CN106610987A - Video image retrieval method, device and system - Google Patents
Video image retrieval method, device and system Download PDFInfo
- Publication number
- CN106610987A CN106610987A CN201510694753.1A CN201510694753A CN106610987A CN 106610987 A CN106610987 A CN 106610987A CN 201510694753 A CN201510694753 A CN 201510694753A CN 106610987 A CN106610987 A CN 106610987A
- Authority
- CN
- China
- Prior art keywords
- video
- color
- frame
- image
- statistical data
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/783—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/7847—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using low-level visual features of the video content
- G06F16/785—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using low-level visual features of the video content using colour or luminescence
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Library & Information Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a video image retrieval method, device and system. The method comprises the steps of: acquiring color statistical data of a target picture, wherein the color statistical data of the target picture is data for characterizing the color of a first feature in the target picture; analyzing an encapsulation layer of a pre-stored video frame to obtain color statistical data of the video frame, wherein the video frame is obtained by encapsulating the color statistical data and video coding data of the video frame; matching the color statistical data of the target picture with the color statistical data of the video frame to obtain a target video frame having the matching result satisfying a preset condition; and performing image retrieval on the target video frame according to the first feature. The method, the device and the system solve the technical problem of low image retrieval efficiency caused by temporary decoding and analysis on an image in the image retrieval process of the prior art.
Description
Technical field
The present invention relates to image process and analysis field, in particular to a kind of video image search method, device
And system.
Background technology
As the fast development of image processing techniquess, the video data volume are more and more, increased from massive video data
Carry out the difficulty of image retrieval.Prior art needs to carry out image interim decoding and analysis in image retrieval procedure,
But picture decoding needs to expend substantial amounts of process time with analysis, this will seriously reduce the efficiency of image retrieval.Additionally,
Image retrieval of the prior art too relies on the efficiency and accuracy of destination object detection, but in video monitoring scene,
Due to the presence of the factors such as different distances, angle and ambient interferences, it will the serious standard for reducing destination object detection
Exactness and reliability, meanwhile, the missing inspection of destination object and flase drop problem also will to the accuracy of destination object detection and
Reliability causes to have a strong impact on.
Need to carry out image the image retrieval that interim decoding is caused with analysis in image retrieval procedure for prior art
The low problem of efficiency, not yet proposes at present effective solution.
The content of the invention
A kind of video image search method, apparatus and system are embodiments provided, at least to solve prior art
Need to carry out image the interim decoding technology low with the image retrieval efficiency that causes is analyzed in image retrieval procedure to ask
Topic.
A kind of one side according to embodiments of the present invention, there is provided method of video image processing, including:Obtain target
The color statistical data of picture, wherein, the color statistical data of Target Photo is to characterize fisrt feature in Target Photo
The data of color;The encapsulated layer of the frame of video to prestoring is parsed, and obtains the color statistical data of frame of video,
Wherein, frame of video is obtained after the color statistical data and video data encoder by frame of video is packaged;By mesh
The color statistical data of piece of marking on a map is matched with the color statistical data of frame of video, is obtained matching result and is met default bar
The target video frame of part;And image retrieval is carried out to target video frame according to fisrt feature.
Further, the color statistical data of Target Photo includes the color information of fisrt feature and the color of fisrt feature
The corresponding pixel quantity of information, wherein, obtaining the color statistical data of Target Photo includes:Obtain the color of fisrt feature
Multimedia message;Obtain the pixel quantity of frame of video;Calculate the shared percentage ratio in Target Photo of the size of fisrt feature;
And the corresponding pixel quantity of color information of calculating fisrt feature, wherein, the corresponding picture of color information of fisrt feature
Prime number amount is the pixel quantity of frame of video and the product of percentage ratio.
Further, the color statistical data of frame of video include frame of video in each characteristics of image color information and each
The corresponding pixel quantity of color information of characteristics of image, wherein, by the color statistical data of Target Photo and frame of video
Color statistical data carries out matching to be included:By the color information of fisrt feature respectively with the color information of each characteristics of image
Matched;And believe the corresponding pixel quantity of the color information of fisrt feature with the color of each characteristics of image respectively
Cease corresponding pixel quantity to be matched.
Further, carrying out image retrieval to target video frame according to fisrt feature includes:By the video of target video frame
Coded data is decoded, and obtains decoded video image;Whether judge in decoded video image comprising first
Feature;And if it is judged that in decoded video image include fisrt feature, then it is decoded video image is true
It is set to the retrieval result of Target Photo.
A kind of other side according to embodiments of the present invention, there is provided video image processing system, including:Video camera,
For obtaining the color statistical data and video data encoder of video image, and by color statistical data and Video coding number
According to being packaged into frame of video;And storage server, for obtaining the color statistical data of Target Photo;To frame of video
Encapsulated layer is parsed, and obtains the color statistical data of frame of video;By the color statistical data and frame of video of Target Photo
Color statistical data matched, obtain matching result meet pre-conditioned target video frame;And according to first
Feature carries out image retrieval to target video frame, wherein, during the color statistical data of Target Photo is to characterize Target Photo
The data of the color of fisrt feature, frame of video is sealed in the color statistical data and video data encoder by frame of video
Obtain after dress.
Further, video camera is used in such a way color statistical data and video data encoder are packaged into into video
Frame:Set up the corresponding relation of color statistical data and video data encoder;And according to corresponding relation by color statistical number
According to being packaged with video data encoder, frame of video is obtained.
Further, video camera is used to obtain the color statistical data of video image in such a way:Obtain video figure
The color information of characteristics of image as in;And each pixel in video image is carried out into color classification, statistical picture
The corresponding pixel quantity of color information of feature.
Further, video camera is used in such a way each pixel in video image is carried out into color classification,
The corresponding pixel quantity of color information of statistical picture feature:Obtain the color-values of the first pixel in video image;Judge
Whether the color-values of the first pixel color-values corresponding with the color information of characteristics of image match;And if it is judged that
The color-values matching corresponding with the color information of characteristics of image of the color-values of one pixel, the color of the first pixel is classified as
The color information of characteristics of image, and the corresponding pixel quantity of color information of accumulated image feature.
Another aspect according to embodiments of the present invention, additionally provides a kind of video image processing device, including:First obtains
Delivery block, for obtaining the color statistical data of Target Photo, wherein, the color statistical data of Target Photo is sign
The data of the color of fisrt feature in Target Photo;Parsing module, for entering to the encapsulated layer of the frame of video for prestoring
Row parsing, obtains the color statistical data of frame of video, wherein, frame of video be in the color statistical data by frame of video and
What video data encoder was obtained after being packaged;Second acquisition module, for by the color statistical data of Target Photo with
The color statistical data of frame of video is matched, and is obtained matching result and is met pre-conditioned target video frame;And inspection
Rope module, for carrying out image retrieval to target video frame according to fisrt feature.
Further, the color statistical data of Target Photo includes the color information of fisrt feature and the color of fisrt feature
The corresponding pixel quantity of information, wherein, the first acquisition module includes:First acquisition submodule, it is special for obtaining first
The color information levied;Second acquisition submodule, for obtaining the pixel quantity of frame of video;First computing module, is used for
Calculate the shared percentage ratio in Target Photo of the size of fisrt feature;And second computing module, for calculating first
The corresponding pixel quantity of color information of feature, wherein, the corresponding pixel quantity of color information of fisrt feature is video
The pixel quantity of frame and the product of percentage ratio.
In embodiments of the present invention, just the color statistical data of video image preserved to storage during video acquisition
In server, when video image retrieval is carried out, the color statistical data for being first according to Target Photo is filtered out and target
The target video frame that the color statistical data of picture matches, then solves to the video data encoder of target video frame
Code, obtains decoded video image, then first in retrieving comprising Target Photo from decoded video image
The video image of feature, has reached in video image retrieving without the need for all of video image is decoded and analyzed
Purpose, it is achieved thereby that improving the technique effect of video image recall precision, and then solve prior art in image
Need to carry out image the interim decoding technical problem low with the image retrieval efficiency that analysis is caused in retrieving.
Description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this
Bright schematic description and description does not constitute inappropriate limitation of the present invention for explaining the present invention.In accompanying drawing
In:
Fig. 1 is the schematic diagram of video image searching system according to embodiments of the present invention;
Fig. 2 is that the handling process of the data message of video image in digital camera according to embodiments of the present invention is illustrated
Figure;
Fig. 3 is that the handling process of the data message of video image in analog video camera according to embodiments of the present invention is illustrated
Figure;
Fig. 4 is the flow chart of video image search method according to embodiments of the present invention;And
Fig. 5 is the schematic diagram that video image according to embodiments of the present invention retrieves device.
Specific embodiment
In order that those skilled in the art more fully understand the present invention program, below in conjunction with the embodiment of the present invention
Accompanying drawing, is clearly and completely described to the technical scheme in the embodiment of the present invention, it is clear that described embodiment
The only embodiment of a present invention part, rather than the embodiment of whole.Based on the embodiment in the present invention, ability
The every other embodiment that domain those of ordinary skill is obtained under the premise of creative work is not made, all should belong to
The scope of protection of the invention.
It should be noted that description and claims of this specification and the term " first " in above-mentioned accompanying drawing, "
Two " it is etc. the object for distinguishing similar, without for describing specific order or precedence.It should be appreciated that this
The data that sample is used can be exchanged in the appropriate case, so as to embodiments of the invention described herein can with except
Here the order beyond those for illustrating or describing is implemented.Additionally, term " comprising " and " having " and they
Any deformation, it is intended that covering is non-exclusive to be included, and for example, contains process, the side of series of steps or unit
Method, system, product or equipment are not necessarily limited to those steps clearly listed or unit, but may include unclear
List or other steps intrinsic for these processes, method, product or equipment or unit.
According to embodiments of the present invention, there is provided a kind of video image searching system, Fig. 1 is according to embodiments of the present invention
The schematic diagram of video image searching system, as shown in figure 1, the system includes:
Video camera 10, for obtaining the color statistical data and video data encoder of video image, and by color statistical number
Frame of video is packaged into according to video data encoder;And
Storage server 20, for obtaining the color statistical data of Target Photo;The encapsulated layer of frame of video is parsed,
Obtain the color statistical data of frame of video;The color statistical data of Target Photo is entered with the color statistical data of frame of video
Row matching, obtains matching result and meets pre-conditioned target video frame;And according to fisrt feature to target video frame
Image retrieval is carried out, wherein, the color statistical data of Target Photo is the color for characterizing fisrt feature in Target Photo
Data, frame of video is obtained after the color statistical data and video data encoder by frame of video is packaged.
Alternatively, the system can also include client 30, for Target Photo can be input into by client, wherein,
Target Photo is the picture that user wants to retrieve according to the actual requirements, and Target Photo can use private client by user
Input, the client can be example, in hardware, or software form.The retrieval source of Target Photo is by video camera 10
The video image composition of (such as digital camera or analog video camera) collection, i.e., from the video of the collection of video camera 10
Searched targets picture in image.Video camera 10 is by the data message of the video image for gathering (including the color of video image
Statistical data and video data encoder) send to storage server 20 and stored, storage server 20 and client
30 are interacted by network (cable network or wireless network).As shown in figure 1, video camera 10 will be collected
Video image processed, and the data message of the video image after process sent carry out into storage server 20
Storage, client 30 sends Target Photo to storage server 20, and storage server 20 is performed to Target Photo
Retrieval, and the retrieval result (video labeling being located including retrieving image and retrieving image etc.) of Target Photo is sent
To client 30.Alternatively, when storage server 20 meets the demand of runs software, can be by storage server 20
It is deployed on same hardware device with client 30, to reduce the biography of the data between client 30 and storage server 20
It is defeated for the consumption of the network bandwidth.
Storage server 20 is used for the data message of store video images, can be stored with storage server 20 one
Or the data message of the video image of multiple videos, wherein, different videos can be distinguished with different video labelings
Represent.Alternatively, the corresponding video of any one video labeling can include multiple frame of video in storage server 20,
Data message per frame video image includes the color statistical data and video data encoder of the frame video image, and color
Color statistical data and video data encoder are stored in storage server 20 with encapsulation format, i.e., in storage server 20
The encapsulation of data of multiple frame of video that are stored with, wherein, the encapsulation of data of frame of video for frame of video color statistical data with
The data that video data encoder is obtained after being packaged.
Frame of video is sent by video camera 10 in the corresponding video of any one video labeling in storage server 20,
Alternatively, video camera 10 is used in such a way by the color statistical data and Video coding of the video image for collecting
Data are packaged into frame of video, and the frame of video after encapsulation is sent to storage server 20:Set up color statistical data and
The corresponding relation of video data encoder;And video camera according to corresponding relation by color statistical data and video data encoder
It is packaged, obtains frame of video.
Alternatively, the color statistical data of video image includes:The color information and figure of characteristics of image in video image
As the corresponding pixel quantity of color information of feature.Wherein, video camera 10 is used to obtain video image in such a way
Color statistical data:Video camera 10 obtains the color information of characteristics of image in video image;And video camera 10 will
Each pixel in video image carries out color classification, the corresponding pixel quantity of color information of statistical picture feature.
Specifically, video camera 10 is used to perform following steps:
Step S1:Obtain the color information of characteristics of image in video image.Video camera, can be right when video image is gathered
Color examination and pixel count statistics of variables are carried out per frame video image, wherein, the number of species of color depends on pixel
The color space of point is calculated, and the search method of the video image of the embodiment is not concerned with color and screens computational methods itself.
Alternatively, obtaining the color information of characteristics of image in video image can occur during video acquisition, by color
Screen and pixel quantity statistics to obtain video image in characteristics of image color information.Obtain image in video image
The color information of feature can also take the video to storing in storage server 20 to decode, then to decoded
Each frame of video in video carries out the mode of color analysis, and to each frame of video in decoded video color is carried out
Analysis includes that color is screened and pixel quantity statistics.
Step S2:By the way that each pixel in video image is carried out into color classification, image is special in statistics video image
The corresponding pixel quantity of color information levied.Assume that the color of the embodiment screens algorithm and can screen out that " reddish orange is yellowish green
Seven kinds of colors of ultramarine purple ", then color classification is carried out to all pixels point of every frame video image, in order to count
The corresponding pixel quantity of every kind of color.
Alternatively, video camera 10 is used in such a way each pixel in video image is carried out into color classification,
The corresponding pixel quantity of color information of statistical picture feature:Video camera 10 obtains the color of the first pixel in video image
Value;Video camera 10 judges whether the color-values of the first pixel color-values corresponding with the color information of characteristics of image match;
And video camera 10 is if it is judged that the color-values of the first pixel and the corresponding color-values of the color information of characteristics of image
Match somebody with somebody, the color of the first pixel is classified as into the color information of characteristics of image, and the color information correspondence of accumulated image feature
Pixel quantity.That is, each pixel in video image is carried out color and is returned by video camera 10 by performing following steps
Class:
Step S21:The color-values of the first pixel in frame of video are obtained, wherein, the first pixel is in video frame images
Any one pixel.Video camera 10 can obtain first the yuv data of video image after collection video image, lead to
Often what collecting device (such as video camera) was exported is the video image of NV12 forms.Then it is the colourity of video image is empty
Between rgb space is converted to by YUV, it is assumed that the YUV values of video image pixel are (y, u, v), then conversion regime
It is shown below:
R=y+1.13983* (v-128)
G=y-0.39465* (u-128) -0.58060* (v-128)
B=y+2.03211* (u-128)
Then video image is converted to into HSV space from rgb space, it is assumed that the rgb value of video image pixel for (r,
G, b), then conversion regime is shown below:
Max=max (r, g, b)
Min=min (r, g, b)
V=max
Step S22:Whether the color-values color-values corresponding with the color information of characteristics of image for judging the first pixel match.
The corresponding color-values of color information of characteristics of image will be stored in default scene color cluster table.Default scene color
Cluster table is the different color cluster table set up previously according to monitoring scene according to actual demand, presets scene color and gathers
One or more colors can be included in class table, such as presetting scene color cluster table can be expressed as:
Color_tab [6]=(hi,si,vi),i∈[0:6]
Wherein, (hi,si,vi) it is " the yellowish green ultramarine of reddish orange the is purple " HSV of this 7 kinds of color center points under this scene mode
Value.
Step S23:If it is judged that the color-values of the first pixel color-values corresponding with the color information of characteristics of image
Match somebody with somebody, the color of the first pixel is classified as into the color information of characteristics of image, and the color information correspondence of accumulated image feature
Pixel quantity.Such as, by the color-values (h, v, s) of the first pixel of video image and default scene color cluster
The corresponding color center point (h of certain color in tablei,si,vi),i∈[0:6] matched, when the match is successful, by
The color of one pixel is sorted out and the color.
The video image searching system of the embodiment considers image under different monitoring scene (such as colour temperature, illumination etc.)
Can there is the situation of colour cast, if carrying out color statistics using identical color template, the accuracy of statistical result will
It is greatly affected.The video image searching system of the embodiment is subtracted using the default scene color cluster table for pre-building
Lack the problem declined because of the image retrieval accuracy that image color deviation is caused, reach the retrieval of raising video image accurate
The effect of exactness.
When corresponding with the color information of the characteristics of image color-values of the color-values for judging the first pixel are matched, by the
While the color of one pixel classifies as the color, by the corresponding color enumerator of the color information of the characteristics of image
ci,i∈[0:6] Jia one.The computational methods of the corresponding pixel quantity of color information of characteristics of image are shown below, its
In, T is empirical value, generally image noise variance:
ci+=1, if D ((h, s, v), (hi,si,vi)) < T
Institute of the method for the color statistical data of above-mentioned acquisition video image suitable for the corresponding video of any video labeling
There is the corresponding video image of frame of video.The video image searching system statistical picture of embodiment color character of overall importance,
Effectively evade the low problem of the image retrieval reliability for only causing to destination object detection, reach raising video figure
As the effect of retrieval reliability.
Alternatively, video camera 10 can be compiled after the color statistical data for obtaining video image to the video image
Code, obtains video data encoder.Video camera 10 sets up the right of the color statistical data of video image and video data encoder
Should be related to so that the color statistical data of video image is corresponded with video data encoder;Will according to the corresponding relation
The color statistical data of video image and video data encoder are packaged, and obtain the corresponding frame of video of the video image,
Wherein, the encapsulation of data of frame of video.The privately owned encapsulation that the embodiment is allowed when form is encapsulated using encapsulation format standard
Method.
Video camera 10 can obtain according to the method described above all frame of video in the corresponding video of any video labeling.Need
Illustrate, for analog video camera and digital camera, the collection of video image, the calculating of color statistical data,
The executive agent of coding and encapsulation flow process is slightly different.Fig. 2 is regarded in digital camera according to embodiments of the present invention
The handling process schematic diagram of the data message of frequency image, as shown in Fig. 2 for digital camera, collection video image,
Calculate color statistical data, coding and encapsulate flow process and complete inside digital camera.Fig. 3 is according to the present invention
The handling process schematic diagram of the data message of video image in the analog video camera of embodiment, as shown in figure 3, for mould
Intend video camera, analog video camera only complete in itself collection video image flow process, calculate color statistical data, coding with
And encapsulation flow process is completed by video encoder.
It should be noted that the data message of the video image in storage server 20 is for conventional live preview, biography
Defeated and storage etc. without any difference, difference be the video data encoder of every frame video image have to should frame regard
The color statistical data of frequency image, and the reading to the frame video image color statistical data need not be to the frame video image
Video data encoder decoding, it is only necessary to parsing to encapsulated layer by obtain, reached raising video image retrieve
The effect of efficiency.
Alternatively, storage server 20 supports addition, modification and the deletion function of the data message of video image, storage
The data message of the video image in server 20 can be prestored, it is also possible to be added in real time, changed
And deletion.After by the data information memory of video image to storage server 20, examine from storage server 20
Rope Target Photo, needs the Target Photo sent to client 30 to carry out color parsing, wherein, the calculating of color parsing
Method is identical with the computational methods of the color statistical data of above-mentioned video image, and here is omitted.
The color statistical data of Target Photo is corresponding including the color information of fisrt feature and the color information of fisrt feature
Pixel quantity.Such as, Target Photo is the picture comprising red car, then fisrt feature is car, fisrt feature
Color information is redness.Alternatively, obtaining the color statistical data of Target Photo includes:Obtain the color of fisrt feature
Information;And the corresponding pixel quantity of color information of acquisition fisrt feature.
Because Target Photo has very much with the video image to be retrieved in storage server 20 on dimension scale, image pixel
May be inconsistent, it is therefore desirable in the corresponding pixel quantity of object color component and storage server 20 of Target Photo
The pixel quantity of video image to be retrieved is adjusted correspondingly according to actual ratio, to obtain more definite aim colour
Color corresponding pixel quantity.Alternatively, storage server 20 obtains in the following manner the color information pair of fisrt feature
The pixel quantity answered:Obtain the pixel quantity of the frame of video prestored in storage server 20;Calculate fisrt feature
The shared percentage ratio in Target Photo of size;And the corresponding pixel quantity of color information of calculating fisrt feature, its
In, the corresponding pixel quantity of color information of fisrt feature is the pixel quantity and percentage of frame of video in storage server 20
The product of ratio.
Such as, it is assumed that Target Photo contains the car of a redness, fisrt feature is car, car picture sheet as
2000000 pixels, the color information of fisrt feature is redness, and frame of video is also 2,000,000 pixels in storage server 20.
In this video scene, need to convert the corresponding pixel quantity of color information of fisrt feature in Target Photo,
Assume that the shared percentage ratio in Target Photo of car is one of percentage, then the color information of the fisrt feature of Target Photo
Corresponding pixel quantity is redness, 20,000 pixels for the color statistical data of 20,000 pixels, i.e. Target Photo.
Alternatively, video image size in the size of fisrt feature in Target Photo and storage server 20 is carried out
Match somebody with somebody, there can be two methods:One kind is artificial matching process;Another kind is image algorithm matching process.Artificial matching
Method is referred to checks the real-time video that storage server 20 is stored by user, when occurring in that target object in real-time video
When (such as fisrt feature, car), picture is intercepted.Picture to intercepting uses grid method, calculates target
Object is the dimension scale in the picture for intercepting, so as to obtain under this video scene target object substantially as
Element value.Image algorithm matching is referred to chooses one section of video record for including all kinds of moving objects by user, by people's car
Thing sorting algorithm is classified to people's car thing, wherein, enter pedestrian's car thing sorting algorithm to video in field of machine vision
Jing is very ripe, is no longer discussed in detail herein.The result of sorting algorithm is carried out to people's car thing includes that people, car, thing are all kinds of
The normalization size of target, just can show that target object is regarded at this with reference to the pixel value of video image in storage server 20
Substantially pixel value under frequency scene.
Storage server 20 obtain Target Photo color statistical data after, according to the color statistical number of Target Photo
According to the target that the color statistical data that color statistical data and Target Photo are obtained from the frame of video for prestoring matches
Frame of video.It should be noted that the color statistical data of the pre-conditioned preferably Target Photo of matching result satisfaction is
The subset of the color statistical data of target video frame, as long as that is, the color statistical data of Target Photo is target video frame
The subset of color statistical data, so that it may meet the color statistical data phase of object color component statistical data and target video frame
Match somebody with somebody.Assume Target Photo color statistical data be:Redness, 20,000 pixels, as long as the then color system of target video frame
Counting includes " red, 20,000 pixels ", such as the color statistical data of target video frame is:Redness, 20,000 pictures
Element;Blueness, 20,000 pixels, you can meet the color statistical data of Target Photo and the color statistical number of target video frame
According to matching.In the searched targets picture from storage server, need to parse the encapsulated layer of video image,
In order to read the color statistical data of video image.
Alternatively, storage server 20 can be by performing following steps by the color statistical data and video of Target Photo
The color statistical data of frame is matched, and to obtain matching result pre-conditioned target video frame is met:
Step S4:The color statistical data of the first frame of video is obtained, wherein, what is stored in storage server 20 is multiple
Frame of video includes the first frame of video, and the first frame of video can be that any one in the corresponding video of any video labeling is regarded
Frequency frame.Obtaining the color statistical data of the first frame of video equally includes:Obtain the color of the first color in the first frame of video
Information;And the corresponding pixel quantity of the first color of acquisition.Wherein, the first color is certain in the first video frame images
The color of one feature.
Step S5:Judge whether the color statistical data of the first frame of video matches with the color statistical data of Target Photo.
Specifically include:Judge whether the color information of the first color matches with the color information of object color component;And judge first
Whether the corresponding pixel quantity of color pixel quantity corresponding with object color component matches.If in the image of the first frame of video
Do not include in object color component, or the first video frame images and include object color component, but the corresponding pixel count of object color component
Amount is far smaller than the corresponding pixel quantity of object color component in Target Photo, then directly skip first frame of video, continues to sentence
Next video frame images in disconnected storage server.
Step S6:If it is judged that the color statistical data of the color statistical data of the first frame of video and Target Photo
Match somebody with somebody, the first frame of video is defined as into target video frame.That is object color component is included in the first video frame images, and first regards
The corresponding pixel quantity of object color component pixel quantity corresponding with object color component in Target Photo is consistent in frequency two field picture, then
First frame of video is defined as into target video frame.
Alternatively, storage server 20 is obtained after target video frame, target video frame will be carried out according to fisrt feature
Image retrieval, specifically, storage server 20 can carry out image retrieval by performing following steps to target video frame:
Step S7, the video data encoder of target video frame is decoded, and obtains decoded video image.
After target video frame is obtained, the video data encoder of 20 pairs of target video frames of storage server is solved
Code, obtains decoded video image.Color statistical data of the storage server 20 of the embodiment according to Target Photo
The target that color statistical data matches with the color statistical data of Target Photo is filtered out from the frame of video for prestoring
Frame of video, color statistical data is filtered out with the obvious unmatched frame of video of the color statistical data of Target Photo, only
The video data encoder of target video frame decoded, it is to avoid decoding is consumed, overcome in prior art to all
Video frame images carry out decoding the low problem of the image retrieval efficiency for causing, and then have reached the substantial amounts of computing resource of saving
Consume, improve the technique effect of video image recall precision.
Whether step S8, judge in decoded video image comprising fisrt feature.
Step S9, if it is judged that fisrt feature is included in decoded video image, then by decoded video image
It is defined as the retrieval result of Target Photo.
With Target Photo of the retrieval comprising red car from any video (fisrt feature as car, the color of fisrt feature
Multimedia message is redness, and the corresponding pixel quantity of color information of fisrt feature is 20,000 pixels) as a example by, in the video figure
The color statistical data for describing every two field picture is read in the encapsulated layer of picture, namely per frame video data COLOR COMPOSITION THROUGH DISTRIBUTION feelings
The private data of condition, if without redness in the two field picture, or red pixel quantity is far smaller than 20,000 pixels, then
Directly skip the frame video, on the contrary then perform it is conventional to scheme to search nomography, will the frame video image Video coding
Whether decoding data, obtains decoded video image, then judge comprising car in decoded video image,
If it is judged that including car in decoded video image, then the frame video image is the retrieval result of Target Photo.
Wherein, retrieval result can include result picture and the corresponding video of result picture.
Alternatively, storage server 20 is additionally operable to send retrieval result after the retrieval result for obtaining Target Photo
To client 30, and then complete once to scheme to search the complete operation of figure.
The video image searching system of the embodiment is according to the color statistical data of Target Photo from the frame of video for prestoring
In filter out the target video frame matched with the color statistical data of Target Photo, then to the video of target video frame
Coded data is decoded, and obtains decoded video image, then is retrieved comprising mesh from decoded video image
The video image of the fisrt feature marked on a map in piece.The video image searching system of the embodiment solves prior art in figure
As needing that the interim decoding technical problem low with the image retrieval efficiency that analysis is caused is carried out to image in retrieving, reach
The technique effect for improving video image recall precision is arrived.Meanwhile, the video image searching system of the embodiment is in video
Just the color statistical data of all video images stored to storage server in gatherer process, by counting video image
Color characteristic of overall importance, effectively evaded only carries out detecting the problem that the reliability for causing is low to destination object, and
And the video image searching system of the embodiment reduces pattern colour using the default scene color cluster table for pre-building
The problem that accuracy rate that color deviation is caused declines, and then reached the accuracy for improving video image retrieval and reliability
Technique effect.
According to embodiments of the present invention, a kind of video image search method is additionally provided, it should be noted that in accompanying drawing
The step of flow process is illustrated can perform in the such as computer system of one group of computer executable instructions, though also,
Logical order is so shown in flow charts, but in some cases, institute can be performed with the order being different from herein
The step of illustrating or describe.
The video image search method of the embodiment is performed by storage server provided in an embodiment of the present invention.Fig. 4 is root
According to the flow chart of the video image search method of the embodiment of the present invention, as shown in figure 4, the method comprises the steps:
Step S402, obtains the color statistical data of Target Photo, wherein, the color statistical data of Target Photo is table
Levy the data of the color of fisrt feature in Target Photo;
Step S404, the encapsulated layer of the frame of video to prestoring is parsed, and obtains the color statistical data of frame of video,
Wherein, frame of video is obtained after the color statistical data and video data encoder by frame of video is packaged;
Step S406, the color statistical data of Target Photo is matched with the color statistical data of frame of video, is obtained
Matching result meets pre-conditioned target video frame;And
Step S408, image retrieval is carried out according to fisrt feature to target video frame.
Above-mentioned steps are the retrieving of Target Photo, and the retrieving can be examined by the video image of the embodiment of the present invention
Storage server in cable system is performed, and Target Photo can be by the video image searching system of the embodiment of the present invention
Client input, Target Photo retrieval source can be by the shooting in the video image searching system of the embodiment of the present invention
Machine is gathered.
Can be stored with storage server one or more video video image data message, wherein, it is different
Video can be distinguished with different video labeling and represent.Alternatively, any one video labeling pair in storage server
The video answered can include multiple frame of video, and the data message per frame video image is including the color of the frame video image
Statistical data and video data encoder, and color statistical data and video data encoder are stored in storage clothes with encapsulation format
In business device, i.e., the encapsulation of data of multiple frame of video that are stored with storage server, wherein, the encapsulation of data of frame of video is
The data that the color statistical data of frame of video is obtained after being packaged with video data encoder.
Frame of video is sent by video camera in the corresponding video of any one video labeling in storage server, optional
The color statistical data and video data encoder of the video image for collecting are packaged into frame of video by ground, video camera, and will
Frame of video after encapsulation sends to storage server and includes:Set up the correspondence pass of color statistical data and video data encoder
System;And video camera is packaged in color statistical data and video data encoder according to corresponding relation, obtains frame of video.
Alternatively, the color statistical data of video image includes:The color information and figure of characteristics of image in video image
As the corresponding pixel quantity of color information of feature.Wherein, the color statistical data of video camera acquisition video image includes:
Step S1:Obtain the color information of characteristics of image in video image.Video camera, can be right when video image is gathered
Color examination and pixel count statistics of variables are carried out per frame video image, wherein, the number of species of color depends on pixel
The color space of point is calculated, and the search method of the video image of the embodiment is not concerned with color and screens computational methods itself.
Alternatively, obtaining the color information of characteristics of image in video image can occur during video acquisition, by color
Screen and pixel quantity statistics to obtain video image in characteristics of image color information.Obtain image in video image
The color information of feature can also take the video to storing in storage server to decode, and then regard to decoded
Each frame of video in frequency carries out the mode of color analysis, and color point is carried out to each frame of video in decoded video
Analysis includes that color is screened and pixel quantity statistics.
Step S2:By the way that each pixel in video image is carried out into color classification, image is special in statistics video image
The corresponding pixel quantity of color information levied.Assume that the color of the embodiment screens algorithm and can screen out that " reddish orange is yellowish green
Seven kinds of colors of ultramarine purple ", then color classification is carried out to all pixels point of every frame video image, in order to count
The corresponding pixel quantity of every kind of color.
Alternatively, each pixel in video image is carried out color classification, the color of statistical picture feature by video camera
The corresponding pixel quantity of information is comprised the following steps:
Step S21:The color-values of the first pixel in frame of video are obtained, wherein, the first pixel is in video frame images
Any one pixel.Video camera can obtain first the yuv data of video image, generally after collection video image
What collecting device (such as video camera) was exported is the video image of NV12 forms.Then by the chrominance space of video image
Rgb space is converted to by YUV, it is assumed that the YUV values of video image pixel are (y, u, v), then conversion regime is such as
Shown in following formula:
R=y+1.13983* (v-128)
G=y-0.39465* (u-128) -0.58060* (v-128)
B=y+2.03211* (u-128)
Then video image is converted to into HSV space from rgb space, it is assumed that the rgb value of video image pixel for (r,
G, b), then conversion regime is shown below:
Max=max (r, g, b)
Min=min (r, g, b)
V=max
Step S22:Whether the color-values color-values corresponding with the color information of characteristics of image for judging the first pixel match.
The corresponding color-values of color information of characteristics of image will be stored in default scene color cluster table.Default scene color
Cluster table is the different color cluster table set up previously according to monitoring scene according to actual demand, presets scene color and gathers
One or more colors can be included in class table, such as presetting scene color cluster table can be expressed as:
Color_tab [6]=(hi,si,vi),i∈[0:6]
Wherein, (hi,si,vi) it is " the yellowish green ultramarine of reddish orange the is purple " HSV of this 7 kinds of color center points under this scene mode
Value.
Step S23:If it is judged that the color-values of the first pixel color-values corresponding with the color information of characteristics of image
Match somebody with somebody, the color of the first pixel is classified as into the color information of characteristics of image, and the color information correspondence of accumulated image feature
Pixel quantity.Such as, by the color-values (h, v, s) of the first pixel of video image and default scene color cluster
The corresponding color center point (h of certain color in tablei,si,vi),i∈[0:6] matched, when the match is successful, by
The color of one pixel is sorted out and the color.
The video image search method of the embodiment considers image under different monitoring scene (such as colour temperature, illumination etc.)
Can there is the situation of colour cast, if carrying out color statistics using identical color template, the accuracy of statistical result will
It is greatly affected.The video image search method of the embodiment is subtracted using the default scene color cluster table for pre-building
Lack the problem declined because of the image retrieval accuracy that image color deviation is caused, reach the retrieval of raising video image accurate
The effect of exactness.
When corresponding with the color information of the characteristics of image color-values of the color-values for judging the first pixel are matched, by the
While the color of one pixel classifies as the color, by the corresponding color enumerator of the color information of the characteristics of image
ci,i∈[0:6] Jia one.The computational methods of the corresponding pixel quantity of color information of characteristics of image are shown below, its
In, T is empirical value, generally image noise variance:
ci+=1, if D ((h, s, v), (hi,si,vi)) < T,
Institute of the method for the color statistical data of above-mentioned acquisition video image suitable for the corresponding video of any video labeling
There is the corresponding video image of frame of video.The video image search method statistical picture of embodiment color character of overall importance,
Effectively evade the low problem of the image retrieval reliability for only causing to destination object detection, reach raising video figure
As the effect of retrieval reliability.
Alternatively, video camera can be encoded after the color statistical data for obtaining video image to the video image,
Obtain video data encoder.Video camera sets up the color statistical data of video image and the corresponding relation of video data encoder,
So that the color statistical data of video image is corresponded with video data encoder;According to the corresponding relation by video image
Color statistical data and video data encoder be packaged, obtain the corresponding frame of video of the video image, wherein, depending on
The encapsulation of data of frequency frame.The privately owned method for packing that the embodiment is allowed when form is encapsulated using encapsulation format standard.
Video camera is obtained after all frame of video in the corresponding video of any video labeling, sends it to storage service
Device.It should be noted that the data message of the video image in storage server is for conventional live preview, transmission
And storage etc. is without any difference, difference is that the video data encoder of every frame video image has to should frame video
The color statistical data of image, and reading without the need for the frame video image to the frame video image color statistical data
Video data encoder is decoded, it is only necessary to obtained by being parsed to encapsulated layer, has reached raising video image retrieval effect
The effect of rate.
Alternatively, storage server supports addition, modification and the deletion function of the data message of video image, storage clothes
The data message of the video image in business device can be prestored, it is also possible to be added in real time, change and deleted
Remove.After by the data information memory of video image to storage server, the searched targets picture from storage server,
The Target Photo sent to client is needed to carry out color parsing, wherein, the computational methods and above-mentioned video of color parsing
The computational methods of the color statistical data of image are identical, and here is omitted.
The color statistical data of Target Photo is corresponding including the color information of fisrt feature and the color information of fisrt feature
Pixel quantity.Such as, Target Photo is the picture comprising red car, then fisrt feature is car, fisrt feature
Color information is redness.Alternatively, obtaining the color statistical data of Target Photo includes:Obtain the color of fisrt feature
Information;And the corresponding pixel quantity of color information of acquisition fisrt feature.
Due to the video image to be retrieved in Target Photo and storage server has very much on dimension scale, image pixel can
Can be inconsistent, it is therefore desirable to be checked in the corresponding pixel quantity of object color component and storage server of Target Photo
The pixel quantity of rope video image is adjusted correspondingly according to actual ratio, to obtain more definite object color component pair
The pixel quantity answered.Alternatively, the corresponding pixel quantity of color information of storage server acquisition fisrt feature includes:
Obtain the pixel quantity of the frame of video prestored in storage server;The size of calculating fisrt feature is in Target Photo
Shared percentage ratio;And the corresponding pixel quantity of color information of calculating fisrt feature, wherein, the color of fisrt feature
The corresponding pixel quantity of multimedia message is the product of the pixel quantity of frame of video and percentage ratio in storage server.
Such as, it is assumed that Target Photo contains the car of a redness, fisrt feature is car, car picture sheet as
2000000 pixels, the color information of fisrt feature is redness, and frame of video is also 2,000,000 pixels in storage server.
In this video scene, need to convert the corresponding pixel quantity of color information of fisrt feature in Target Photo, it is false
If the shared percentage ratio in Target Photo of car is one of percentage, then the color information pair of the fisrt feature of Target Photo
The pixel quantity answered is redness, 20,000 pixels for the color statistical data of 20,000 pixels, i.e. Target Photo.
Alternatively, the size of fisrt feature in Target Photo is matched with video image size in storage server,
There can be two methods:One kind is artificial matching process;Another kind is image algorithm matching process.Artificial matching process
Refer to and the real-time video that storage server is stored is checked by user, when occurring in that target object (such as the in real-time video
One feature, car) when, intercept picture.Picture to intercepting uses grid method, calculates target object and is
Dimension scale in the picture for intercepting, so as to obtain the substantially pixel value of the target object under this video scene.Figure
One section of video record for including all kinds of moving objects is chosen by user as algorithmic match is referred to, is calculated by the classification of people's car thing
Method is classified to people's car thing, wherein, enter pedestrian's car thing sorting algorithm to video in field of machine vision very ripe
, no longer it is discussed in detail herein.Carrying out to people's car thing the result of sorting algorithm includes that people, car, all kinds of targets of thing are returned
One changes size, just can draw target object under the video scene with reference to the pixel value of video image in storage server
Substantially pixel value.
Storage server is after the color statistical data for obtaining Target Photo, it will execution step S406, i.e., will root
The color of color statistical data and Target Photo is obtained from the frame of video for prestoring according to the color statistical data of Target Photo
The target video frame that color statistical data matches.It should be noted that the pre-conditioned preferably mesh that matching result meets
Mark on a map piece color statistical data for target video frame color statistical data subset, as long as the i.e. color of Target Photo
Statistical data is the subset of the color statistical data of target video frame, so that it may meets object color component statistical data and regards with target
The color statistical data of frequency frame matches.Assume Target Photo color statistical data be:Redness, 20,000 pixels, then
As long as the color statistical data of target video frame includes " red, 20,000 pixels ", such as the color system of target video frame
Count for:Redness, 20,000 pixels;Blueness, 20,000 pixels, you can meet the color statistical data and mesh of Target Photo
The color statistical data of mark frame of video matches.In the searched targets picture from storage server, need to video figure
The encapsulated layer of picture is parsed, in order to read the color statistical data of video image.
Alternatively, the color statistical data of frame of video includes the color information and each figure of each characteristics of image in frame of video
As the corresponding pixel quantity of color information of feature, wherein, by the color statistical data of Target Photo and the color of frame of video
Color statistical data carries out matching to be included:The color information of the color information of fisrt feature and each characteristics of image is carried out
Match somebody with somebody;And by the corresponding pixel quantity of the color information of fisrt feature picture corresponding with the color information of each characteristics of image
Prime number amount is matched.Specifically, step S406 storage server is by the color statistical data and frame of video of Target Photo
Color statistical data matched, with obtain matching result meet pre-conditioned target video frame can include it is following
Step:
Step S4:The color statistical data of the first frame of video is obtained, wherein, the multiple videos stored in storage server
Frame includes the first frame of video, and the first frame of video can be any one frame of video in the corresponding video of any video labeling.
Obtaining the color statistical data of the first frame of video equally includes:Obtain the color information of the first color in the first frame of video;
And the corresponding pixel quantity of the first color of acquisition.Wherein, the first color is a certain feature in the first video frame images
Color.
Step S5:Judge whether the color statistical data of the first frame of video matches with the color statistical data of Target Photo.
Specifically include:Judge whether the color information of the first color matches with the color information of object color component, wherein, aim colour
Coloured silk is the color of fisrt feature in Target Photo;And judge that the corresponding pixel quantity of the first color is corresponding with object color component
Pixel quantity whether match.If object color component is not included in the image of the first frame of video, or the first frame of video
Object color component is included in image, but the corresponding pixel quantity of object color component is far smaller than object color component correspondence in Target Photo
Pixel quantity, then directly skip first frame of video, continuation judges the next video frame images in storage server.
Step S6:If it is judged that the color statistical data of the color statistical data of the first frame of video and Target Photo
Match somebody with somebody, the first frame of video is defined as into target video frame.That is object color component is included in the first video frame images, and first regards
The corresponding pixel quantity of object color component pixel quantity corresponding with object color component in Target Photo is consistent in frequency two field picture, then
First frame of video is defined as into target video frame.
Alternatively, step S408 carries out image retrieval and comprises the following steps according to fisrt feature to target video frame:
Step S7, the video data encoder of target video frame is decoded, and obtains decoded video image.
After target video frame is obtained, storage server is decoded to the video data encoder of the target video frame,
Obtain decoded video image.The storage server of the embodiment is according to the color statistical data of Target Photo from advance
The target video that color statistical data matches with the color statistical data of Target Photo is filtered out in the frame of video of storage
Frame, color statistical data is filtered out with the obvious unmatched frame of video of the color statistical data of Target Photo, only to mesh
The video data encoder of mark frame of video is decoded, it is to avoid decoding is consumed, and is overcome in prior art to all videos
Two field picture carries out decoding the low problem of the image retrieval efficiency for causing, and then has reached the substantial amounts of computing resource consumption of saving,
Improve the technique effect of video image recall precision.
Whether step S8, judge in decoded video image comprising fisrt feature.
Step S9, if it is judged that fisrt feature is included in decoded video image, then by decoded video image
It is defined as the retrieval result of Target Photo.
With Target Photo of the retrieval comprising red car from any video (fisrt feature as car, the color of fisrt feature
Multimedia message is redness, and the corresponding pixel quantity of color information of fisrt feature is 20,000 pixels) as a example by, in the video figure
The color statistical data for describing every two field picture is read in the encapsulated layer of picture, namely per frame video data COLOR COMPOSITION THROUGH DISTRIBUTION feelings
The private data of condition, if without redness in the two field picture, or red pixel quantity is far smaller than 20,000 pixels, then
Directly skip the frame video, on the contrary then perform it is conventional to scheme to search nomography, will the frame video image Video coding
Whether decoding data, obtains decoded video image, then judge comprising car in decoded video image,
If it is judged that including car in decoded video image, then the frame video image is the retrieval result of Target Photo.
Wherein, retrieval result can include result picture and the corresponding video of result picture.
Alternatively, retrieval result can be sent to client by storage server after the retrieval result for obtaining Target Photo
End, and then complete once to scheme to search the complete operation of figure.
The video image search method of the embodiment is according to the color statistical data of Target Photo from the frame of video for prestoring
In filter out the target video frame matched with the color statistical data of Target Photo, then to the video of target video frame
Coded data is decoded, and obtains decoded video image, then is retrieved comprising mesh from decoded video image
The video image of the fisrt feature marked on a map in piece.The video image search method of the embodiment solves prior art in figure
As needing that the interim decoding technical problem low with the image retrieval efficiency that analysis is caused is carried out to image in retrieving, reach
The technique effect for improving video image recall precision is arrived.
According to embodiments of the present invention, a kind of video image retrieval device is additionally provided, it should be noted that the embodiment
In video image retrieval device can be used for performing the video image search method in the embodiment of the present invention, the present invention is real
Perform in the video image retrieval device that the video image search method applied in example can also be in this embodiment.
Fig. 5 is the schematic diagram that video image according to embodiments of the present invention retrieves device, as shown in figure 5, the video figure
As retrieval device includes:First acquisition module 52, parsing module 54, the second acquisition module 56 and retrieval module 58.
First acquisition module 52, for obtaining the color statistical data of Target Photo, wherein, the color system of Target Photo
The data of the color of fisrt feature in counting to characterize Target Photo.
Alternatively, the color statistical data of Target Photo includes the color information of fisrt feature and the color letter of fisrt feature
Corresponding pixel quantity is ceased, wherein, the first acquisition module 52 includes:First acquisition submodule, it is special for obtaining first
The color information levied;Second acquisition submodule, for obtaining the pixel quantity of frame of video;First computing module, is used for
Calculate the shared percentage ratio in Target Photo of the size of fisrt feature;And second computing module, for calculating first
The corresponding pixel quantity of color information of feature, wherein, the corresponding pixel quantity of color information of fisrt feature is video
The pixel quantity of frame and the product of percentage ratio.
Parsing module 54, for parsing to the encapsulated layer of the frame of video for prestoring, obtains the color system of frame of video
Count, wherein, frame of video is obtained after the color statistical data and video data encoder by frame of video is packaged
's.
Second acquisition module 56, for the color statistical data of Target Photo to be carried out with the color statistical data of frame of video
Matching, obtains matching result and meets pre-conditioned target video frame.
Alternatively, the color statistical data of frame of video includes the color information and each figure of each characteristics of image in frame of video
As the corresponding pixel quantity of color information of feature, wherein, the second acquisition module 56 includes:First matching module, uses
In the color information of fisrt feature is matched with the color information of each characteristics of image;And second matching module,
For by the corresponding pixel quantity of the color information of fisrt feature pixel count corresponding with the color information of each characteristics of image
Amount is matched.
Retrieval module 58, for carrying out image retrieval to target video frame according to fisrt feature.
Alternatively, retrieving module 58 includes:Decoder module, for the video data encoder of target video frame to be solved
Code, obtains decoded video image;Judge module, for judging decoded video image in whether include first
Feature;And determining module, for when fisrt feature is included in judging decoded video image, then decoding
Video image afterwards is defined as the retrieval result of Target Photo.
The first acquisition module 52 in the video image retrieval device of the embodiment can be used for performing the embodiment of the present invention
Video image search method in step S402, parsing module 54 can be used for perform the embodiment of the present invention video
Step S404 in image search method, the second acquisition module 56 can be used for performing the video figure of the embodiment of the present invention
As step S406 in search method, retrieve module 58 and can be used for performing the video image retrieval of the embodiment of the present invention
Step S408 in method.Device is retrieved by the video image of the embodiment, prior art is solved in image retrieval
During need to carry out image the interim decoding technical problem low with the image retrieval efficiency that analysis is caused, reached and carried
The technique effect of high video image recall precision.
By above-mentioned analysis, the present invention reaches following technique effect:
First, it is of the invention by advance the color statistical data of video image being stored to storage server, in video figure
As in retrieving, filtering out the target matched with the color statistical data of Target Photo from storage server first
Frame of video, then carries out scheming to search nomography to target video frame, obtains the retrieval result of Target Photo, and the present invention is only
Pair video frame images matched with the color statistical data of Target Photo are needed to decode, it is to avoid to all videos
Image is decoded and is analyzed the low problem of the recall precision for causing.
Secondly, the present invention has effectively been evaded only to destination object by statistics video image color information of overall importance
Enter the low problem of retrieval reliability that line retrieval causes.
Again, the present invention adopts the corresponding pixel quantity of default scene color cluster table analysis object color component for pre-building,
Effectively reduce the low problem of the retrieval rate caused because of image color deviation.
The present invention can be used for video image retrieval, can especially effectively improve the recall precision of massive video image, utilize
The present invention can efficiently reduce artificial investigation cost, when shortening the positioning of critical eventss and target in massive video image
Between.
The embodiments of the present invention are for illustration only, do not represent the quality of embodiment.
In the above embodiment of the present invention, the description to each embodiment all emphasizes particularly on different fields, and does not have in certain embodiment
The part of detailed description, may refer to the associated description of other embodiment.
In several embodiments provided herein, it should be understood that disclosed technology contents, other can be passed through
Mode realize.Wherein, device embodiment described above is only schematic, such as division of described unit,
Can be a kind of division of logic function, can there is other dividing mode, such as multiple units or component when actually realizing
Can with reference to or be desirably integrated into another system, or some features can be ignored, or not perform.It is another, institute
The coupling each other for showing or discussing or direct-coupling or communication connection can be by some interfaces, unit or mould
The INDIRECT COUPLING of block or communication connection, can be electrical or other forms.
The unit as separating component explanation can be or may not be it is physically separate, it is aobvious as unit
The part for showing can be or may not be physical location, you can with positioned at a place, or can also be distributed to
On multiple units.Some or all of unit therein can be according to the actual needs selected to realize this embodiment scheme
Purpose.
In addition, each functional unit in each embodiment of the invention can be integrated in a processing unit, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a unit.It is above-mentioned integrated
Unit both can be realized in the form of hardware, it would however also be possible to employ the form of SFU software functional unit is realized.
If the integrated unit is realized using in the form of SFU software functional unit and as independent production marketing or used
When, during a computer read/write memory medium can be stored in.Based on such understanding, technical scheme
The part for substantially contributing to prior art in other words or all or part of the technical scheme can be produced with software
The form of product is embodied, and the computer software product is stored in a storage medium, including some instructions are to make
Obtain a computer equipment (can be personal computer, server or network equipment etc.) and perform each enforcement of the present invention
The all or part of step of example methods described.And aforesaid storage medium includes:USB flash disk, read only memory (ROM,
Read-Only Memory), random access memory (RAM, Random Access Memory), portable hard drive, magnetic
Dish or CD etc. are various can be with the medium of store program codes.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improve and moisten
Decorations also should be regarded as protection scope of the present invention.
Claims (10)
1. a kind of video image search method, it is characterised in that include:
The color statistical data of Target Photo is obtained, wherein, the color statistical data of the Target Photo is sign
The data of the color of fisrt feature in the Target Photo;
The encapsulated layer of the frame of video to prestoring is parsed, and obtains the color statistical data of the frame of video,
Wherein, the frame of video is after the color statistical data and video data encoder by the frame of video is packaged
Obtain;
The color statistical data of the Target Photo is matched with the color statistical data of the frame of video, is obtained
Take matching result and meet pre-conditioned target video frame;And
Image retrieval is carried out to the target video frame according to the fisrt feature.
2. video image search method according to claim 1, it is characterised in that the color system of the Target Photo
Count the corresponding pixel count of the color information of color information and the fisrt feature including the fisrt feature
Amount, wherein, obtaining the color statistical data of Target Photo includes:
Obtain the color information of the fisrt feature;
Obtain the pixel quantity of the frame of video;
Calculate the shared percentage ratio in the Target Photo of the size of the fisrt feature;And
The corresponding pixel quantity of color information of the fisrt feature is calculated, wherein, the color of the fisrt feature
The corresponding pixel quantity of information is the product of the pixel quantity with the percentage ratio of the frame of video.
3. video image search method according to claim 2, it is characterised in that the color statistics of the frame of video
Data include the color information and the color information of each characteristics of image of each characteristics of image in the frame of video
Corresponding pixel quantity, wherein, the color statistical data of the Target Photo is united with the color of the frame of video
Counting carries out matching and includes:
The color information of the fisrt feature is matched respectively with the color information of each characteristics of image;
And
By the corresponding pixel quantity of the color information of the fisrt feature respectively with the color of each characteristics of image
The corresponding pixel quantity of information is matched.
4. video image search method according to claim 1, it is characterised in that according to the fisrt feature to institute
Stating target video frame and carrying out image retrieval includes:
The video data encoder of the target video frame is decoded, decoded video image is obtained;
Whether judge in the decoded video image comprising the fisrt feature;And
If it is judged that the fisrt feature is included in the decoded video image, then will be described decoded
Video image is defined as the retrieval result of the Target Photo.
5. a kind of video image searching system, it is characterised in that include:
Video camera, for obtaining the color statistical data and video data encoder of video image, and by the color
Statistical data and the video data encoder are packaged into frame of video;And
Storage server, for obtaining the color statistical data of Target Photo;The encapsulated layer of the frame of video is entered
Row parsing, obtains the color statistical data of the frame of video;By the color statistical data of the Target Photo and institute
The color statistical data for stating frame of video is matched, and is obtained matching result and is met pre-conditioned target video frame;
And image retrieval is carried out to the target video frame according to fisrt feature, wherein, the color of the Target Photo
Statistical data is the data for characterizing the color of fisrt feature described in the Target Photo, and the frame of video is to incite somebody to action
What the color statistical data of the frame of video and video data encoder were obtained after being packaged.
6. video image searching system according to claim 5, it is characterised in that the video camera be used for according to
The color statistical data and the video data encoder are packaged into frame of video by under type:
Set up the corresponding relation of the color statistical data and the video data encoder;And
The color statistical data and the video data encoder are packaged according to the corresponding relation, are obtained
The frame of video.
7. video image searching system according to claim 6, it is characterised in that the video camera be used for according to
Under type obtains the color statistical data of the video image:
Obtain the color information of characteristics of image in the video image;And
Each pixel in the video image is carried out into color classification, the color letter of described image feature is counted
Cease corresponding pixel quantity.
8. video image searching system according to claim 7, it is characterised in that the video camera be used for according to
Each pixel in the video image is carried out color classification by under type, counts the color of described image feature
The corresponding pixel quantity of information:
Obtain the color-values of the first pixel in the video image;
Judge the color-values color-values corresponding with the color information of described image feature of first pixel whether
Match somebody with somebody;And
If it is judged that the color-values of first pixel color-values corresponding with the color information of described image feature
Matching, by the color of first pixel color information of described image feature, and cumulative described image are classified as
The corresponding pixel quantity of color information of feature.
9. a kind of video image retrieves device, it is characterised in that include:
First acquisition module, for obtaining the color statistical data of Target Photo, wherein, the Target Photo
Color statistical data is the data for characterizing the color of fisrt feature in the Target Photo;
Parsing module, for parsing to the encapsulated layer of the frame of video for prestoring, obtains the frame of video
Color statistical data, wherein, the frame of video is in the color statistical data and Video coding by the frame of video
What data were obtained after being packaged;
Second acquisition module, for the color statistical data of the Target Photo to be united with the color of the frame of video
Count and matched, obtain matching result and meet pre-conditioned target video frame;And
Retrieval module, for carrying out image retrieval to the target video frame according to the fisrt feature.
10. video image according to claim 9 retrieves device, it is characterised in that the color system of the Target Photo
Count the corresponding pixel count of the color information of color information and the fisrt feature including the fisrt feature
Amount, wherein, first acquisition module includes:
First acquisition submodule, for obtaining the color information of the fisrt feature;
Second acquisition submodule, for obtaining the pixel quantity of the frame of video;
First computing module, the shared percentage in the Target Photo of the size for calculating the fisrt feature
Than;And
Second computing module, the corresponding pixel quantity of the color information for calculating the fisrt feature, wherein,
The corresponding pixel quantity of color information of the fisrt feature is the pixel quantity and the percentage ratio of the frame of video
Product.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510694753.1A CN106610987B (en) | 2015-10-22 | 2015-10-22 | Video image retrieval method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510694753.1A CN106610987B (en) | 2015-10-22 | 2015-10-22 | Video image retrieval method, device and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106610987A true CN106610987A (en) | 2017-05-03 |
CN106610987B CN106610987B (en) | 2020-02-21 |
Family
ID=58613584
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510694753.1A Active CN106610987B (en) | 2015-10-22 | 2015-10-22 | Video image retrieval method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106610987B (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107481187A (en) * | 2017-09-29 | 2017-12-15 | 康佳集团股份有限公司 | A kind of method of video image processing, intelligent terminal and storage medium |
CN107609104A (en) * | 2017-09-12 | 2018-01-19 | 中广热点云科技有限公司 | The method and system of associated video is searched according to video image material |
CN109584317A (en) * | 2018-12-24 | 2019-04-05 | 天津天地伟业机器人技术有限公司 | Body color recognition methods based on HSV color space histogram |
CN110598048A (en) * | 2018-05-25 | 2019-12-20 | 北京中科寒武纪科技有限公司 | Video retrieval method and video retrieval mapping relation generation method and device |
CN111177190A (en) * | 2018-11-13 | 2020-05-19 | 杭州海康威视数字技术股份有限公司 | Data processing method and device, electronic equipment and readable storage medium |
CN111311697A (en) * | 2020-03-19 | 2020-06-19 | 北京搜狐新媒体信息技术有限公司 | Method for detecting color richness of picture and related device |
CN111382306A (en) * | 2018-12-28 | 2020-07-07 | 杭州海康威视数字技术股份有限公司 | Method and device for inquiring video frame |
CN113343033A (en) * | 2021-06-29 | 2021-09-03 | 平安普惠企业管理有限公司 | Video searching method and device, computer equipment and storage medium |
CN114741553A (en) * | 2022-03-31 | 2022-07-12 | 慧之安信息技术股份有限公司 | Image feature-based picture searching method |
WO2022170866A1 (en) * | 2021-02-10 | 2022-08-18 | Oppo广东移动通信有限公司 | Data transmission method and apparatus, and storage medium |
US11995556B2 (en) | 2018-05-18 | 2024-05-28 | Cambricon Technologies Corporation Limited | Video retrieval method, and method and apparatus for generating video retrieval mapping relationship |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1340178A (en) * | 1999-08-17 | 2002-03-13 | 皇家菲利浦电子有限公司 | System and method for performing region-based image retrieval using color-based segmentation |
CN1426002A (en) * | 2001-11-27 | 2003-06-25 | 三星电子株式会社 | Image research method and device not affected by light change |
CN101566994A (en) * | 2008-04-22 | 2009-10-28 | 王磊 | Image and video retrieval method |
CN101996212A (en) * | 2009-08-31 | 2011-03-30 | 英业达股份有限公司 | Image search system and method |
CN103198110A (en) * | 2013-03-28 | 2013-07-10 | 广州中国科学院软件应用技术研究所 | Method and system for rapid video data characteristic retrieval |
WO2013140037A1 (en) * | 2012-03-19 | 2013-09-26 | P2S Media Group Oy | Method and apparatus for reducing duplicates of multimedia data items in service system |
CN103824075A (en) * | 2014-02-18 | 2014-05-28 | 深圳天源迪科信息技术股份有限公司 | Image recognition system and method |
CN104504121A (en) * | 2014-12-29 | 2015-04-08 | 北京奇艺世纪科技有限公司 | Video retrieval method and device |
US20150193534A1 (en) * | 2010-04-13 | 2015-07-09 | At&T Intellectual Property I, L.P. | Systemand method for generating media thumbscapes |
-
2015
- 2015-10-22 CN CN201510694753.1A patent/CN106610987B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1340178A (en) * | 1999-08-17 | 2002-03-13 | 皇家菲利浦电子有限公司 | System and method for performing region-based image retrieval using color-based segmentation |
CN1426002A (en) * | 2001-11-27 | 2003-06-25 | 三星电子株式会社 | Image research method and device not affected by light change |
CN101566994A (en) * | 2008-04-22 | 2009-10-28 | 王磊 | Image and video retrieval method |
CN101996212A (en) * | 2009-08-31 | 2011-03-30 | 英业达股份有限公司 | Image search system and method |
US20150193534A1 (en) * | 2010-04-13 | 2015-07-09 | At&T Intellectual Property I, L.P. | Systemand method for generating media thumbscapes |
WO2013140037A1 (en) * | 2012-03-19 | 2013-09-26 | P2S Media Group Oy | Method and apparatus for reducing duplicates of multimedia data items in service system |
CN103198110A (en) * | 2013-03-28 | 2013-07-10 | 广州中国科学院软件应用技术研究所 | Method and system for rapid video data characteristic retrieval |
CN103824075A (en) * | 2014-02-18 | 2014-05-28 | 深圳天源迪科信息技术股份有限公司 | Image recognition system and method |
CN104504121A (en) * | 2014-12-29 | 2015-04-08 | 北京奇艺世纪科技有限公司 | Video retrieval method and device |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107609104B (en) * | 2017-09-12 | 2020-09-08 | 中广热点云科技有限公司 | Method and system for searching related video according to video image material |
CN107609104A (en) * | 2017-09-12 | 2018-01-19 | 中广热点云科技有限公司 | The method and system of associated video is searched according to video image material |
CN107481187A (en) * | 2017-09-29 | 2017-12-15 | 康佳集团股份有限公司 | A kind of method of video image processing, intelligent terminal and storage medium |
US11995556B2 (en) | 2018-05-18 | 2024-05-28 | Cambricon Technologies Corporation Limited | Video retrieval method, and method and apparatus for generating video retrieval mapping relationship |
CN110598048A (en) * | 2018-05-25 | 2019-12-20 | 北京中科寒武纪科技有限公司 | Video retrieval method and video retrieval mapping relation generation method and device |
CN111177190B (en) * | 2018-11-13 | 2023-05-30 | 杭州海康威视数字技术股份有限公司 | Data processing method, device, electronic equipment and readable storage medium |
CN111177190A (en) * | 2018-11-13 | 2020-05-19 | 杭州海康威视数字技术股份有限公司 | Data processing method and device, electronic equipment and readable storage medium |
CN109584317A (en) * | 2018-12-24 | 2019-04-05 | 天津天地伟业机器人技术有限公司 | Body color recognition methods based on HSV color space histogram |
CN111382306A (en) * | 2018-12-28 | 2020-07-07 | 杭州海康威视数字技术股份有限公司 | Method and device for inquiring video frame |
CN111382306B (en) * | 2018-12-28 | 2023-12-01 | 杭州海康威视数字技术股份有限公司 | Method and device for inquiring video frame |
CN111311697A (en) * | 2020-03-19 | 2020-06-19 | 北京搜狐新媒体信息技术有限公司 | Method for detecting color richness of picture and related device |
CN111311697B (en) * | 2020-03-19 | 2023-11-17 | 北京搜狐新媒体信息技术有限公司 | Picture color richness detection method and related device |
WO2022170866A1 (en) * | 2021-02-10 | 2022-08-18 | Oppo广东移动通信有限公司 | Data transmission method and apparatus, and storage medium |
CN114945019A (en) * | 2021-02-10 | 2022-08-26 | Oppo广东移动通信有限公司 | Data transmission method, device and storage medium |
CN114945019B (en) * | 2021-02-10 | 2023-11-21 | Oppo广东移动通信有限公司 | Data transmission method, device and storage medium |
CN113343033A (en) * | 2021-06-29 | 2021-09-03 | 平安普惠企业管理有限公司 | Video searching method and device, computer equipment and storage medium |
CN114741553A (en) * | 2022-03-31 | 2022-07-12 | 慧之安信息技术股份有限公司 | Image feature-based picture searching method |
Also Published As
Publication number | Publication date |
---|---|
CN106610987B (en) | 2020-02-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106610987A (en) | Video image retrieval method, device and system | |
US8238605B2 (en) | Digital video target moving object segmentation method and system | |
CN108206917B (en) | Image processing method and device, storage medium and electronic device | |
US11403762B2 (en) | Image processing method and related camera device | |
CN108764208A (en) | Image processing method and device, storage medium, electronic equipment | |
CN108875619B (en) | Video processing method and device, electronic equipment and computer readable storage medium | |
Gargi et al. | Evaluation of video sequence indexing and hierarchical video indexing | |
CN109119009A (en) | Screen defect inspection method, detection device, system and storage medium | |
CN105654057A (en) | Picture auditing system and picture auditing method based on picture contents | |
CN104636759B (en) | A kind of method and picture filter information recommendation system for obtaining picture and recommending filter information | |
CN108024107A (en) | Image processing method, device, electronic equipment and computer-readable recording medium | |
CN110830787B (en) | Method and device for detecting screen-patterned image | |
CN102035988A (en) | System and method for testing photographic effect of mobile phone camera | |
CN108806638B (en) | Image display method and device | |
CN108615030A (en) | A kind of title consistency detecting method, device and electronic equipment | |
CN106886789A (en) | A kind of image recognition sorter and method | |
Toylan et al. | A Real‐Time Apple Grading System Using Multicolor Space | |
EP1315104A2 (en) | Image retrieval method and apparatus independent of illumination change | |
CN108062554B (en) | Method and device for identifying color of vehicle annual inspection label | |
CN111310727A (en) | Object detection method and device, storage medium and electronic device | |
US20020172417A1 (en) | Image processing apparatus for and method of improving an image and an image display apparatus comprising the image processing apparatus | |
CN113850809A (en) | Method for detecting appearance defects of protective film of chip resistor | |
CN109903265A (en) | A kind of image change area detecting threshold value setting method, system and its electronic device | |
Roy et al. | New tampered features for scene and caption text classification in video frame | |
CN111294594A (en) | Security inspection method, device, system and storage medium |
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 |