CN106469306A - Many people image extract real-time based on infrared structure light and synthetic method - Google Patents
Many people image extract real-time based on infrared structure light and synthetic method Download PDFInfo
- Publication number
- CN106469306A CN106469306A CN201610856895.8A CN201610856895A CN106469306A CN 106469306 A CN106469306 A CN 106469306A CN 201610856895 A CN201610856895 A CN 201610856895A CN 106469306 A CN106469306 A CN 106469306A
- Authority
- CN
- China
- Prior art keywords
- image
- target
- structure light
- infrared
- infrared structure
- 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
- 239000000284 extract Substances 0.000 title claims abstract description 18
- 238000010189 synthetic method Methods 0.000 title claims abstract description 13
- 238000000034 method Methods 0.000 claims description 27
- 230000011218 segmentation Effects 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 3
- 238000003708 edge detection Methods 0.000 claims description 2
- 230000003287 optical effect Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 5
- 230000004927 fusion Effects 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000006002 Pepper Substances 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
-
- 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
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of many people image extract real-time based on infrared structure light and synthetic method, depth information be increased to target image by infrared structure light, detect the border of target using depth information, in real time the participant of all video conferences can be extracted from respective video image, then it is synthesized in a sub-picture, thus increasing the acquisition of information of participant, improve Consumer's Experience.The device that the present invention adopts is simple, and cost is relatively low, and algorithm complex is little, can have good practical value with real-time processing video image.
Description
Technical field
The invention belongs to image procossing, computer vision and optical engineering technical field, refer in particular to one kind and be based on infrared structure
Many people image extract real-time of light and synthetic method.
Background technology
Video conference is to be built between 2 points or multiple place by transmission channel using computer technology and communication equipment
Stand visual multimedia communication, realize a kind of conferencing form of image, voice and data exchange.When holding video conference,
The representatives participating in the conference being in different location can receive sound and the meeting-place scene in other side meeting-place, makes representatives participating in the conference same just as being in
One meeting-place participation meeting is the same, so can significantly improve work efficiency.
Currently, with regard to video conference research with application tentatively achieve cross-region people carry out aspectant investigation with
Collaborative work, but still suffer from following weak point：
(1) acquisition of information is imperfect, and in video conference, participant cannot obtain all participants' in synchronization
Audiovisual information；
(2) spatial impression of meeting is not strong with sense of reality, and the participant in different meeting-place does not have space to the perception of whole meeting
Sense talks environment it is impossible to construction is real.
Content of the invention
In view of the shortcomings of the prior art, the present invention proposes a kind of many people image extract real-time based on infrared structure light
With synthetic method.The participant of all video conferences is extracted from respective video image by the present invention, is then synthesized to
In one sub-picture, thus increasing the acquisition of information of participant, improve Consumer's Experience.The device that the present invention adopts is simple, and cost is relatively
Low, algorithm complex is little, can have good practical value with real-time processing video image.
In order to realize above-mentioned technical purpose, the technical solution used in the present invention is：
A kind of many people image extract real-time based on infrared structure light and synthetic method, comprise the following steps：
S1, the infrared structure light image of collection target (one of participant) and RGB image
Image capture device using Infrared laser emission device and with infrared fileter gathers the infrared structure of target
Light image and RGB image, method is as follows：
When infrared generating laser is opened, launch a plurality of vertical bar straight line to the surface of target, at this moment image acquisition sets
The standby infrared structure light image collecting target；When infrared generating laser cuts out, what image capture device collected is mesh
Target RGB image；
When image acquisition is carried out to target, Infrared laser emission device is opened and closed according to certain frequency,
Make image capture device obtain infrared structure light image and the RGB image of target successively, note is adjacent once open and close red
The infrared structure light image that outer generating laser obtains is F (x, y), and RGB image is P (x, y, z), and the size of two width images is M
×N；
S2, pretreatment is carried out to infrared structure light image F (x, y), obtain gray level image F2(x,y)；
S3, to image F2(x, y) carries out binaryzation, and note binary image is F3(x,y)；
S4, to F3(x, y) carries out rim detection, is then merged four edge detection results images, obtains edge
Frame, is designated as FD (x, y)；
S5, image procossing is carried out to FD (x, y), remove some isolated points, the new images obtaining are designated as FD4(x,y)；
S6, bianry image FD4The non-zero points of (x, y) are exactly the external boundary of target, and some borders have fracture, and this is to select
Two points and two points below before breaking part, altogether 4 points carry out averagely, the result obtaining enters row interpolation at this, this
Sample obtains the external boundary image FD of a closure5(x,y)；
S7, according to external boundary image FD5(x, y) carries out Target Segmentation to target image for P (x, y, z), extracts target
Partial image.
S8, using the method in S1 to S7, respectively image zooming-out is carried out to the participant of different location, then extracting
Image put on the big figure of setting in advance, thus obtained one inclusion all participants composograph.
In the S2 of the present invention, pretreatment is carried out to infrared structure light image F (x, y), method is as follows：
First, all pixels of traversal F (x, y), adopt the template of 3 × 3, to institute in template window to each pixel
Some pixels are ranked up according to the pixel value size of pixel, find out maximum therein and minima, are designated as noise, enter
Row is deleted, and then remaining 7 points is carried out averagely, average being assigned to the center pixel that current pixel is template window, note is processed
Infrared structure light image afterwards is F1(x,y)；
Then, for filtered image F1(x, y) adopts Laplace operator to F1(x, y) is filtered, and obtains new
Image is designated as F2(x, y), formula is as follows：
WhereinRepresent and do convolution algorithm.
In the S3 of the present invention, using classical maximum variance between clusters to image F2(x, y) carries out binaryzation, only retains figure
As F2Bright fringess information on (x, y), note binary image is F3(x,y).(maximum variance between clusters are by the big Tianjin of Japanese scholars
(Nobuyuki Otsu) proposed in 1979, is a kind of method that adaptive threshold value determines, is Da-Jin algorithm, referred to as again
OTSU.This patent operates fully according to this method, and the online source code of this method is downloaded, and directly invokes)
In the present invention, the implementation method of S4 is as follows：
S41, according to the feature of infrared vertical bar structure light image, designs four kinds of new template, as follows：
Binary image is F by S423(x, y) respectively with four kinds of new template carry out convolution, then carry out two-value by threshold value
Change, obtain four binary image { FDi(x, y) | i=1,2,3,4 }, concrete formula is as follows
WhereinRepresent convolution, { Di| i=1,2,3,4 } it is above-mentioned four kinds of new template, TH is the threshold value of binaryzation, typically takes
Value 5.
S43 is by four binary image { FDi(x, y) | i=1,2,3,4 } merged, each pixel is according to OR
Operated, thus obtained the more complete marginal information image of ratio, be designated as FD (x, y).
FD (x, y)=FD1(x,y)|FD2(x,y)|FD3(x,y)|FD4(x,y)
Wherein | it is step-by-step or operator, this illustrates only when the whole four width binary images in a certain position are 0,
The fusion results of this position are 0 eventually, and other situation fusion results are 1.
In S5 of the present invention, in order to prevent the interference of some of them noise, need to delete some isolated points.Concretely comprise the following steps：
S51, for bianry image FD (x, y), each of traversing graph picture non-zero pixels point, calculates centered on it, size
The sum of all pixels value in the field of 5x5, if less than 2 then it is assumed that this non-zero pixels point is noise, its value is changed to 0；At note
Image after reason is FD2(x,y).
S52 in order to obtain the external border information of target, to FD2(x, y) is carried out by column scan, and every string only retains line number
Two big and minimum non-zero pixels values, other non-zero pixels values all reset to zero, and the new images obtaining are designated as FD3(x,y).
S53 calculates image FD3Two nonzero values (namely line number minimum and maximum two nonzero values) of (x, y) every string
Distance, set up array array to keep result, then array is a 1 dimension group, length be image columns N.Pick
Except the catastrophe point in array array, method is as follows：Each element of traversal array, if the difference of current point and former and later two points
Different it is all higher than 10 then it is assumed that this puts as catastrophe point.In image FD3Two on corresponding for all of catastrophe point row on (x, y)
Nonzero value resets to 0, and the new images obtaining are designated as FD4(x,y).
In S7 of the present invention, according to the external boundary image FD obtaining in S65(x, y) carries out mesh to target image for P (x, y, z)
Mark segmentation, method is as follows：Processed by row, according to external boundary image FD5(x, y) its each column only has two nonzero values, for
External boundary image FD5Either rank y on (x, y)m, external boundary image FD5In y in (x, y)mThe corresponding row of two nonzero values of row
Number is respectively xn1And xn2, then to the y for P (x, y, z) for the target imagemRow, only retain line number from xn1To xn2Between picture
Element, using same method external boundary image FD5After (x, y) upper all of row are processed, the image of target part also by
Extract.
The present invention increased depth information by infrared structure light to target image, detects the side of target using depth information
The participant of all video conferences can be extracted from respective video image by boundary in real time, is then synthesized to a secondary figure
In picture, thus increasing the acquisition of information of participant, improve Consumer's Experience.The device that the present invention adopts is simple, and cost is relatively low, algorithm
Complexity is little, can have good practical value with real-time processing video image.
Brief description
Fig. 1 is that the image capture device using Infrared laser emission device and with infrared fileter gathers the infrared of target
Structure light image and the schematic diagram of RGB image；
Fig. 2 is the flow chart of the present invention；
Fig. 3 is the schematic diagram of image synthesis.
Specific embodiment
The invention will be further described with reference to the accompanying drawings and detailed description.
Structure light refers to have certain projection structure, includes the Rhizoma Dioscoreae (peeled) of certain information.These fringe projections are to measured object
After on body, if the shape of projection surface is uneven, then the Rhizoma Dioscoreae (peeled) that video camera photographs is just no longer straight line, but there occurs
Deflection, deformation.Actually these deformation contain the dimensional surface information of object, therefore carry out three-dimensional measurement based on structure light,
It is widely used in the fields such as the reconstruction of threedimensional model, the measurement of profiling object surface.
As shown in figure 1, the image acquisition first with Infrared laser emission device and with infrared fileter for the present invention sets
The infrared structure light image of standby collection target and RGB image.When infrared generating laser is opened, launch a plurality of vertical bar straight
Line is to the surface of target, the at this moment infrared structure light image of the target that image capture device obtains；When infrared generating laser closes
When closing, no longer emitting structural light is to the surface of target, the at this moment RGB image of the target that image capture device obtains.To target
When carrying out image acquisition, Infrared laser emission device is opened according to certain frequency (typically taking 50ms to 100ms) and is closed
Close so that image capture device obtains infrared structure light image and the RGB image of target successively.Note is adjacent once to be opened and closed
The infrared structure light image that Infrared laser emission device obtains is F (x, y), and RGB image is P (x, y, z), and the size of two width images is
M×N.
Then the present invention is by carrying out processing the boundary information obtaining target to F (x, y), then mesh from P (x, y, z)
Mark extracts, and detailed process is as follows：
(1) infrared structure light image F (x, y) has a lot of noises, including impulse type noise, Gaussian noise, salt-pepper noise
Deng if directly to splitting containing noisy image, a lot of image details can be lost, or pseudo-edge occurs.Therefore first
To carrying out Image semantic classification, denoising is carried out to image by filtering.
First, infrared structure light image F (x, y) is filtered processing, comprises the following steps that：
The all pixels of traversal F (x, y), adopt the template of 3 × 3, to all of in template window to each pixel
Pixel is ranked up according to the value size of pixel, finds out maximum therein and minima, is designated as noise, is deleted
Remove, then remaining 7 points are carried out averagely, average being assigned to current pixel (namely center pixel of window).After note is processed
Infrared structure light image is F1(x,y).
Above-mentioned process is effectively reduced the random noise of image, but there is also the negative effect that part edge obscures.For
Solve this problem, can adopt again for filtered image and sharpen spatial filter, project the details in image
Or strengthen, the present invention adopts Laplace operator to F1(x, y) is filtered, and obtains new image and is designated as F2(x, y), formula
As follows：
WhereinRepresent and do convolution algorithm.
(2) after filtering after gray level image F2(x, y), with dark fringe and bright fringess, but in analysis stripe pattern
When, only interested in bright fringess, and the transitional region between dark fringe inside, light and shade striped and background are lost interest in.Therefore
Target striped namely bright fringess is needed to extract.Here the present invention adopts classical maximum variance between clusters (otsu) to figure
As F2(x, y) carries out binaryzation, only retains bright fringess information, and note binary image is F3(x,y).(maximum variance between clusters be by
Big Tianjin of Japanese scholars (Nobuyuki Otsu) proposed in 1979, was a kind of method that adaptive threshold value determines, cried big again
Tianjin method, abbreviation OTSU.This patent operates fully according to this method, and the online source code of this method is downloaded, and directly adjusts
With)
(3) to F3(x, y) carries out rim detection.
The present invention, according to the feature of infrared vertical bar structure light image, devises four kinds of new templates, as follows：
Binary image is F3(x, y) respectively with four kinds of new template carry out convolution, then carry out binaryzation by threshold value,
Obtain four binary image { FDi(x, y) | i=1,2,3,4 }, concrete formula is as follows
WhereinRepresent convolution, { Di| i=1,2,3,4 } it is above-mentioned four kinds of new template, TH is the threshold value of binaryzation, typically
Value 5.
Then by four binary image { FDi(x, y) | i=1,2,3,4 } merged, each pixel is according to " or fortune
Calculate " operated, thus obtain the more complete marginal information image of ratio, be designated as FD (x, y).
FD (x, y)=FD1(x,y)|FD2(x,y)|FD3(x,y)|FD4(x,y)
Wherein | it is step-by-step or operator, this illustrates only when the whole four width binary images in a certain position are 0,
The fusion results of this position are 0 eventually, and other situation fusion results are 1.
In order to prevent the interference of some of them noise, need to delete some isolated points.Concretely comprise the following steps：
For bianry image FD (x, y), travel through each non-zero pixels point, calculate centered on it, the field of size 5x5
The sum of interior all pixels value, if less than 2 then it is assumed that this non-zero pixels point is noise, its value is changed to 0.Figure after note process
Picture is FD2(x,y).
In order to obtain the external border information of target, to FD2(x, y) is carried out by column scan, and it is maximum that every string only retains line number
With the pixel value of two minimum non-zero pixels points, other nonzero values all reset to zero, and the new images obtaining are designated as FD3(x,y).
The external border information disappearance being arranged due to some, leads to the pixel finally retaining in fact to be in the inside of target, institute
To also need to screen further.Calculate every string two nonzero values (namely line number minimum and maximum two nonzero values) away from
From, set up array array to keep result, then array is a 1 dimension group, length be image columns N.Reject number
Catastrophe point in group array, comprises the following steps that：Each element of traversal array, if the difference of current point and former and later two points
Different it is all higher than 10 then it is assumed that this puts as catastrophe point.In image FD3Two on corresponding for all of catastrophe point row on (x, y)
Nonzero value resets to 0, and the new images obtaining are designated as FD4(x,y).
So bianry image FD4The non-zero points of (x, y) are exactly the external boundary of target, and some borders have fracture, at this moment select
Select before breaking part two points and two points below, 4 points are carried out averagely altogether, and the result obtaining enters row interpolation at this,
The external boundary image FD of a closure so can be obtained5(x,y).
(6) according to external boundary image FD5(x, y) carries out Target Segmentation to target image for P (x, y, z), extracts target
Partial image.
The external boundary image FD being obtained according to step 55(x, y), carries out Target Segmentation to target image for P (x, y, z).Press
Row are processed, according to external boundary image FD5(x, y) its each column only has two nonzero values, for external boundary image FD5On (x, y)
Either rank ym, external boundary image FD5In y in (x, y)mThe corresponding line number of two nonzero values of row is respectively xn1And xn2, then
To the y for P (x, y, z) for the target imagemRow, only retain line number from xn1To xn2Between pixel, external using same method
Boundary image FD5After (x, y) upper all of row are processed, the image of target part is also extracted.
(7) according to above-mentioned steps 1 to step 6, all image zooming-out can be carried out to the participant of different location, then extraction
The image going out is put on the big figure of setting in advance, as shown in figure 3, indicating the position of each target in advance on the composite image
Confidence ceases, and so after extracting a target, directly target image is pasted into corresponding position in composograph, thus
Composograph to all participants.
Claims (7)
1. a kind of many people image extract real-time based on infrared structure light with synthetic method it is characterised in that comprising the following steps：
S1, the infrared structure light image of collection target and RGB image
Image capture device using Infrared laser emission device and with infrared fileter gathers the infrared structure light figure of target
Picture and RGB image, method is as follows：
When infrared generating laser is opened, launch a plurality of vertical bar straight line to the surface of target, at this moment image capture device is adopted
Collect the infrared structure light image of target；When infrared generating laser cuts out, what image capture device collected is target
RGB image；
When image acquisition is carried out to target, Infrared laser emission device is opened and closed according to certain frequency so that
Image capture device obtains infrared structure light image and the RGB image of target successively, and note is adjacent once to be opened and closed and infrared swash
The infrared structure light image that optical transmitting set obtains is F (x, y), and RGB image is P (x, y, z), and the size of two width images is M × N；
S2, pretreatment is carried out to infrared structure light image F (x, y), obtain gray level image F2(x,y)；
S3, to image F2(x, y) carries out binaryzation, and note binary image is F3(x,y)；
S4, to F3(x, y) carries out rim detection, is then merged four edge detection results images, obtains marginal information figure
Picture, is designated as FD (x, y)；
S5, image procossing is carried out to FD (x, y), remove some isolated points, the new images obtaining are designated as FD4(x,y)；
S6, bianry image FD4The non-zero points of (x, y) are exactly the external boundary of target, and some borders have fracture, and this is to select fracture
Place's above two points and two points below, altogether 4 points carry out averagely, the result obtaining enters row interpolation at this, such must
External boundary image FD to a closure5(x,y)；
S7, according to external boundary image FD5(x, y) carries out Target Segmentation to target image for P (x, y, z), extracts target part
Image；
S8, the method adopting in S1 to S7, carry out image zooming-out to the participant of different location, respectively then the figure extracting
As putting on the big figure of setting in advance, thus obtain the composograph of all participants of inclusion.
2. the many people image extract real-time based on infrared structure light according to claim 1 and synthetic method, its feature exists
In the method for S2 is as follows：
First, all pixels of traversal F (x, y), adopt the template of 3 × 3, to all of in template window to each pixel
Pixel is ranked up according to the pixel value size of pixel, finds out maximum therein and minima, is designated as noise, is deleted
Remove, then remaining 7 points are carried out averagely, average being assigned to the center pixel that current pixel is template window, after note is processed
Infrared structure light image is F1(x,y)；
Then, for filtered image F1(x, y) adopts Laplace operator to F1(x, y) is filtered, and obtains new image
It is designated as F2(x, y), formula is as follows：
WhereinRepresent and do convolution algorithm.
3. the many people image extract real-time based on infrared structure light according to claim 1 and 2 and synthetic method, its feature
It is, in S3, using classical maximum variance between clusters to image F2(x, y) carries out binaryzation, only retains image F2On (x, y)
Bright fringess information, note binary image be F3(x,y).
4. the many people image extract real-time based on infrared structure light according to claim 3 and synthetic method, its feature exists
In the implementation method of S4 is as follows：
S41, according to the feature of infrared vertical bar structure light image, designs four kinds of new template, as follows：
Binary image is F by S423(x, y) respectively with four kinds of new template carry out convolution, then carry out binaryzation by threshold value, obtain
To four binary image { FDi(x, y) | i=1,2,3,4 }, concrete formula is as follows
WhereinRepresent convolution, { Di| i=1,2,3,4 } it is above-mentioned four kinds of new template, TH is the threshold value of binaryzation；
S43 is by four binary image { FDi(x, y) | i=1,2,3,4 } merged, each pixel is carried out according to OR
Operation, thus obtains the more complete marginal information image of ratio, is designated as FD (x, y)：
FD (x, y)=FD1(x,y)|FD2(x,y)|FD3(x,y)|FD4(x,y)
Wherein | it is step-by-step or operator.
5. the many people image extract real-time based on infrared structure light according to claim 4 and synthetic method, its feature exists
In, in S42, TH value is 5.
6. the many people image extract real-time based on infrared structure light according to claim 4 or 5 and synthetic method, its feature
It is, the method for S5 is：
S51, for bianry image FD (x, y), each of traversing graph picture non-zero pixels point, calculates centered on it, size 5x5
Field in all pixels value sum, if less than 2 then it is assumed that this non-zero pixels point be noise, its value is changed to 0；Note is processed
Image afterwards is FD2(x,y)；
S52 in order to obtain the external border information of target, to FD2(x, y) is carried out by column scan, every string only retain line number maximum and
Two minimum non-zero pixels values, other non-zero pixels values all reset to zero, and the new images obtaining are designated as FD3(x,y)；
S53 calculates image FD3The distance of two nonzero values of (x, y) every string, sets up array array to keep result, then
Array is a 1 dimension group, and length is the columns N of image；Reject the catastrophe point in array array, method is as follows：Traversal number
Each element of group, if the difference of current point and former and later two points is all higher than 10 then it is assumed that this puts as catastrophe point；In image
FD3On (x, y), two nonzero values on corresponding for all of catastrophe point row are reset to 0, the new images obtaining are designated as FD4(x,
y).
7. the many people image extract real-time based on infrared structure light according to claim 6 and synthetic method, its feature exists
In the method for S7 is：According to the external boundary image FD obtaining in S65(x, y) carries out target to target image for P (x, y, z) and divides
Cut, method is as follows：Processed by row, according to external boundary image FD5(x, y) its each column only has two nonzero values, for outside
Boundary image FD5Either rank y on (x, y)m, external boundary image FD5In y in (x, y)mThe corresponding line number of two nonzero values of row is divided
Wei not xn1And xn2, then to the y for P (x, y, z) for the target imagemRow, only retain line number from xn1To xn2Between pixel, adopt
With same method external boundary image FD5After (x, y) upper all of row are processed, the image of target part is also extracted
Come.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610856895.8A CN106469306B (en) | 2016-09-28 | 2016-09-28 | More people's image extract real-times and synthetic method based on infrared structure light |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610856895.8A CN106469306B (en) | 2016-09-28 | 2016-09-28 | More people's image extract real-times and synthetic method based on infrared structure light |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106469306A true CN106469306A (en) | 2017-03-01 |
| CN106469306B CN106469306B (en) | 2019-07-09 |
Family
ID=58230710
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610856895.8A Active CN106469306B (en) | 2016-09-28 | 2016-09-28 | More people's image extract real-times and synthetic method based on infrared structure light |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106469306B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109215046A (en) * | 2018-08-06 | 2019-01-15 | 浙江工贸职业技术学院 | A kind of Laplace operator edge detection method based on image interpolation arithmetic |
| CN111524088A (en) * | 2020-05-06 | 2020-08-11 | 北京未动科技有限公司 | Method, device and equipment for image acquisition and computer-readable storage medium |
| CN112204605A (en) * | 2019-08-29 | 2021-01-08 | 深圳市大疆创新科技有限公司 | Extreme point extraction method, extreme point extraction device, and computer-readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1877599A (en) * | 2006-06-29 | 2006-12-13 | 南京大学 | Face setting method based on structured light |
| CN103795961A (en) * | 2012-10-30 | 2014-05-14 | 三亚中兴软件有限责任公司 | Video conference telepresence system and image processing method thereof |
-
2016
- 2016-09-28 CN CN201610856895.8A patent/CN106469306B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1877599A (en) * | 2006-06-29 | 2006-12-13 | 南京大学 | Face setting method based on structured light |
| CN103795961A (en) * | 2012-10-30 | 2014-05-14 | 三亚中兴软件有限责任公司 | Video conference telepresence system and image processing method thereof |
Non-Patent Citations (4)
| Title |
|---|
| 俞柏峰 等: "多画面视频图像分割拼接实现方法", 《长江大学学报（自然科学版）》 * |
| 周阿珍: "浅析多点视频会议中的多画面合成方法", 《电子世界》 * |
| 安平 等: "视频会议***中基于图像拼合的中间视合成", 《上海大学学报（自然科学版）》 * |
| 汪陈伍 等: "一种在像素域内实现多画面合成的方法", 《西安邮电学院学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109215046A (en) * | 2018-08-06 | 2019-01-15 | 浙江工贸职业技术学院 | A kind of Laplace operator edge detection method based on image interpolation arithmetic |
| CN112204605A (en) * | 2019-08-29 | 2021-01-08 | 深圳市大疆创新科技有限公司 | Extreme point extraction method, extreme point extraction device, and computer-readable storage medium |
| WO2021035621A1 (en) * | 2019-08-29 | 2021-03-04 | 深圳市大疆创新科技有限公司 | Extreme point extraction method and apparatus, and computer-readable storage medium |
| CN111524088A (en) * | 2020-05-06 | 2020-08-11 | 北京未动科技有限公司 | Method, device and equipment for image acquisition and computer-readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106469306B (en) | 2019-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113643289B (en) | Fabric surface defect detection method and system based on image processing | |
| CN104050471B (en) | Natural scene character detection method and system | |
| CN101443791B (en) | Improved foreground/background separation in digitl images | |
| CN104361314B (en) | Based on infrared and transformer localization method and device of visual image fusion | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| WO2018023916A1 (en) | Shadow removing method for color image and application | |
| CN107133969B (en) | A kind of mobile platform moving target detecting method based on background back projection | |
| CN105160310A (en) | 3D (three-dimensional) convolutional neural network based human body behavior recognition method | |
| CN105894484A (en) | HDR reconstructing algorithm based on histogram normalization and superpixel segmentation | |
| CN110008832A (en) | Based on deep learning character image automatic division method, information data processing terminal | |
| CN103955949B (en) | Moving target detecting method based on Mean-shift algorithm | |
| CN104077577A (en) | Trademark detection method based on convolutional neural network | |
| CN102184534B (en) | Method for image fusion by using multi-scale top-hat selective transform | |
| CN101527043B (en) | Video picture segmentation method based on moving target outline information | |
| CN106296744A (en) | A kind of combining adaptive model and the moving target detecting method of many shading attributes | |
| CN106709438A (en) | Method for collecting statistics of number of people based on video conference | |
| CN106469306A (en) | Many people image extract real-time based on infrared structure light and synthetic method | |
| CN105913407A (en) | Method for performing fusion optimization on multi-focusing-degree image base on difference image | |
| CN103514608A (en) | Movement target detection and extraction method based on movement attention fusion model | |
| CN106529441A (en) | Fuzzy boundary fragmentation-based depth motion map human body action recognition method | |
| CN111080574A (en) | Fabric defect detection method based on information entropy and visual attention mechanism | |
| CN111222432A (en) | Face living body detection method, system, equipment and readable storage medium | |
| CN114005081A (en) | Intelligent detection device and method for foreign matters in tobacco shreds | |
| CN105046670A (en) | Image rain removal method and system | |
| Zhu et al. | Towards automatic wild animal detection in low quality camera-trap images using two-channeled perceiving residual pyramid networks |
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 |