CN110033488A - A kind of adaptive light source Orientation method and system based on compression information - Google Patents
A kind of adaptive light source Orientation method and system based on compression information Download PDFInfo
- Publication number
- CN110033488A CN110033488A CN201910291948.XA CN201910291948A CN110033488A CN 110033488 A CN110033488 A CN 110033488A CN 201910291948 A CN201910291948 A CN 201910291948A CN 110033488 A CN110033488 A CN 110033488A
- Authority
- CN
- China
- Prior art keywords
- illumination
- image
- frame
- block
- short
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- 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
- G06T2207/10024—Color 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
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
The present invention proposes a kind of adaptive light source Orientation method and system based on compression information.Whether the method for the present invention belongs to the scene that illumination analysis has been established first with the determining current identification video of object of reference matching method, starts the analytical model of different direction of illuminations by system then in conjunction with matching result for the system for supporting the analysis of multi-method direction of illumination;Then, for different direction of illumination analysis methods, the direction of illumination analyzed area of present image is quickly marked off using information such as the block division of code stream, prediction modes, to promote analysis speed.
Description
Technical field
The present invention relates to field of image processing more particularly to a kind of adaptive light source Orientation sides based on compression information
Method and system.
Background technique
In practical application, illumination condition can generate tremendous influence, such as backlight, polarization light source to Face datection precision.
At this point, illumination compensation can be carried out in advance if light source position can be obtained, accuracy of identification is promoted.And currently used light source direction
Analytic approach has the Classification and Identification method of the database based on different direction of illuminations, the light based on image informations such as illumination model, inverted images
Source direction detection method.Wherein first kind method accuracy of identification is high, but has scene limitation, and the second class method difficult point is
Actual image information is complicated, so that algorithm will spend a large amount of calculation amount extracting inverted image from image, be used for light source direction point
On the step of region of analysis.On the other hand, when being scanned for video, if ignoring the division and prediction letter that video code flow carries
Breath can be used as the auxiliary information of light source analysis, it will the calculation amount of full frame search frame by frame is caused to waste.
Summary of the invention
The purpose of the embodiment of the present invention is to propose a kind of adaptive light source Orientation method based on compression information, purport
Solving the problems, such as that prior art light source Orientation method is computationally intensive.
The embodiments of the present invention are implemented as follows, a kind of adaptive light source Orientation method based on compression information, institute
The method of stating includes:
Determine whether current video to be detected belongs to the scene of preset illumination analysis using object of reference matching method;
Start corresponding direction of illumination analytical model according to judgement result.
Further, described to determine whether current video to be detected belongs to preset illumination and analyze using object of reference matching method
Scene include:
According to the rule of one image of a scene, the corresponding mould of each scene is selected from preset Classification and Identification template
Formula screening figure, while dividing the recognition template of same scene into a subclass;
Using current video first frame image to be detected and mode screening figure, object of reference matching is carried out.
The another object of the embodiment of the present invention is to propose adaptive light source Orientation of the kind based on compression information
System, the system comprises:
Scene decision maker is analyzed in illumination, pre- for determining whether current video to be detected belongs to using object of reference matching method
If illumination analysis scene;
Direction of illumination analytical equipment, for starting corresponding direction of illumination analysis module according to judgement result.
Further, the illumination analysis scene decision maker includes:
Mode screens figure selecting module, for the rule according to one image of a scene, from preset Classification and Identification mould
The corresponding mode screening figure of each scene is selected in plate, while dividing the recognition template of same scene into a subclass;
Object of reference coalignment, for carrying out reference using current video first frame image to be detected and mode screening figure
Object matching.
Beneficial effects of the present invention
The present invention proposes a kind of adaptive light source Orientation method and system based on compression information.The method of the present invention needle
To the system for supporting the analysis of multi-method direction of illumination, determine whether current identification video belongs to first with object of reference matching method
The scene for establishing illumination analysis, starts the analytical model of different direction of illuminations then in conjunction with matching result by system;Then, needle
To different direction of illumination analysis methods, the light of present image is quickly marked off using information such as the block division of code stream, prediction modes
According to Orientation region, to promote analysis speed.
Detailed description of the invention
Fig. 1 is a kind of adaptive light source Orientation method flow diagram based on compression information of the preferred embodiment of the present invention;
Fig. 2 is that a kind of utilization object of reference matching method provided in an embodiment of the present invention determines whether current video to be detected belongs to
The method flow diagram of the scene of preset illumination analysis;
Fig. 3 be it is provided in an embodiment of the present invention it is a kind of using current video first frame image to be detected and mode screen figure into
The matched method flow diagram of row object of reference;
Fig. 4 is that another kind provided in an embodiment of the present invention utilizes current video first frame image to be detected and mode screening figure
Carry out the matched method flow diagram of object of reference;
Fig. 5 is a kind of adaptive light source Orientation system construction drawing based on compression information of the preferred embodiment of the present invention;
Fig. 6 is that scene decision maker structure chart is analyzed in a kind of illumination provided in an embodiment of the present invention;
Fig. 7 is a kind of object of reference coalignment structure chart provided in an embodiment of the present invention;
Fig. 8 is another object of reference coalignment structure chart provided in an embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, right below in conjunction with drawings and examples
The present invention is further elaborated, and for ease of description, only parts related to embodiments of the present invention are shown.It should manage
Solution, the specific embodiment that this place is described are used only for explaining the present invention, be not intended to limit the invention.
The present invention proposes a kind of adaptive light source Orientation method and system based on compression information.The method of the present invention needle
To the system for supporting the analysis of multi-method direction of illumination, determine whether current identification video belongs to first with object of reference matching method
The scene for establishing illumination analysis, starts the analytical model of different direction of illuminations then in conjunction with matching result by system;Then, needle
To different direction of illumination analysis methods, the light of present image is quickly marked off using information such as the block division of code stream, prediction modes
According to Orientation region, to promote analysis speed.
Embodiment one
Fig. 1 is a kind of adaptive light source Orientation method flow diagram based on compression information of the preferred embodiment of the present invention;
Present invention method is applied to support the encoder of multi-method direction of illumination analysis;
Present invention method the following steps are included:
S1 determines whether current video to be detected belongs to the scene of preset illumination analysis using object of reference matching method;
If object of reference successful match, then it represents that the scene for belonging to preset illumination analysis, if object of reference matching is unsuccessful,
Indicate the scene for being not belonging to preset illumination analysis;
Fig. 2 is that a kind of utilization object of reference matching method provided in an embodiment of the present invention determines whether current video to be detected belongs to
The method of the scene of preset illumination analysis, specifically includes are as follows:
S101: according to the rule of one image of a scene, each scene pair is selected from preset Classification and Identification template
The mode screening figure answered, while dividing the recognition template of same scene into a subclass.
The Classification and Identification template is image recognition mould that the pattern-recognition training stage obtains, being classified according to direction of illumination
Plate is stored in advance in the encoder, and each image corresponds to the screening figure of a mode in Classification and Identification template.
S102: using current video first frame image to be detected and mode screening figure, object of reference matching is carried out.
Fig. 3 be it is provided in an embodiment of the present invention it is a kind of using current video first frame image to be detected and mode screen figure into
The matched method of row object of reference:
Step A1: current video first frame image to be detected is schemed into respectively one two frame of composition with the screening of each mode respectively
The short-sighted frequency group of image, wherein enabling current video first frame image to be detected is the second frame image of each short-sighted frequency.
Step A2: each short-sighted frequency is encoded.
Step A3: the block of positioning references divides position first, if having block with object of reference institute in current short-sighted the second frame of frequency
The block at place is reference block, then currently the corresponding mode screening figure of short-sighted frequency first frame is that the mode of successful match screens figure for setting,
Subsequently into step S2, A4 is otherwise entered step.
Step A4: if all short-sighted frequencies judge to terminate, S2 is entered step;Otherwise, return step A3 is to next short
Video is determined.
Fig. 4 is that another kind provided in an embodiment of the present invention utilizes current video first frame image to be detected and mode screening figure
Carry out the matched method of object of reference:
Step B1: current video first frame image to be detected is formed into short-sighted frequency with each mode screening figure respectively together
Group, wherein enabling current video first frame image to be detected is the first frame image of short-sighted frequency.
Step B2: the short-sighted frequency is encoded;
Wherein, during encoding to the short-sighted frequency, to guarantee that first frame image exists in reference frame span
It screens in range (i.e. reference frame includes first frame image).
Step B3: setting present frame is short-sighted the second frame of frequency.
Step B4: if the prediction mode of object of reference corresponding blocks is inter-frame forecast mode in present frame, and object of reference is corresponding
Block using the block in short-sighted frequency first frame as reference block, then the mould that the corresponding mode screening figure of present frame is successful match is set
Formula screening figure, subsequently into step S2;Otherwise, B5 is entered step;
Step B5: if the judgement of all frames terminates, S2 is entered step;Otherwise, the next frame that present frame is arranged is present frame,
Then return step B4.
Described to be schemed using current video first frame image to be detected and mode screening, carrying out object of reference matching also can be used this skill
Other methods well known to art field.
S2 starts corresponding direction of illumination analytical model according to judgement result;
Specifically: if object of reference successful match (scene for belonging to preset illumination analysis) enters the first direction of illumination
Analytical model;If object of reference matching is unsuccessful (scene for being not belonging to preset illumination analysis), enter the second direction of illumination point
Analysis mode;
First direction of illumination analytical model: subclass belonging to the mode screening figure using successful match, to current to be detected
Video first frame image carries out Classification and Identification template matching, determines specific direction of illumination;Then surplus to current video to be detected
Remaining frame image carries out direction of illumination detection.
Second direction of illumination analytical model: first according to direction of illumination figures different in preset light source direction analytic approach encapsulation package
The characteristics of picture, divides the direction of illumination analyzed area of present image, then to institute in conjunction with the code stream information of current video to be detected
It states direction of illumination analyzed area and carries out direction of illumination detection, determine specific direction of illumination;Then surplus to current video to be detected
Remaining frame image carries out direction of illumination detection;
The code stream information includes block division, prediction mode etc..
The light source direction analytic approach encapsulation package is encoder in advance setting encapsulation for different images information illumination side
To detection method, Classification and Identification method including the database based on different direction of illuminations, based on images such as illumination model, inverted images
The direction of illumination detection method of information;It is the prior art.
Specifically in the application, if light source direction analytic approach encapsulation package is the direction of illumination determination method based on inverted image,
By feature of image known to encapsulation package: inverted image compares the spy that simple with texture, brightness is low, chrominance information is deficient with object itself
Point;
The then direction of illumination analyzed area for dividing present image, then carries out light to the direction of illumination analyzed area
According to angle detecting, specific direction of illumination is determined specifically:
Step C1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step C2: followed by the luminance mean value of block, the block that luminance mean value is more than luminance threshold is further deleted;
Step C3: and then recycling inverted image is further to delete the block of chromatic component non-zero without the colorful one feature,
Invalid data is completed to delete;
Step C4: finally, to the image-region of current residual, disclosed method extracts inverted image in the industry for utilization, carries out illumination
Orientation determines specific direction of illumination.
Specifically in the application, if light source direction analytic approach encapsulation package is that the direction of illumination based on illumination model tests and analyzes,
The feature of image known to encapsulation package: the brightness of image changes with the variation of angle between body surface direction and direction of illumination,
It is (thus specific it may determine that direction of illumination out, illumination that any direction of illumination can produce a specific image brightness distribution
It needs to filter out to illuminant direction estimation effective body surface when analysis);Either concave surface or convex surface, in video code flow
All have the characteristics that texture is simple, to there is illumination, then brightness will not be completely black;
The then direction of illumination analyzed area for dividing present image, then carries out light to the direction of illumination analyzed area
According to angle detecting, specific direction of illumination is determined specifically:
Step D1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step D2: followed by the luminance mean value of block, the block that luminance mean value is less than luminance threshold is further deleted;
Step D3: followed by connectivity is utilized, isolated largest block is further deleted, and wait the block of connection as one
The male and fomale(M&F) detection zone of choosing is completed invalid data and is deleted;
Step D4: finally, to each candidate male and fomale(M&F) detection zone in the image-region of current residual, industry is utilized
Interior disclosed Pentland method or improved Pentland method calculate angle and obtain direction of illumination:
Further, if wanting to obtain more accurate as a result, described " utilize disclosed Pentland method or improvement in the industry
Pentland method calculate angle and obtain direction of illumination " can be replaced " using Pentland method extracted valid data, then
Direction of illumination is calculated again ".
It is described that direction of illumination detection is carried out to current video residue frame image to be detected specifically: to keep current view to be detected
The direction of illumination of frequency is analyzed as a result, until I frame (first frame default code mould in Video coding occurs again in current video to be detected
Formula is exactly I frame), all frames before the current video to be detected I frame are deleted at this time, return S1;If current video to be detected is not
Occur I frame again, then keeps the direction of illumination analysis result of current video to be detected until terminating.
Embodiment two
Fig. 5 is a kind of adaptive light source Orientation system construction drawing based on compression information of the preferred embodiment of the present invention;
The system comprises:
Scene decision maker is analyzed in illumination, pre- for determining whether current video to be detected belongs to using object of reference matching method
If illumination analysis scene;
Specifically, if object of reference successful match, then it represents that belong to the scene of preset illumination analysis, if object of reference matching is not
Success, then it represents that be not belonging to the scene of preset illumination analysis;
Direction of illumination analytical equipment, for starting corresponding direction of illumination analysis module according to judgement result;The illumination
Orientation module includes the first direction of illumination analysis module and the second direction of illumination analysis module
Specifically, if object of reference successful match (scene for belonging to preset illumination analysis), starts the first direction of illumination
Analysis module;If object of reference matching is unsuccessful (scene for being not belonging to preset illumination analysis), start the second direction of illumination point
Analyse module;
Further, Fig. 6 is that scene decision maker structure chart is analyzed in illumination provided in an embodiment of the present invention;The illumination point
Analysing scene decision maker includes:
Mode screens figure selecting module, for the rule according to one image of a scene, from preset Classification and Identification mould
The corresponding mode screening figure of each scene is selected in plate, while dividing the recognition template of same scene into a subclass.
The Classification and Identification template is image recognition mould that the pattern-recognition training stage obtains, being classified according to direction of illumination
Plate is stored in advance in the encoder, and each image corresponds to the screening figure of a mode in Classification and Identification template.
Object of reference coalignment, for carrying out reference using current video first frame image to be detected and mode screening figure
Object matching.
Further, Fig. 7 is object of reference coalignment structure chart provided in an embodiment of the present invention;
First short-sighted frequency sets up module, schemes for screening current video first frame image to be detected with each mode respectively
The respectively short-sighted frequency group of one two field pictures of composition, wherein enabling current video first frame image to be detected is the of each short-sighted frequency
Two frame images.
First short video encoding module, for being encoded to each short-sighted frequency.
Object of reference positioning and the first reference block determination module, the block for positioning references first divides position, if currently
There is block using block locating for object of reference as reference block in short-sighted the second frame of frequency, then enters the first successful match mode screening figure setting mould
Block;Otherwise enter the first judging treatmenting module;
First successful match mode screens figure setup module, for the corresponding mode screening of current short-sighted frequency first frame to be arranged
Figure is the mode screening figure of successful match, subsequently into direction of illumination analytical equipment
First judging treatmenting module, if entering direction of illumination analysis dress for judging that all short-sighted frequencies judge to terminate
It sets;Otherwise, referring back to object positioning and the first reference block determination module.
Further, Fig. 8 is another object of reference coalignment structure chart provided in an embodiment of the present invention.
Second short-sighted frequency sets up module, schemes for screening current video first frame image to be detected with each mode respectively
Short-sighted frequency group is formed together, wherein enabling current video first frame image to be detected is the first frame image of short-sighted frequency;
Second short video encoding module, for being encoded to the short-sighted frequency;
During encoding to the short-sighted frequency, to guarantee first frame image in screening model in reference frame span
In enclosing (i.e. reference frame includes first frame image);
Second short-sighted the second frame of frequency setup module is short-sighted the second frame of frequency for present frame to be arranged;
Object of reference prediction mode and the second reference block determination module, if for judging the pre- of object of reference corresponding blocks in present frame
Survey mode is inter-frame forecast mode, and object of reference corresponding blocks using the block in short-sighted frequency first frame as reference block, then enter
Second successful match mode screens figure setup module;Otherwise, into the second judging treatmenting module;
Second successful match mode screen figure setup module, for be arranged the corresponding mode screening figure of present frame be match at
The mode of function screens figure, subsequently into direction of illumination analytical equipment;
Second judging treatmenting module, if entering direction of illumination analytical equipment for judging that all frame judgements terminate;It is no
Then, the next frame that present frame is arranged is present frame, then referring back to object prediction mode and the second reference block determination module;
Further, the first direction of illumination analysis module, for belonging to the mode screening figure using successful match
Subclass carries out Classification and Identification template matching to current video first frame image to be detected, determines specific direction of illumination;Then right
Current video residue frame image to be detected carries out direction of illumination detection.
Second direction of illumination analysis analysis module, for first according to illumination different in preset light source direction analytic approach encapsulation package
The characteristics of directional image, divides the direction of illumination analyzed area of present image, connects in conjunction with the code stream information of current video to be detected
To the direction of illumination analyzed area carry out direction of illumination detection, determine specific direction of illumination;Then to current to be detected
Video residue frame image carries out direction of illumination detection;
The code stream information includes block division, prediction mode etc..
The light source direction analytic approach encapsulation package is encoder in advance setting encapsulation for different images information illumination side
To detection method, Classification and Identification method including the database based on different direction of illuminations, based on images such as illumination model, inverted images
The direction of illumination detection method of information;It is the prior art.
Specifically in the application, if light source direction analytic approach encapsulation package is the direction of illumination determination method based on inverted image,
By feature of image known to encapsulation package: inverted image compares the spy that simple with texture, brightness is low, chrominance information is deficient with object itself
Point;
The then direction of illumination analyzed area for dividing present image, then carries out light to the direction of illumination analyzed area
According to angle detecting, specific direction of illumination is determined specifically:
Step C1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step C2: followed by the luminance mean value of block, the block that luminance mean value is more than luminance threshold is further deleted;
Step C3: and then recycling inverted image is further to delete the block of chromatic component non-zero without the colorful one feature,
Invalid data is completed to delete;
Step C4: finally, to the image-region of current residual, disclosed method extracts inverted image in the industry for utilization, carries out illumination
Orientation determines specific direction of illumination.
Specifically in the application, if light source direction analytic approach encapsulation package is that the direction of illumination based on illumination model tests and analyzes,
The feature of image known to encapsulation package: the brightness of image changes with the variation of angle between body surface direction and direction of illumination,
It is (thus specific it may determine that direction of illumination out, illumination that any direction of illumination can produce a specific image brightness distribution
It needs to filter out to illuminant direction estimation effective body surface when analysis);Either concave surface or convex surface, in video code flow
All have the characteristics that texture is simple, to there is illumination, then brightness will not be completely black;
The then direction of illumination analyzed area for dividing present image, then carries out light to the direction of illumination analyzed area
According to angle detecting, specific direction of illumination is determined specifically:
Step D1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step D2: followed by the luminance mean value of block, the block that luminance mean value is less than luminance threshold is further deleted;
Step D3: followed by connectivity is utilized, isolated largest block is further deleted, and wait the block of connection as one
The male and fomale(M&F) detection zone of choosing is completed invalid data and is deleted;
Step D4: finally, to each candidate male and fomale(M&F) detection zone in the image-region of current residual, industry is utilized
Interior disclosed Pentland method or improved Pentland method calculate angle and obtain direction of illumination;
Further, if wanting to obtain more accurate as a result, described " utilize disclosed Pentland method or improvement in the industry
Pentland method calculate angle and obtain direction of illumination " can be replaced " using Pentland method extracted valid data, then
Direction of illumination is calculated again ".
It is described that direction of illumination detection is carried out to current video residue frame image to be detected specifically: to keep current view to be detected
The direction of illumination of frequency is analyzed as a result, until I frame (first frame default code mould in Video coding occurs again in current video to be detected
Formula is exactly I frame), all frames before the current video to be detected I frame are deleted at this time, return illumination analysis scene decision maker;
If I frame no longer occurs in current video to be detected, keep the direction of illumination analysis result of current video to be detected until terminating.
It will be understood by those skilled in the art that implement the method for the above embodiments be can
With what is completed by program instruction related hardware, the program be can be stored in a computer readable storage medium,
The storage medium can be ROM, RAM, disk, CD etc..
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (17)
1. a kind of adaptive light source Orientation method based on compression information, which is characterized in that the described method includes:
Determine whether current video to be detected belongs to the scene of preset illumination analysis using object of reference matching method;
Start corresponding direction of illumination analytical model according to judgement result.
2. the adaptive light source Orientation method as described in claim 1 based on compression information, which is characterized in that the benefit
Determine whether current video to be detected belongs to the scene that preset illumination is analyzed and include: with object of reference matching method
According to the rule of one image of a scene, the corresponding mode sieve of each scene is selected from preset Classification and Identification template
Choosing figure, while dividing the recognition template of same scene into a subclass;
Using current video first frame image to be detected and mode screening figure, object of reference matching is carried out.
3. the adaptive light source Orientation method as claimed in claim 2 based on compression information, which is characterized in that the benefit
With current video first frame image to be detected and mode screening figure, carry out object of reference matching comprising steps of
Step A1: current video first frame image to be detected is schemed into respectively one two field pictures of composition with the screening of each mode respectively
Short-sighted frequency group, wherein enabling current video first frame image to be detected is the second frame image of each short-sighted frequency;
Step A2: each short-sighted frequency is encoded;
Step A3: the block of positioning references divides position first, if having block locating for object of reference in current short-sighted the second frame of frequency
Block is reference block, then currently the corresponding mode screening figure of short-sighted frequency first frame is that the mode screening of successful match is schemed for setting, then
Into according to the corresponding direction of illumination analytical model of result starting is determined, A4 is otherwise entered step;
Step A4: if all short-sighted frequencies judge to terminate, enter and corresponding direction of illumination analysis mould is started according to judgement result
Formula;Otherwise, return step A3 determines next short-sighted frequency.
4. the adaptive light source Orientation method as claimed in claim 2 based on compression information, which is characterized in that the benefit
With current video first frame image to be detected and mode screening figure, carry out object of reference matching comprising steps of
Step B1: forming short-sighted frequency group with each mode screening figure respectively for current video first frame image to be detected together,
The middle first frame image for enabling current video first frame image to be detected as short-sighted frequency;
Step B2: the short-sighted frequency is encoded;Wherein, during being encoded to the short-sighted frequency, in reference frame
To guarantee first frame image in screening range in span;
Step B3: setting present frame is short-sighted the second frame of frequency;
Step B4: if the prediction mode of object of reference corresponding blocks is inter-frame forecast mode in present frame, and object of reference corresponding blocks
Using the block in short-sighted frequency first frame as reference block, then the mode sieve that the corresponding mode screening figure of present frame is successful match is set
Choosing figure starts corresponding direction of illumination analytical model subsequently into according to judgement result;Otherwise, B5 is entered step;
Step B5: if the judgement of all frames terminates, enter and corresponding direction of illumination analytical model is started according to judgement result;It is no
Then, the next frame that present frame is arranged is present frame, then return step B4.
5. the adaptive light source Orientation method based on compression information as described in claim 1-4 any one, feature
It is, it is described that corresponding direction of illumination analytical model is started according to judgement result specifically: if object of reference successful match, enters
First direction of illumination analytical model;If object of reference matching is unsuccessful, enter the second direction of illumination analytical model;
First direction of illumination analytical model: subclass belonging to the mode screening figure using successful match, to current video to be detected
First frame image carries out Classification and Identification template matching, determines specific direction of illumination;Then to current video residue frame to be detected
Image carries out direction of illumination detection;
Second direction of illumination analytical model: first according to direction of illumination images different in preset light source direction analytic approach encapsulation package
Feature divides the direction of illumination analyzed area of present image, then to the light in conjunction with the code stream information of current video to be detected
Direction of illumination detection is carried out according to Orientation region, determines specific direction of illumination;Then to current video residue frame to be detected
Image carries out direction of illumination detection.
6. the adaptive light source Orientation method as claimed in claim 5 based on compression information, which is characterized in that if light source
Orientation method encapsulation package is the direction of illumination determination method based on inverted image,
The then direction of illumination analyzed area for dividing present image, then carries out illumination side to the direction of illumination analyzed area
To detection, specific direction of illumination is determined specifically:
Step C1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step C2: followed by the luminance mean value of block, the block that luminance mean value is more than luminance threshold is further deleted;
Step C3: and then recycling inverted image is further to delete the block of chromatic component non-zero without the colorful one feature, complete
Invalid data is deleted;
Step C4: finally, to the image-region of current residual, disclosed method extracts inverted image in the industry for utilization, carries out direction of illumination
Analysis, determines specific direction of illumination.
7. the adaptive light source Orientation method as claimed in claim 5 based on compression information, which is characterized in that if light source
Orientation method encapsulation package is that the direction of illumination based on illumination model tests and analyzes,
The then direction of illumination analyzed area for dividing present image, then carries out illumination side to the direction of illumination analyzed area
To detection, specific direction of illumination is determined specifically:
Step D1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step D2: followed by the luminance mean value of block, the block that luminance mean value is less than luminance threshold is further deleted;
Step D3: followed by connectivity is utilized, further deleting isolated largest block, and candidate using the block of connection as one
Male and fomale(M&F) detection zone is completed invalid data and is deleted;
Step D4: finally, to each candidate male and fomale(M&F) detection zone in the image-region of current residual, utilization is public in the industry
The Pentland method opened or improved Pentland method calculate angle and obtain direction of illumination.
8. the adaptive light source Orientation method as claimed in claim 7 based on compression information, which is characterized in that
It is described " to calculate angle using disclosed Pentland method or improved Pentland method in the industry and obtain illumination side
To " replace with " using Pentland method extracted valid data, then calculating direction of illumination again ".
9. a kind of adaptive light source Orientation system based on compression information, which is characterized in that the system comprises:
Scene decision maker is analyzed in illumination, preset for determining whether current video to be detected belongs to using object of reference matching method
The scene of illumination analysis;
Direction of illumination analytical equipment, for starting corresponding direction of illumination analysis module according to judgement result.
10. a kind of adaptive light source Orientation system based on compression information as claimed in claim 9, which is characterized in that
Scene decision maker is analyzed in the illumination
Mode screens figure selecting module, for the rule according to one image of a scene, from preset Classification and Identification template
The corresponding mode screening figure of each scene is selected, while dividing the recognition template of same scene into a subclass;
Object of reference coalignment, for carrying out object of reference using current video first frame image to be detected and mode screening figure
Match.
11. a kind of adaptive light source Orientation system based on compression information as claimed in claim 10, which is characterized in that
The object of reference coalignment includes:
First short-sighted frequency sets up module, for current video first frame image to be detected is respective with each mode screening figure respectively
The short-sighted frequency group an of two field pictures is formed, wherein enabling current video first frame image to be detected is the second frame of each short-sighted frequency
Image;
First short video encoding module, for being encoded to each short-sighted frequency;
Object of reference positioning and the first reference block determination module, the block for positioning references first divide position, if current short-sighted
Frequently there is block using block locating for object of reference as reference block in the second frame, then enter the first successful match mode and screen figure setup module;
Otherwise enter the first judging treatmenting module;
First successful match mode screens figure setup module, is for the corresponding mode screening figure of current short-sighted frequency first frame to be arranged
The mode of successful match screens figure, subsequently into direction of illumination analytical equipment;
First judging treatmenting module, if entering direction of illumination analytical equipment for judging that all short-sighted frequencies judge to terminate;It is no
Then, referring back to object positioning and the first reference block determination module.
12. a kind of adaptive light source Orientation system based on compression information as claimed in claim 10, which is characterized in that
The object of reference coalignment includes:
Second short-sighted frequency sets up module, for current video first frame image to be detected to be screened figure together with each mode respectively
Short-sighted frequency group is formed, wherein enabling current video first frame image to be detected is the first frame image of short-sighted frequency;
Second short video encoding module, for being encoded to the short-sighted frequency;In the process encoded to the short-sighted frequency
In, to guarantee first frame image in screening range in reference frame span;
Second short-sighted the second frame of frequency setup module is short-sighted the second frame of frequency for present frame to be arranged;
Object of reference prediction mode and the second reference block determination module, if for judging the prediction mould of object of reference corresponding blocks in present frame
Formula is inter-frame forecast mode, and object of reference corresponding blocks using the block in short-sighted frequency first frame as reference block, then enter second
Successful match mode screens figure setup module;Otherwise, into the second judging treatmenting module;
Second successful match mode screens figure setup module, is successful match for the corresponding mode screening figure of present frame to be arranged
Mode screening figure, subsequently into direction of illumination analytical equipment;
Second judging treatmenting module, if entering direction of illumination analytical equipment for judging that all frame judgements terminate;Otherwise, if
The next frame for setting present frame is present frame, then referring back to object prediction mode and the second reference block determination module.
13. a kind of adaptive light source Orientation system based on compression information as described in claim 9-12 any one,
It is characterized in that,
The first direction of illumination analysis module, for using successful match mode screening figure belonging to subclass, to currently to
It detects video first frame image and carries out Classification and Identification template matching, determine specific direction of illumination;Then to current view to be detected
Frequency residue frame image carries out direction of illumination detection;
The second direction of illumination analysis analysis module, for first according to illumination different in preset light source direction analytic approach encapsulation package
The characteristics of directional image, divides the direction of illumination analyzed area of present image, connects in conjunction with the code stream information of current video to be detected
To the direction of illumination analyzed area carry out direction of illumination detection, determine specific direction of illumination;Then to current to be detected
Video residue frame image carries out direction of illumination detection.
14. a kind of adaptive light source Orientation system based on compression information as claimed in claim 13, which is characterized in that
The light source direction analytic approach encapsulation package is encoder in advance setting encapsulation for different images information direction of illumination
Detection method, Classification and Identification method including the database based on different direction of illuminations, based on image informations such as illumination model, inverted images
Direction of illumination detection method.
15. a kind of adaptive light source Orientation system based on compression information as claimed in claim 14, which is characterized in that
If light source direction analytic approach encapsulation package is the direction of illumination determination method based on inverted image,
The then direction of illumination analyzed area for dividing present image, then carries out illumination side to the direction of illumination analyzed area
To detection, specific direction of illumination is determined specifically:
Step C1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step C2: followed by the luminance mean value of block, the block that luminance mean value is more than luminance threshold is further deleted;
Step C3: and then recycling inverted image is further to delete the block of chromatic component non-zero without the colorful one feature, complete
Invalid data is deleted;
Step C4: finally, to the image-region of current residual, disclosed method extracts inverted image in the industry for utilization, carries out direction of illumination
Analysis, determines specific direction of illumination.
16. a kind of adaptive light source Orientation system based on compression information as claimed in claim 14, which is characterized in that
If light source direction analytic approach encapsulation package is that the direction of illumination based on illumination model tests and analyzes,
The then direction of illumination analyzed area for dividing present image, then carries out illumination side to the direction of illumination analyzed area
To detection, specific direction of illumination is determined specifically:
Step D1: the image-region that block in current video first frame image to be detected divides non-largest block is deleted first;
Step D2: followed by the luminance mean value of block, the block that luminance mean value is less than luminance threshold is further deleted;
Step D3: followed by connectivity is utilized, further deleting isolated largest block, and candidate using the block of connection as one
Male and fomale(M&F) detection zone is completed invalid data and is deleted;
Step D4: finally, to each candidate male and fomale(M&F) detection zone in the image-region of current residual, utilization is public in the industry
The Pentland method opened or improved Pentland method calculate angle and obtain direction of illumination.
17. a kind of adaptive light source Orientation system based on compression information as claimed in claim 16, which is characterized in that
It is described " to calculate angle using disclosed Pentland method or improved Pentland method in the industry and obtain illumination side
To " it can be replaced " using Pentland method extracted valid data, then calculating direction of illumination again ".
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910291948.XA CN110033488B (en) | 2019-04-09 | 2019-04-09 | Self-adaptive light source direction analysis method and system based on compressed information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910291948.XA CN110033488B (en) | 2019-04-09 | 2019-04-09 | Self-adaptive light source direction analysis method and system based on compressed information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110033488A true CN110033488A (en) | 2019-07-19 |
CN110033488B CN110033488B (en) | 2023-09-15 |
Family
ID=67238094
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910291948.XA Active CN110033488B (en) | 2019-04-09 | 2019-04-09 | Self-adaptive light source direction analysis method and system based on compressed information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110033488B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116071807A (en) * | 2023-03-06 | 2023-05-05 | 深圳市网联天下科技有限公司 | Campus card intelligent early warning method and system based on video monitoring |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102761737A (en) * | 2011-04-28 | 2012-10-31 | 财团法人工业技术研究院 | Systems and methods for processing shadows in compressed video images |
US20160366415A1 (en) * | 2015-06-09 | 2016-12-15 | Qualcomm Incorporated | Systems and methods of determining illumination compensation parameters for video coding |
CN108765463A (en) * | 2018-05-30 | 2018-11-06 | 河海大学常州校区 | A kind of moving target detecting method calmodulin binding domain CaM extraction and improve textural characteristics |
CN109446967A (en) * | 2018-10-22 | 2019-03-08 | 深圳市梦网百科信息技术有限公司 | A kind of method for detecting human face and system based on compression information |
-
2019
- 2019-04-09 CN CN201910291948.XA patent/CN110033488B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102761737A (en) * | 2011-04-28 | 2012-10-31 | 财团法人工业技术研究院 | Systems and methods for processing shadows in compressed video images |
US20160366415A1 (en) * | 2015-06-09 | 2016-12-15 | Qualcomm Incorporated | Systems and methods of determining illumination compensation parameters for video coding |
CN108765463A (en) * | 2018-05-30 | 2018-11-06 | 河海大学常州校区 | A kind of moving target detecting method calmodulin binding domain CaM extraction and improve textural characteristics |
CN109446967A (en) * | 2018-10-22 | 2019-03-08 | 深圳市梦网百科信息技术有限公司 | A kind of method for detecting human face and system based on compression information |
Non-Patent Citations (1)
Title |
---|
张屹凌: "基于光度立体算法的图像建模系统的设计及实现", 《中国优秀博硕士学位论文全文数据库(硕士)》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116071807A (en) * | 2023-03-06 | 2023-05-05 | 深圳市网联天下科技有限公司 | Campus card intelligent early warning method and system based on video monitoring |
Also Published As
Publication number | Publication date |
---|---|
CN110033488B (en) | 2023-09-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR100997064B1 (en) | Multi-mode region-of-interest video object segmentation | |
KR100983204B1 (en) | Inter-mode region-of-interest video object segmentation | |
CN109063694B (en) | Video target detection and identification method | |
CN105608455B (en) | A kind of license plate sloped correcting method and device | |
US20060039605A1 (en) | Method and apparatus for recognition of character string in scene image | |
US8155396B2 (en) | Method, apparatus, and program for detecting faces | |
JP2009526495A (en) | Region of interest image object segmentation between modes | |
JP2006505853A (en) | Method for generating quality-oriented importance map for evaluating image or video quality | |
KR20040028911A (en) | Method and apparatus for motion estimation between video frames | |
KR20100073189A (en) | Apparatus and method for detecting face image | |
CN102609724B (en) | Method for prompting ambient environment information by using two cameras | |
CN101029823B (en) | Method for tracking vehicle based on state and classification | |
CN110232323A (en) | A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device | |
JP3490910B2 (en) | Face area detection device | |
JPH06231254A (en) | High-speed recognition and retrieval method for moving image | |
US7715632B2 (en) | Apparatus and method for recognizing an image | |
CN103020949A (en) | Facial image detection method | |
CN110033488A (en) | A kind of adaptive light source Orientation method and system based on compression information | |
CN101827224A (en) | Detection method of anchor shot in news video | |
JP2978039B2 (en) | Video cut screen detection method | |
JPH10222678A (en) | Device for detecting object and method therefor | |
CN108711139B (en) | One kind being based on defogging AI image analysis system and quick response access control method | |
Love et al. | An empirical study of block matching techniques for the detection of moving objects | |
CN109784176B (en) | Vehicle-mounted thermal imaging pedestrian detection Rois extraction method and device | |
CN114463242A (en) | Image detection method, device, storage medium and device |
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 | ||
CB02 | Change of applicant information | ||
CB02 | Change of applicant information |
Address after: 518000 Guangdong city of Shenzhen province Nanshan District Guangdong streets high in the four Longtaili Technology Building Room 325 No. 30 Applicant after: Shenzhen mengwang video Co.,Ltd. Address before: 518000 Guangdong city of Shenzhen province Nanshan District Guangdong streets high in the four Longtaili Technology Building Room 325 No. 30 Applicant before: SHENZHEN MONTNETS ENCYCLOPEDIA INFORMATION TECHNOLOGY Co.,Ltd. |
|
GR01 | Patent grant | ||
GR01 | Patent grant |