the first position, the second longitudinal direction fuzziness is institute
State the fuzziness in region between the second starting point and the second place；
Determine whether vertical direction needs segmentation according to the first longitudinal direction fuzziness and second longitudinal direction fuzziness, vertical
When direction need not be split, wrapped the step of perform detection horizontal direction fuzziness, the step of the detection level direction fuzziness
Include：
The histogram Hh [i] of the horizontal direction is normalized, histogram Hh [i] ` is obtained；
Histogram Hh [i] ` is begun stepping through from the 3rd starting point, the value for obtaining first Hh [i] ` is more than the first Nogata
The position of figure threshold value is the 3rd position, and histogram Hh [i] ` is begun stepping through from fourth initial point, obtains first Hh [i] `
Value be more than the first histogram thresholding position be the 4th position；Wherein, the 3rd starting point and fourth described initial point are
It is located at the end points of histogram Hh [i] the ` both sides respectively；
The the first landscape blur degree and the second landscape blur degree of calculated level direction two side areas, wherein, described first is horizontal
To fuzziness of the fuzziness for region between the 3rd starting point and the 3rd position, the second landscape blur degree is described the
The fuzziness in region between four starting points and the 4th position.
Preferably, according to the vertical edge information and longitudinal fuzziness, the vertical of the video frame images is determined
The step of region segmentation coordinate, includes：
Judge whether the distance between the first position and described second place are less than the first preset length, and described the
One longitudinal fuzziness is less than the first fuzziness threshold value, and the second longitudinal direction fuzziness is less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation
Position；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Preferably, it is described according to the vertical edge information and first fuzziness, determine the video frame images
The step of vertical area splits coordinate also includes：
Judge that whether the first distance is more than the second preset length, and the first longitudinal direction fuzziness is small with second distance sum
In the second fuzziness threshold value, and the second longitudinal direction fuzziness is less than the second fuzziness threshold value；Wherein, first distance is institute
The distance between first position and described first starting point are stated, the second distance is the second place and the described first starting
The distance between point, the second fuzziness threshold value is less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation
Position；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Preferably, it is described according to the vertical edge information and first fuzziness, determine the video frame images
The step of vertical area splits coordinate also includes：
If the first position is overlapped with the second place, centered on the 5th position, in the 5th position
In default span, the 6th position is judged whether；Wherein, the 5th position and the first position are on described straight
Point symmetry in side's figure Hv [i] ` transverse axis, the 6th position is that the value of histogram Hv [i] ` is more than the second histogram thresholding
Position, second histogram thresholding is less than first histogram thresholding；
If in the absence of the 6th position, the step of performing the detection level direction fuzziness；
If there is the 6th position, the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are calculated；Wherein, it is described
3rd longitudinal fuzziness is the fuzziness of a side region corresponding with the 6th position；
Calculate the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness；
Judge whether the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are more than the first fuzziness threshold value；
If it is not, the step of performing the detection level direction fuzziness；
If so, determining that the first position is the first vertical segmentation position, the 6th position is the second vertical segmentation position
Put；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Preferably, after the step of detection level direction fuzziness, in addition to：
Whether horizontal direction is determined according to the first landscape blur degree and the second landscape blur degree of horizontal direction two side areas
Segmentation is needed, when need not split in the horizontal direction, judges that the video frame images need not be split.
The present invention also provides a kind of segmenting device of video frame images, and the segmenting device of the video frame images includes：
Image edge information extraction module, horizontal edge information and vertical edge letter for obtaining the video frame images
Breath；
Position information process module, for calculating the video frame images vertical area according to the vertical edge information
Longitudinal fuzziness, and, the landscape blur degree of the video frame images horizontal zone is calculated according to the horizontal edge information；
Split coordinate obtaining module, for according to the vertical edge information and longitudinal fuzziness, it is determined that described regard
The vertical area segmentation coordinate of frequency two field picture, and, according to the horizontal edge information and the landscape blur degree, it is determined that described
The horizontal zone segmentation coordinate of video frame images；
Split module, for splitting coordinate and horizontal zone segmentation coordinate using the vertical area, from the frame of video
Effective coverage is partitioned into image.
Preferably, the video frame images edge extraction module includes：
Color space acquiring unit, the color space for obtaining the video frame images；
Converting unit, for the color space to be converted into gray space or any brightness and color separated space；
Gradient operator computing unit, for calculated level direction edge gradient operator and vertical direction edge gradient operator；
Edge graph acquiring unit, for the gradient operator to be separated into sky with the gray space or any brightness and color
Between carry out convolution, obtain horizontal edge figure Eh and vertical edge figure Ev；
Horizontal level analytic unit, the projection of horizontal direction is carried out for each row to horizontal direction edge graph, is obtained
The histogram Hh [i] of horizontal direction；
Upright position analytic unit, the projection of vertical direction is carried out for each row to vertical direction edge graph, is obtained
The histogram Hv [i] of vertical direction；
Horizontal edge information acquisition unit, for determining the frame of video using the histogram Hh [i] of the horizontal direction
The horizontal edge information of image；
Vertical edge information acquisition unit, for determining the frame of video using the histogram Hv [i] of the vertical direction
The vertical edge information of image.
Preferably, the segmentation coordinate obtaining module includes：
Line analysis unit, is used for：
The histogram Hv [i] of the vertical direction is normalized, histogram Hv [i] ` is obtained；
Histogram Hv [i] ` is begun stepping through from the first starting point, the value for obtaining first Hv [i] ` is more than the first Nogata
The position of figure threshold value is first position, and histogram Hv [i] ` is begun stepping through from the second starting point, obtains first Hv [i] `
Value be more than the first histogram thresholding position be the second place；Wherein, first starting point and second starting point are
It is located at the end points of histogram Hv [i] the ` both sides respectively；
Calculate the first longitudinal direction fuzziness and second longitudinal direction fuzziness of vertical direction two side areas；Wherein, described first indulge
To fuzziness of the fuzziness for region between first starting point and the first position, the second longitudinal direction fuzziness is institute
State the fuzziness in region between the second starting point and the second place；
Horizontal segmentation analytic unit, is used for：
The histogram Hh [i] of the horizontal direction is normalized, histogram Hh [i] ` is obtained；
Histogram Hh [i] ` is begun stepping through from the 3rd starting point, the value for obtaining first Hh [i] ` is more than the first Nogata
The position of figure threshold value is the 3rd position, and histogram Hh [i] ` is begun stepping through from fourth initial point, obtains first Hh [i] `
Value be more than the first histogram thresholding position be the 4th position；Wherein, the 3rd starting point and fourth described initial point are
It is located at the end points of histogram Hh [i] the ` both sides respectively；
The the first landscape blur degree and the second landscape blur degree of calculated level direction two side areas, wherein, described first is horizontal
To fuzziness of the fuzziness for region between the 3rd starting point and the 3rd position, the second landscape blur degree is described the
The fuzziness in region between four starting points and the 4th position；
Whether horizontal direction is determined according to the first landscape blur degree and the second landscape blur degree of horizontal direction two side areas
Segmentation is needed, when need not split in the horizontal direction, judges that the video frame images need not be split；
The vertical segmentation analytic unit, for also including：
Judge whether the distance between the first position and described second place are less than the first preset length, and described the
One longitudinal fuzziness is less than the first fuzziness threshold value, and the second longitudinal direction fuzziness is less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation
Position.
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Or,
Judge that whether the first distance is more than the second preset length, and the first longitudinal direction fuzziness is small with second distance sum
In the second fuzziness threshold value, and the second longitudinal direction fuzziness is less than the second fuzziness threshold value；Wherein, first distance is institute
The distance between first position and described first starting point are stated, the second distance is the second place and the described first starting
The distance between point, the second fuzziness threshold value is less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation
Position.
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Or,
If the first position is overlapped with the second place, centered on the 5th position, in the 5th position
In default span, the 6th position is judged whether；Wherein, the 5th position and the first position are on described straight
Point symmetry in side's figure Hv [i] ` transverse axis, the 6th position is that the value of histogram Hv [i] ` is more than the second histogram thresholding
Position, second histogram thresholding is less than first histogram thresholding；
If in the absence of the 6th position, the step of performing the detection level direction fuzziness；
If there is the 6th position, the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are calculated；Wherein, it is described
3rd longitudinal fuzziness is the fuzziness of a side region corresponding with the 6th position；
Calculate the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness；
Judge whether the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are more than the first fuzziness threshold value；
If it is not, the step of performing the detection level direction fuzziness；
If so, determining that the first position is the first vertical segmentation position, the 6th position is the second vertical segmentation position
Put；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
The dividing method of video frame images proposed by the present invention, can be to frosted glass effect video frame and ordinary video frame
Edge distribution information and fog-level are analyzed, and automatic decision image belongs to frosted glass effect video frame or ordinary video
Frame, and the frame of video of frosted glass effect is split automatically, obtain wherein original, effective image-region.For video
Website, a kind of pre-treatment means of Video processing are provided for it, the inactive area of frame of video is automatically removed, and contribute to poster figure to give birth to
The poster figure of video frame content can be more preferably expressed into system generation, poster plot quality is improved.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the embodiment of dividing method one of video frame images of the present invention；
Fig. 2 for video frame images of the present invention dividing method in acquisition video frame images horizontal edge information and vertical edges
The step schematic flow sheet of edge information；
Fig. 3 is the effect diagram of vertical edge histogram, horizontal edge histogram and edge graph in the present invention；
Fig. 4 for video frame images of the present invention the embodiment of dividing method one according to vertical edge infomation detection image level
The step schematic flow sheet of the fuzziness of region and vertical area；
Fig. 5 shows for the step flow of output vertical position coordinate in the embodiment of dividing method one of video frame images of the present invention
It is intended to；
Fig. 6 is the image segmentation schematic diagram after being split using video frame images dividing method of the present invention；
Fig. 7 is the modular structure schematic diagram of the embodiment of segmenting device one of video frame images of the present invention；
Fig. 8 shows for the structure of image edge information extraction module in the embodiment of segmenting device one of video frame images of the present invention
It is intended to；
Fig. 9 for video frame images of the present invention the embodiment of segmenting device one in segmentation coordinate obtaining module structural representation
Figure.
The realization, functional characteristics and advantage of the object of the invention will be described further referring to the drawings in conjunction with the embodiments.
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.
A kind of dividing method of video frame images of present invention offer, reference picture 1, in one embodiment, the video frame images
Dividing method include：
Step S10, obtains the horizontal edge information and vertical edge information of the video frame images；The embodiment of the present invention
In, using image processing method, the marginal information of coloured image is extracted, and it is special to be converted into a kind of directive statistics of tool
Levy.
Step S20, longitudinal fuzziness of the video frame images vertical area is calculated according to the vertical edge information, with
And, the landscape blur degree of the video frame images horizontal zone is calculated according to the horizontal edge information；In vertical edge distribution
In feature base, vertical area fuzziness is calculated,.On the basis of horizontal edge distribution characteristics, calculated level region blur degree.
Step S30, according to the vertical edge information and longitudinal fuzziness, determines the vertical of the video frame images
Region segmentation coordinate, and, according to the horizontal edge information and the landscape blur degree, determine the water of the video frame images
Flat region segmentation coordinate；According to vertical edge distribution characteristics, and the vertical area fuzziness calculated, automatic decision vertical area
In whether there is frosted glass region, and export the region of vertical segmentation.According to horizontal edge distribution characteristics, and the level calculated
It whether there is frosted glass region, and the region of output level segmentation in region blur degree, automatic decision horizontal zone.
Step S40, splits coordinate using the vertical area and horizontal zone splits coordinate, from the video frame images
It is partitioned into effective coverage.The segmentation coordinate obtained according to preceding step S30, obtains effective coverage from video frame images.
The present invention proposes classification and partitioning algorithm of the frosted glass effect video frame with ordinary video frame, according to image border
Distributed intelligence and image blur information carry out intelligent classification, and carry out vertical or water to image according to cutting position is automatic
Square to segmentation, obtain frame of video effective coverage, overall process is without manual intervention.For poster figure generation system,
Region that is meaningless, manually adding is eliminated, helps to select the image-region being more of practical significance, the interior of frame of video is expressed
Hold, lift the quality of the poster figure of generation.Pre-treatment can be provided for poster figure production system, poster figure production system will be inputted
Frame of video be automatically classified into frosted glass effect video frame and ordinary video frame, directly produce poster figure for ordinary video frame；
Split, obtained behind the effective coverage in frame of video for frosted glass effect video frame, then carry out the production of poster figure, lifted
Poster plot quality, reduces the workload of later stage artificial screening.
Reference picture 2, shows the horizontal edge information that image is obtained in the dividing method of video frame images of the present invention and hangs down
The step schematic flow sheet of straight edge information includes：
Step S11, obtains the color space of the video frame images；Obtain the color space of video frame images, Ke Yishi
Rgb color space.
Step S12, gray space or any brightness and color separated space are converted into by the color space；By input figure
As being converted into gray scale/or any brightness and color separated space (such as YUV, HSV, HSL, LAB) by rgb color space, for ash
Degree space changes formula and is：Gray=R*0.299+G*0.587+B*0.114；For brightness and color separated space, illustrated with HSL,
Brightness L (Lightness) conversion formula is：L=(max (R, G, B)+min (R, G, B))/2.
Step S13, calculated level direction edge gradient operator and vertical direction edge gradient operator；Horizontal direction and vertical
The edge gradient operator in direction is by taking Sobel operators (Sobel Operator) as an example, and other operators are equally applicable.
Step S14, convolution is carried out by the gradient operator and the gray space or any brightness and color separated space,
Obtain horizontal edge figure Eh and vertical edge figure Ev；By horizontal direction edge gradient operator and vertical direction edge gradient operator with
Gray space or any brightness and color separated space, carry out convolution algorithm, respectively obtain horizontal edge figure Eh and vertical edge
Scheme Ev.
Step S15, carries out the projection of horizontal direction to each row of horizontal direction edge graph, obtains the Nogata of horizontal direction
Scheme Hh [i]；For the influence at the edge that excludes vertical direction, only statistics horizontal edge number, i.e., for any point on image
(x, y), only counts Eh (x, y)>Th1 and Ev (x, y)<Th2 edge.
Each row of vertical direction edge graph are carried out the projection of vertical direction, obtain the Nogata of vertical direction by step S16
Scheme Hv [i]；For the purposes of the influence at the edge for excluding horizontal direction, vertical edge number is only counted, i.e., for appointing on image
A bit (x, y), Ev (x, y) is only counted>Th1 and Eh (x, y)<Th2 edge, wherein Th1 and Th2 are first threshold and the second threshold
Value.
Wherein, image edge information is calculated as follows：
E (x, y)=sqrt (Ev (x, y)2+Eh(x,y)2)；Wherein E (x, y) is any point on edge graph.
Step S17, the horizontal edge for determining the video frame images using the histogram Hh [i] of the horizontal direction is believed
Breath；And, the vertical edge information of the video frame images is determined using the histogram Hv [i] of the vertical direction.Pass through water
Square to histogram Hh [i] and vertical direction histogram Hv [i], can clearly obtain the horizontal edge of video frame images
Information and vertical edge information.
In the embodiment of the present invention, the effect of vertical edge histogram, horizontal edge histogram and edge graph refer to Fig. 3.
In the embodiment of the present invention, abovementioned steps S20 includes：According to the fuzziness of vertical edge infomation detection vertical area,
Coordinate is split according to the vertical area of testing result output image；And according to the fuzzy of horizontal edge infomation detection horizontal zone
Degree, splits coordinate according to the horizontal zone of testing result output image.In the embodiment of the present invention, the flow basic one of two steps
Cause, therefore, below with the fuzziness according to vertical edge infomation detection vertical area, according to the vertical of testing result output image
Elaborated exemplified by the step of region segmentation coordinate.Reference picture 4, in one embodiment, abovementioned steps S20 include：
Step 21, the histogram Hv [i] of vertical direction is normalized；Normalizing is carried out to vertical edge histogram
Change, histogrammic each Hv [i] divided by histogrammic maximum Hmax that vertical edge is distributed, Hv [i]=Hv [i]/
Hmax。
Step 22, the histogram after normalized is traveled through respectively from the both sides of vertical direction, from histogram Hv [i], i=
To the first split position seg1 of searching at i=length-1 at 0, to the second framing bits of searching at i=0 at i=length-1
Put seg2；Travel through the histogram Hv [i] after normalization respectively from both sides, find and be more than preset threshold value Hv [i] in histogram>
Thseg position, once searching out the two positions, record both sides position is respectively seg1, seg2, is then log out, does not continue
Found.
Step 23, the fuzziness B1, B2 of vertical direction two side areas are calculated.The fuzziness of areas at both sides is to add up per side
E (x, y) value in region, divided by the number of pixels per side region, obtain B1, B2, it is noted that the method is most simple
The method of single evaluation image region blur degree, can use more complicated method to obtain the fuzziness per side region.
In the embodiment of the present invention, the foregoing fuzziness according to horizontal edge infomation detection horizontal zone, according to testing result
The step of horizontal zone segmentation coordinate of output image, includes：
The histogram Hh [i] of horizontal direction is normalized；
Both sides from horizontal direction travel through the histogram after normalized respectively, to i=at histogram Hh [i] i=0
The first split position seg1 is found at length-1, to the second split position seg2 of searching at i=0 at i=length-1；
The fuzziness B1`, B2` of calculated level direction two side areas.
In the embodiment of the present invention, the fuzziness of vertical direction is first detected, whether process decision chart picture can be split in vertical direction,
The fuzziness in detection level direction again when vertical direction need not be split.In other embodiments, can also first detection level
The fuzziness in direction, then detect vertical direction.In the embodiment of the present invention, it can also include the steps of：According to horizontal direction both sides
The fuzziness B1` and B2` in region value determine whether horizontal direction can be split, when cannot split in the horizontal direction, judge
Image cannot be split.The embodiment of the present invention includes following several situations：1st, the fuzziness of vertical direction is first detected, vertical
When direction need not be split, the fuzziness in detection level direction, if horizontal direction need not be split, process decision chart picture need not divide
Cut.2nd, the fuzziness in first detection level direction, if horizontal direction need not be split, process decision chart picture need not be split.Level side
To or vertical direction need not split and refer to that algorithm can not export segmentation coordinate.If horizontal direction need not be split, it was demonstrated that
The frame of video of this input is ordinary video frame, directly exports baseline results.The embodiment of the present invention is imitated for polytype frosted glass
Fruit has good segmentation effect, while normal video frame can be distinguished, it is to avoid normal video frame is divided.In practical application
In, pre-treatment can be provided for poster figure production system, the frame of video for inputting poster figure production system is automatically classified into a mao glass
Glass effect video frame and ordinary video frame, poster figure is directly produced for ordinary video frame；Enter for frosted glass effect video frame
Row is split, and is obtained behind the effective coverage in frame of video, then carries out the production of poster figure, is lifted poster plot quality, is reduced later stage people
The workload of work screening.
Specifically, reference picture 5, abovementioned steps S20 includes：
Step S201, to searching seg1 at i=length-1 at the histogram Hv [i] of vertical direction, i=0, from i=
To searching seg2 at i=0 at length-1；If both sides position | seg1-seg2 |<Ths and B1<ThB and B2<ThB, i.e. both sides
When smooth in split position almost symmetry and region, it is believed that vertical direction can be split, output seg1 and seg2 is used as framing bits
Put, turn segmentation step S30, otherwise go to step S202.
Step S202, if two side positions | seg1-seg2 |<Ths and B1<ThB and B2<ThB, then export seg1 and seg2
It is used as split position；Work as seg1+seg2<0.7*length and B1<ThB2 and B2<ThB2(ThB2<When ThB), it is believed that Vertical Square
To can split, output seg1 and seg2 goes to step S30 as split position, and (physical significance now is, both sides frosted glass
Effect is asymmetric to cause seg1 and seg2 difference than larger), otherwise go to step S203.
Step S203, works as seg1+seg2<0.7*length and B1<ThB2 and B2<ThB2(ThB2<When ThB), output
Seg1 and seg2 is used as split position；When the situation more than Thseg is found only at seg1, if now H [seg1]>Thseg and
Seg1=length-seg2, are in left and right [- gap, gap] place from i=length-seg1 and find Hv [i]>Thseg2(Thseg2
<Thseg position) is used as new seg2.If finding new seg2, S204 is gone to step, otherwise vertical direction is indivisible, turned
Horizontal level analytical procedure S13.
Using step S24 method, B1, the B2 in new region are recalculated, if B1<ThB and B2<ThB, it is believed that vertical
Direction can be split, and output seg1 and seg2 turns segmentation module, otherwise vertical direction is indivisible, turns water as split position
Flat analytical procedure S13.
Step S204, has found to be more than Thseg, and H [seg2] at seg2>During Thseg and seg1=length-seg2,
Left and right [- gap, gap] place, which is in, from i=length-seg2 finds Hv [i]>Thseg2 position is used as new seg1；Calculate new
Region B1, B2, in B1<ThB and B2<During ThB, output seg1 and seg2 is as split position, the Thseg2<Thseg.
When the situation more than Thseg is found only at seg1, if now H [seg1]>Thseg and seg1=length-seg2, from i=
Length-seg1 is in left and right [- gap, gap] place and finds Hv [i]>Thseg2 position is used as new seg2；Recalculate new
Region B1, B2, in B1<ThB and B2<During ThB, output seg1 and seg2 is as split position, if finding new seg1,
S205 is gone to step, otherwise vertical direction is indivisible, turn horizontal level analytical procedure S13.
Step S205, recalculates B1, the B2 in new region, if B1<ThB and B2<ThB, then export seg1 and seg2
As split position, turn segmentation step S30, otherwise vertical direction is indivisible, turn horizontal level analytical procedure S13.
In the embodiment of the present invention, horizontal level analysis method is identical with upright position analysis method, the basis of all analyses
It is horizontal edge distribution histogram, if horizontal direction discovery is indivisible, it was demonstrated that frame of video of this input is ordinary video frame,
Directly export baseline results.
In one embodiment, abovementioned steps S30 includes：
For the split position seg1 and seg2 of vertical direction, choose rectangular area [seg1,0, width-seg1-seg2,
Height] image and output；
For the split position seg1 and seg2 of horizontal direction, rectangular area [0, seg1, width, height- are chosen
Seg1-seg2] image and output.In the embodiment of the present invention, Fig. 6 is refer to the image effect after segmentation before segmentation.
The present invention also provides a kind of segmenting device of video frame images, to realize the above method.It is shown in Figure 7,
In one embodiment, the segmenting device of the video frame images includes：
Image edge information extraction module 10, horizontal edge information and vertical edge for obtaining the video frame images
Information；In the embodiment of the present invention, image edge information extraction module 10 can utilize image processing method, extract the side of coloured image
Edge information, and it is converted into a kind of directive statistical nature of tool.
Position information process module 20, for calculating the video frame images vertical area according to the vertical edge information
Longitudinal fuzziness, and, the landscape blur degree of the video frame images horizontal zone is calculated according to the horizontal edge information；
Position information process module 20 includes vertical information processing unit 21, horizontal information processing unit 22, horizontal edge acquisition of information
Unit 23 and vertical edge information acquisition unit 24, vertical information processing unit 21 are used on vertical edge distribution characteristics basis
On, vertical area fuzziness is calculated, horizontal information processing unit 22 is used on the basis of horizontal edge distribution characteristics, calculated level
Region blur degree, horizontal edge information acquisition unit 23, for being regarded described in histogram Hh [i] determinations using the horizontal direction
The horizontal edge information of frequency two field picture, vertical edge information acquisition unit 24, for the histogram Hv using the vertical direction
[i] determines the vertical edge information of the video frame images.
Split coordinate obtaining module 30, for according to the vertical edge information and longitudinal fuzziness, it is determined that described
The vertical area segmentation coordinate of video frame images, and, according to the horizontal edge information and the landscape blur degree, determine institute
State the horizontal zone segmentation coordinate of video frame images.
Split module 40, split the input of coordinate for receiving the vertical area segmentation coordinate and horizontal zone, to original
Figure is split to obtain effective coverage.Specifically, segmentation module 40 is used to be determined according to anterior locations message processing module 30
Segmentation coordinate, effective coverage is partitioned into from video frame images.
Shown in Figure 8, in one embodiment, image edge information extraction module 10 includes：
Color space acquiring unit 11, obtains the color space of the video frame images；Obtain the color of video frame images
Space, including rgb color space.
Converting unit 12, for the color space to be converted into gray space or any brightness and color separated space；
Specifically, the color space (such as rgb color space) can be converted into gray scale/or any brightness and color point by converting unit 12
From space (such as YUV, HSV, HSL, LAB), changing formula for gray space is：Gray=R*0.299+G*0.587+B*0.114；
For brightness and color separated space, illustrated with HSL, brightness L (Lightness) conversion formula is：L=(max (R, G, B)+
min(R,G,B))/2。
Gradient operator computing unit 13, is calculated for calculated level direction edge gradient operator and vertical direction edge gradient
Son；Edge gradient operator horizontally and vertically is by taking Sobel operators as an example, and other operators are equally applicable.
Edge graph acquiring unit 14, for the gradient operator to be separated with the gray space or any brightness and color
Space carries out convolution, obtains horizontal edge figure Eh and vertical edge figure Ev；By horizontal direction edge gradient operator and vertical direction
Edge gradient operator and gray space or any brightness and color separated space, carry out convolution algorithm, respectively obtain horizontal edge
Scheme Eh and vertical edge figure Ev.
Horizontal level analytic unit 15, the projection of horizontal direction is carried out for each row to horizontal direction edge graph, is obtained
Obtain the histogram Hh [i] of horizontal direction；For the influence at the edge that excludes vertical direction, only statistics horizontal edge number, i.e., pair
In any point (x, y) on image, Eh (x, y) is only counted>Th1 and Ev (x, y)<Th2 edge.
Upright position analytic unit 16, the projection of vertical direction is carried out for each row to vertical direction edge graph, is obtained
Obtain the histogram Hv [i] of vertical direction；For the purposes of the influence at the edge for excluding horizontal direction, upright position analytic unit 16
Vertical edge number is counted, i.e., for any point (x, y) on image, only counts Ev (x, y)>Th1 and Eh (x, y)<Th2 side
Edge, wherein Th1 and Th2 are first threshold and Second Threshold.
Wherein, image edge information is calculated as follows：
E (x, y)=sqrt (Ev (x, y)2+Eh(x,y)2)；Wherein E (x, y) is any point on edge graph.
Horizontal edge information acquisition unit 17, for determining the video using the histogram Hh [i] of the horizontal direction
The horizontal edge information of two field picture；By the histogram Hh [i] of horizontal direction, the level of video frame images can be clearly obtained
Marginal information.
Vertical edge information acquisition unit 18, for determining the video using the histogram Hv [i] of the vertical direction
The vertical edge information of two field picture；By the histogram Hv [i] of vertical direction, the vertical of video frame images can be clearly obtained
Marginal information.
Reference picture 9, in one embodiment, segmentation coordinate obtaining module 30 include：
Vertical segmentation analytic unit 31, is used for：
The histogram Hv [i] of vertical direction is normalized；
The histogram after normalized is traveled through respectively from the both sides of vertical direction, to i at histogram Hv [i], i=0
The first split position seg1 is found at=length-1, to the second split position seg2 of searching at i=0 at i=length-1；
Calculate the fuzziness B1, B2 of vertical direction two side areas；
Horizontal segmentation analytic unit 32, is used for：
The histogram Hh [i] of horizontal direction is normalized；
Both sides from horizontal direction travel through the histogram after normalized respectively, to i=at histogram Hh [i] i=0
The first split position seg1 is found at length-1, to the second split position seg2 of searching at i=0 at i=length-1；
The fuzziness B1`, B2` of calculated level direction two side areas.
Vertical segmentation analytic unit 31 is additionally operable to realize abovementioned steps S201 to S205：
To searching seg1 at i=length-1 at the histogram Hv [i] of vertical direction, i=0, at i=length-1
Seg2 is found to i=0；
If two side positions | seg1-seg2 |<Ths and B1<ThB and B2<ThB, then export seg1 and seg2 and be used as segmentation
Position；
Work as seg1+seg2<0.7*length and B1<ThB2 and B2<ThB2(ThB2<When ThB), output seg1 and seg2 makees
For split position；
When the situation more than Thseg is found only at seg1, if now H [seg1]>Thseg and seg1=length-
Seg2, is in left and right [- gap, gap] place from i=length-seg1 and finds Hv [i]>Thseg2 position is used as new seg2；
B1, the B2 in new region are recalculated, in B1<ThB and B2<During ThB, output seg1 and seg2 is described as split position
Thseg2<Thseg；
Find to be more than Thseg, and H [seg2] at seg2>During Thseg and seg1=length-seg2, from i=
Length-seg2 are in left and right [- gap, gap] place and find Hv [i]>Thseg2 position is used as new seg1；Calculate new area
B1, the B2 in domain, in B1<ThB and B2<During ThB, output seg1 and seg2 is used as split position.
In one embodiment, segmentation module 40 is used for：
For the split position seg1 and seg2 of vertical direction, choose rectangular area [seg1,0, width-seg1-seg2,
Height] image and output；
For the split position seg1 and seg2 of horizontal direction, rectangular area [0, seg1, width, height- are chosen
Seg1-seg2] image and output.
In present invention, horizontal segmentation analytic unit 32 is additionally operable to：According to the horizontal direction two side areas
Fuzziness B1` and B2` value determine whether horizontal direction can be split, when cannot split in the horizontal direction, process decision chart picture
It cannot split.
The preferred embodiments of the present invention are these are only, are not intended to limit the scope of the invention, it is every to utilize this hair
Equivalent structure or equivalent flow conversion that bright specification and accompanying drawing content are made, or directly or indirectly it is used in other related skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of dividing method of video frame images, it is characterised in that the dividing method of the video frame images includes following step
Suddenly：
Obtain the horizontal edge information and vertical edge information of the video frame images；
Longitudinal fuzziness of the video frame images vertical area is calculated according to the vertical edge information, and, according to described
Horizontal edge information calculates the landscape blur degree of the video frame images horizontal zone；
According to the vertical edge information and longitudinal fuzziness, determine that the vertical area segmentation of the video frame images is sat
Mark, and, according to the horizontal edge information and the landscape blur degree, determine the horizontal zone segmentation of the video frame images
Coordinate；
Coordinate and horizontal zone segmentation coordinate are split using the vertical area, effective district is partitioned into from the video frame images
Domain.
2. the dividing method of video frame images as claimed in claim 1, it is characterised in that the acquisition video frame images
Horizontal edge information and vertical edge information the step of include：
Obtain the color space of the video frame images；
The color space is converted into gray space or any brightness and color separated space；
Calculated level direction edge gradient operator and vertical direction edge gradient operator；
The gradient operator and the gray space or any brightness and color separated space are subjected to convolution, horizontal edge is obtained
Scheme Eh and vertical edge figure Ev；
The projection of horizontal direction is carried out to each row of the horizontal direction edge graph Eh, the histogram Hh of horizontal direction is obtained
[i]；
The projection of vertical direction is carried out to each row of the vertical direction edge graph Ev, the histogram Hv of vertical direction is obtained
[i]；The horizontal edge information of the video frame images is determined using the histogram Hh [i] of the horizontal direction；And, using institute
The histogram Hv [i] for stating vertical direction determines the vertical edge information of the video frame images.
3. the dividing method of video frame images as claimed in claim 2, it is characterised in that described according to vertical edge information meter
Longitudinal fuzziness of the video frame images vertical area is calculated, and, the video frame images are calculated according to horizontal edge information
The step of landscape blur of horizontal zone is spent includes：
The histogram Hv [i] of the vertical direction is normalized, histogram Hv [i] ` is obtained；
Histogram Hv [i] ` is begun stepping through from the first starting point, the value for obtaining first Hv [i] ` is more than the first histogram threshold
The position of value is first position, and histogram Hv [i] ` is begun stepping through from the second starting point, obtains first Hv [i] ` value
Position more than the first histogram thresholding is the second place；Wherein, first starting point and second starting point are difference
End points positioned at histogram Hv [i] the ` both sides；
Calculate the first longitudinal direction fuzziness and second longitudinal direction fuzziness of vertical direction two side areas；Wherein, the first longitudinal direction mould
Paste degree is the fuzziness in region between first starting point and the first position, and the second longitudinal direction fuzziness is described the
The fuzziness in region between two starting points and the second place；
Determine whether vertical direction needs segmentation according to the first longitudinal direction fuzziness and second longitudinal direction fuzziness, in vertical direction
When need not split, include the step of perform detection horizontal direction fuzziness, the step of the detection level direction fuzziness：
The histogram Hh [i] of the horizontal direction is normalized, histogram Hh [i] ` is obtained；
Histogram Hh [i] ` is begun stepping through from the 3rd starting point, the value for obtaining first Hh [i] ` is more than the first histogram threshold
The position of value is the 3rd position, and histogram Hh [i] ` is begun stepping through from fourth initial point, obtains first Hh [i] ` value
Position more than the first histogram thresholding is the 4th position；Wherein, the 3rd starting point and fourth described initial point are difference
End points positioned at histogram Hh [i] the ` both sides；
The the first landscape blur degree and the second landscape blur degree of calculated level direction two side areas, wherein, the described first horizontal mould
Paste degree for region between the 3rd starting point and the 3rd position fuzziness, the second landscape blur degree for it is described fourth
The fuzziness in region between initial point and the 4th position.
4. the dividing method of video frame images as claimed in claim 3, it is characterised in that according to the vertical edge information and
The step of longitudinal fuzziness, vertical area segmentation coordinate for determining the video frame images, includes：
Judge whether the distance between the first position and described second place are less than the first preset length, and described first indulges
It is less than the first fuzziness threshold value to fuzziness, and the second longitudinal direction fuzziness is less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation position；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
5. the dividing method of video frame images as claimed in claim 4, it is characterised in that described to be believed according to the vertical edge
The step of breath and first fuzziness, vertical area segmentation coordinate for determining the video frame images, also includes：
Judge whether the first distance and second distance sum are more than the second preset length, and the first longitudinal direction fuzziness is less than the
Two fuzziness threshold values, and the second longitudinal direction fuzziness is less than the second fuzziness threshold value；Wherein, first distance is described
The distance between one position and described first starting point, the second distance be the second place and first starting point it
Between distance, the second fuzziness threshold value be less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation position；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
6. the dividing method of the video frame images as described in claim 4 or 5, it is characterised in that described according to the vertical edges
The step of edge information and first fuzziness, vertical area segmentation coordinate for determining the video frame images, also includes：
If the first position is overlapped with the second place, centered on the 5th position, in the default of the 5th position
In span, the 6th position is judged whether；Wherein, the 5th position and the first position are on the histogram
Point symmetry in Hv [i] ` transverse axis, the 6th position is that the value of histogram Hv [i] ` is more than the position of the second histogram thresholding
Put, second histogram thresholding is less than first histogram thresholding；
If in the absence of the 6th position, the step of performing the detection level direction fuzziness；
If there is the 6th position, the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are calculated；Wherein, the described 3rd
Longitudinal fuzziness is the fuzziness of a side region corresponding with the 6th position；
Calculate the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness；
Judge whether the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are more than the first fuzziness threshold value；
If it is not, the step of performing the detection level direction fuzziness；
If so, determining that the first position is the first vertical segmentation position, the 6th position is the second vertical segmentation position；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
7. the dividing method of video frame images as claimed in claim 3, it is characterised in that detection level direction fuzziness
The step of after, in addition to：
According to the first landscape blur degree and the second landscape blur degree of horizontal direction two side areas determine horizontal direction whether needs
Segmentation, when need not split in the horizontal direction, judges that the video frame images need not be split.
8. a kind of segmenting device of video frame images, it is characterised in that the segmenting device of the video frame images includes：
Image edge information extraction module, horizontal edge information and vertical edge information for obtaining the video frame images；
Position information process module, the longitudinal direction for calculating the video frame images vertical area according to the vertical edge information
Fuzziness, and, the landscape blur degree of the video frame images horizontal zone is calculated according to the horizontal edge information；
Split coordinate obtaining module, for according to the vertical edge information and longitudinal fuzziness, determining the frame of video
The vertical area segmentation coordinate of image, and, according to the horizontal edge information and the landscape blur degree, determine the video
The horizontal zone segmentation coordinate of two field picture；
Split module, for splitting coordinate and horizontal zone segmentation coordinate using the vertical area, from the video frame images
In be partitioned into effective coverage.
9. the segmenting device of video frame images as claimed in claim 8, it is characterised in that the video frame images marginal information
Extraction module includes：
Color space acquiring unit, the color space for obtaining the video frame images；
Converting unit, for the color space to be converted into gray space or any brightness and color separated space；
Gradient operator computing unit, for calculated level direction edge gradient operator and vertical direction edge gradient operator；
Edge graph acquiring unit, for the gradient operator to be entered with the gray space or any brightness and color separated space
Row convolution, obtains horizontal edge figure Eh and vertical edge figure Ev；
Horizontal level analytic unit, the projection of horizontal direction is carried out for each row to horizontal direction edge graph, level is obtained
The histogram Hh [i] in direction；
Upright position analytic unit, the projection of vertical direction is carried out for each row to vertical direction edge graph, obtains vertical
The histogram Hv [i] in direction；
Horizontal edge information acquisition unit, for determining the video frame images using the histogram Hh [i] of the horizontal direction
Horizontal edge information；
Vertical edge information acquisition unit, for determining the video frame images using the histogram Hv [i] of the vertical direction
Vertical edge information.
10. the segmenting device of video frame images as claimed in claim 8, it is characterised in that the segmentation coordinate obtaining module
Including：
Vertical segmentation analytic unit, is used for：
The histogram Hv [i] of the vertical direction is normalized, histogram Hv [i] ` is obtained；
Histogram Hv [i] ` is begun stepping through from the first starting point, the value for obtaining first Hv [i] ` is more than the first histogram threshold
The position of value is first position, and histogram Hv [i] ` is begun stepping through from the second starting point, obtains first Hv [i] ` value
Position more than the first histogram thresholding is the second place；Wherein, first starting point and second starting point are difference
End points positioned at histogram Hv [i] the ` both sides；
Calculate the first longitudinal direction fuzziness and second longitudinal direction fuzziness of vertical direction two side areas；Wherein, the first longitudinal direction mould
Paste degree is the fuzziness in region between first starting point and the first position, and the second longitudinal direction fuzziness is described the
The fuzziness in region between two starting points and the second place；
Horizontal segmentation analytic unit is used for：
The histogram Hh [i] of the horizontal direction is normalized, histogram Hh [i] ` is obtained；
Histogram Hh [i] ` is begun stepping through from the 3rd starting point, the value for obtaining first Hh [i] ` is more than the first histogram threshold
The position of value is the 3rd position, and histogram Hh [i] ` is begun stepping through from fourth initial point, obtains first Hh [i] ` value
Position more than the first histogram thresholding is the 4th position；Wherein, the 3rd starting point and fourth described initial point are difference
End points positioned at histogram Hh [i] the ` both sides；
The the first landscape blur degree and the second landscape blur degree of calculated level direction two side areas, wherein, the described first horizontal mould
Paste degree for region between the 3rd starting point and the 3rd position fuzziness, the second landscape blur degree for it is described fourth
The fuzziness in region between initial point and the 4th position；
According to the first landscape blur degree and the second landscape blur degree of horizontal direction two side areas determine horizontal direction whether needs
Segmentation, when need not split in the horizontal direction, judges that the video frame images need not be split；
The vertical segmentation analytic unit, for also including：
Judge whether the distance between the first position and described second place are less than the first preset length, and described first indulges
It is less than the first fuzziness threshold value to fuzziness, and the second longitudinal direction fuzziness is less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation position.
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Or,
Judge whether the first distance and second distance sum are more than the second preset length, and the first longitudinal direction fuzziness is less than the
Two fuzziness threshold values, and the second longitudinal direction fuzziness is less than the second fuzziness threshold value；Wherein, first distance is described
The distance between one position and described first starting point, the second distance be the second place and first starting point it
Between distance, the second fuzziness threshold value be less than the first fuzziness threshold value；
If, it is determined that the first position is the first vertical segmentation position, and the second place is the second vertical segmentation position.
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Or,
If the first position is overlapped with the second place, centered on the 5th position, in the default of the 5th position
In span, the 6th position is judged whether；Wherein, the 5th position and the first position are on the histogram
Point symmetry in Hv [i] transverse axis, the 6th position is that the value of the histogram Hv [i] is more than the position of the second histogram thresholding,
Second histogram thresholding is less than first histogram thresholding；
If in the absence of the 6th position, the step of performing the detection level direction fuzziness；
If there is the 6th position, the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are calculated；Wherein, the described 3rd
Longitudinal fuzziness is the fuzziness of a side region corresponding with the 6th position；
Calculate the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness；
Judge whether the first longitudinal direction fuzziness and the 3rd longitudinal fuzziness are more than the first fuzziness threshold value；
If it is not, the step of performing the detection level direction fuzziness；
If so, determining that the first position is the first vertical segmentation position, the 6th position is the second vertical segmentation position；
Using the first vertical segmentation position and the second vertical segmentation position, the vertical area segmentation coordinate is calculated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710486311.7A CN107292892B (en) | 2017-06-23 | 2017-06-23 | Video frame image segmentation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710486311.7A CN107292892B (en) | 2017-06-23 | 2017-06-23 | Video frame image segmentation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107292892A true CN107292892A (en) | 2017-10-24 |
| CN107292892B CN107292892B (en) | 2020-04-21 |
Family
ID=60102109
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710486311.7A Active CN107292892B (en) | 2017-06-23 | 2017-06-23 | Video frame image segmentation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107292892B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108769803A (en) * | 2018-06-29 | 2018-11-06 | 北京字节跳动网络技术有限公司 | Recognition methods, method of cutting out, system, equipment with frame video and medium |
| CN108939584A (en) * | 2018-08-14 | 2018-12-07 | 浙江大丰实业股份有限公司 | The huge screen content analysis platform of background |
| CN112363844A (en) * | 2021-01-12 | 2021-02-12 | 之江实验室 | Convolutional neural network vertical segmentation method for image processing |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007074695A (en) * | 2005-08-09 | 2007-03-22 | Mitsubishi Electric Corp | Image processor |
| CN102254172A (en) * | 2011-06-16 | 2011-11-23 | 电子科技大学 | Method for segmenting fingerprint image based on cellular neural network and morphology |
| CN103426144A (en) * | 2012-05-17 | 2013-12-04 | 佳能株式会社 | Method and device for deblurring image having perspective distortion |
| CN104166841A (en) * | 2014-07-24 | 2014-11-26 | 浙江大学 | Rapid detection identification method for specified pedestrian or vehicle in video monitoring network |
| CN104200475A (en) * | 2014-09-05 | 2014-12-10 | 中国传媒大学 | Novel no-reference image blur degree estimation method |
| CN106485727A (en) * | 2016-09-08 | 2017-03-08 | 乐视控股（北京）有限公司 | The dividing method of object and device in a kind of video |
| CN106504261A (en) * | 2016-10-31 | 2017-03-15 | 北京奇艺世纪科技有限公司 | A kind of image partition method and device |
-
2017
- 2017-06-23 CN CN201710486311.7A patent/CN107292892B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007074695A (en) * | 2005-08-09 | 2007-03-22 | Mitsubishi Electric Corp | Image processor |
| CN102254172A (en) * | 2011-06-16 | 2011-11-23 | 电子科技大学 | Method for segmenting fingerprint image based on cellular neural network and morphology |
| CN103426144A (en) * | 2012-05-17 | 2013-12-04 | 佳能株式会社 | Method and device for deblurring image having perspective distortion |
| CN104166841A (en) * | 2014-07-24 | 2014-11-26 | 浙江大学 | Rapid detection identification method for specified pedestrian or vehicle in video monitoring network |
| CN104200475A (en) * | 2014-09-05 | 2014-12-10 | 中国传媒大学 | Novel no-reference image blur degree estimation method |
| CN106485727A (en) * | 2016-09-08 | 2017-03-08 | 乐视控股（北京）有限公司 | The dividing method of object and device in a kind of video |
| CN106504261A (en) * | 2016-10-31 | 2017-03-15 | 北京奇艺世纪科技有限公司 | A kind of image partition method and device |
Non-Patent Citations (1)
| Title |
|---|
| 沙莎 等: "边缘信息引导的阈值图像分割算法", 《中国图象图形学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108769803A (en) * | 2018-06-29 | 2018-11-06 | 北京字节跳动网络技术有限公司 | Recognition methods, method of cutting out, system, equipment with frame video and medium |
| CN108769803B (en) * | 2018-06-29 | 2021-06-22 | 北京字节跳动网络技术有限公司 | Recognition method, cutting method, system, equipment and medium for video with frame |
| CN108939584A (en) * | 2018-08-14 | 2018-12-07 | 浙江大丰实业股份有限公司 | The huge screen content analysis platform of background |
| CN112363844A (en) * | 2021-01-12 | 2021-02-12 | 之江实验室 | Convolutional neural network vertical segmentation method for image processing |
| CN112363844B (en) * | 2021-01-12 | 2021-04-09 | 之江实验室 | Convolutional neural network vertical segmentation method for image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107292892B (en) | 2020-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10088600B2 (en) | Weather recognition method and device based on image information detection | |
| CN107705288B (en) | Infrared video detection method for dangerous gas leakage under strong interference of pseudo-target motion | |
| WO2017088365A1 (en) | Skin-colour detection method and apparatus | |
| CN106846339A (en) | Image detection method and device | |
| CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
| CN108805116B (en) | Image text detection method and system | |
| CN106934386B (en) | A kind of natural scene character detecting method and system based on from heuristic strategies | |
| CN104361357B (en) | Photo album categorizing system and sorting technique based on image content analysis | |
| US9881202B2 (en) | Providing visual effects for images | |
| KR20150079576A (en) | Depth map generation from a monoscopic image based on combined depth cues | |
| CN102956029B (en) | Image processing apparatus and image processing method | |
| CN102402918B (en) | Method for improving picture quality and liquid crystal display (LCD) | |
| CN107977645B (en) | Method and device for generating video news poster graph | |
| CN104463134B (en) | A kind of detection method of license plate and system | |
| CN109948566A (en) | A kind of anti-fraud detection method of double-current face based on weight fusion and feature selecting | |
| CN107292892A (en) | The dividing method and device of video frame images | |
| CN103377272A (en) | Automatic selection method and automatic selection system for representative thumbnail of photo data folder | |
| CN105678301B (en) | method, system and device for automatically identifying and segmenting text image | |
| CN106910195A (en) | A kind of web page layout monitoring method and device | |
| CN109165592B (en) | Real-time rotatable face detection method based on PICO algorithm | |
| CN111080722A (en) | Color migration method and system based on significance detection | |
| CN109784357A (en) | A kind of image based on statistical model retakes detection method | |
| WO2024016632A1 (en) | Bright spot location method, bright spot location apparatus, electronic device and storage medium | |
| CN105721738B (en) | A kind of chromoscan file and picture preprocess method | |
| CN203970354U (en) | A kind of Tongue analytical system based on mobile terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |