CN105957111B - The color tone consistency bearing calibration of sequence remote sensing image and system - Google Patents
The color tone consistency bearing calibration of sequence remote sensing image and system Download PDFInfo
- Publication number
- CN105957111B CN105957111B CN201610269941.4A CN201610269941A CN105957111B CN 105957111 B CN105957111 B CN 105957111B CN 201610269941 A CN201610269941 A CN 201610269941A CN 105957111 B CN105957111 B CN 105957111B
- Authority
- CN
- China
- Prior art keywords
- image
- remote sensing
- sequence
- color
- node
- 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.)
- Active
Links
Landscapes
- Facsimile Image Signal Circuits (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A kind of color tone consistency bearing calibration of sequence remote sensing image and system include that colored sequential images are gone to L α β color spaces from RGB color, carry out tone and transmit pretreatment, the global gain compensation of luminance channel；The Histogram Mapping of Color Channel separately adjusts image greyscale in Color Channel α and β；All images are gone into RGB color from L α β color spaces, exports and preserves handling result.The present invention utilizes the tone correspondence of overlapping region between adjacent image, can effectively inhibit or eliminate existing brightness and color distortion between image；The problem of image fault brought due to Inter-channel Correlation can largely be reduced, the mode of the maximum coherent image subset of search is taken to choose reference color automatically, avoid artificial selection standard reference color.
Description
Technical field
The invention belongs to Remote Sensing Image Processing Technology fields, are corrected more particularly to the color tone consistency of sequence remote sensing image
Method and system.
Background technology
Remote sensing image has more and more extensive in geographic mapping, disaster monitoring, resource investigation and urban planning etc.
Using.Quickly splice the ring in flow as remote sensing image, compared to for other modules, the light and color homogenization of sequential images is handled
The achievement in research of aspect is less, and is concentrated mainly on to the dodging in terms of image brilliance, includes mainly：The even light of MASK is calculated
Method, Wallis filter even smooth algorithm etc..It is more not classical for the color distortions problem such as existing colour cast aberration between image
With the processing method of mainstream.For large-scale measured zone, due to the influence of lighting angle and cloud cover etc., image
Between brightness and color distortion be difficult to avoid, this is even more so for the remote sensing image of multidate.Sequence remote sensing image
Light and color homogenization processing is related to several key technologies：The isolation technics of chromatic image luminance channel and Color Channel, the color of image
Transfer Technology is adjusted, the optimal consistent color of sequential images is with reference to subset search technology；
The isolation technics of brightness and Color Channel：The color space of presently described color has very much, is shown for equipment
It is exactly the rgb space of classics.Due between tri- channels RBG there is larger correlation, this image processing to subchannel
Algorithm is made troubles, therefore image is now gone to the very low space of Inter-channel Correlation by many image algorithms, after the completion of processing again
Go back to rgb space.Chromatic image mainly there are into HSV space and L α β by luminance channel and color for the separated color space in channel
Space.Wherein, the correlation of each interchannel is minimum in the spaces L α β, and the channels L indicate that luminance channel, the channels α and β are indicating two just
The Color Channel of friendship is an ideal the color space of image processing.
The tone Transfer Technology of image：The tone Transfer Technology of image refers to that will pass to wait for reference to the tone characteristics of image
The image of processing makes pending image have the tone with reference to image under the premise of keeping original presentation content.Main method
There are two classes：Transformation based on gray-scale statistical parameter (mean value and variance) and the transformation based on grey level histogram mapping.
There is presently no disclosed, and the consistent color reference subset search based on sequential images is theoretical.Light and color homogenization it is final
Purpose is that spliced remote sensing image is allowed to show unified tone.When there are a variety of inconsistent tones in sequence remote sensing image
When, it is difficult to determine that any is best standard colour tone.
Invention content
The tone inconsistence problems that the present invention is likely to occur for sequence remote sensing image in splicing, propose sequence remote sensing shadow
The color tone consistency bearing calibration of picture and system.
The present invention provides a kind of color tone consistency bearing calibration of sequence remote sensing image, includes the following steps：
Step 1, colored sequential images are gone into L α β color spaces from RGB color；
Step 2, tone transmits pretreatment, including is based on referring to image, using the Wallis Filtering Formulas of cum rights to image
All pixels carry out tone transmission respectively in channel L α β；
Step 3, the global gain compensation of luminance channel, including following sub-step is executed in luminance channel L to image greyscale
It adjusts,
Step 3.1, all adjacent images are counted to the gray average μ (I in overlapped public domainij), wherein
IijIndicate the lap with image j in image i；
Step 3.2, according to the energy of gain compensation, optimize the gray scale gain coefficient a of every imagei, i ∈ [1, N], wherein
N is image quantity；
Step 3.3, according to gain compensation factor optimal obtained by step 3.2, gain is done to the luminance channel of every image
Compensation is as follows,
Wherein, Ii(p) andGray value of any pixel p before and after gain compensation on image i is indicated respectively；
Step 4, the Histogram Mapping of Color Channel, including execute following sub-step Color Channel α and β separately
Image greyscale is adjusted,
Step 4.1, undirected weighted graph is built, including individual image is considered as a node, by the shadow with overlapping region
As being connected with side between node, side right is set as the image to the gray difference value D in overlapping regionH；
Step 4.2, given threshold TD, work as DH(Ii,Ij)≤TD, then IiAnd IjSide between image node is that tone is consistent
Otherwise side is the non-uniform side of tone；
Step 4.3, optimal reference image set is searched for, being included in search in undirected weighted graph has the consistent side of tone
Maximal connected subgraphs, the reference image set mapped using the image node set of the subgraph as color channel histograms；
Step 4.4, determine every image directly refers to image, include by the reference image node in undirected weighted graph it
Between weights be set as 0, appoint take one of node as root node；Determine each image node to root using shortest path first
The shortest path of node obtains binary tree, and the father node of each image node is exactly that it carries out the direct of grey level histogram mapping
With reference to image；
Step 4.5, the grey scale mapping sequence for determining image sequence, includes the binary tree to being obtained in step 4.4, from root section
Breadth First or the mode binary tree traversal of depth-first are pressed, the image node sequence of actual access is exactly between carrying out picture pair
Grey scale mapping executes sequence；
Step 4.6, according to the sequence that executes searched out, every image is established and it is with reference to the Histogram Mapping between image
Relationship, and adjust the gray scale of each pixel in image；
Step 5, all images are gone into RGB color from L α β color spaces, exports and preserves handling result.
Moreover, the calculating process of gray difference value is as follows between adjacent image pair,
Step 4.1.1, for image to IiAnd Ij, I is counted respectivelyiAnd IjGray accumulation probability distribution in overlapping region
Scheme CDF, K probability value p is equally spaced chosen in main sectionk, the corresponding image greyscale values of k ∈ [1, K] are as sample point, note
ForWithK is preset value；
Step 4.1.2, with IiSample point and IjSample point be reference axis establish plane right-angle coordinate ovivj, obtain
K discrete point is fitted to obtain solid-line curve with Quadric Spline；
Step 4.1.3 calculates area between block curve section and datum line, is denoted as As, then IiAnd IjGray difference value DH
(Ii,Ij)=As/(pK-p1)。
The present invention provides a kind of color tone consistency correction system of sequence remote sensing image, comprises the following modules：
First module, for colored sequential images to be gone to L α β color spaces from RGB color；
Second module transmits pretreatment for tone, including is based on referring to image, utilizes the Wallis Filtering Formulas of cum rights
Tone transmission is carried out respectively in channel L α β to all pixels of image；
Third module, the global gain for luminance channel compensates, including is arranged with lower unit in luminance channel L to image
Gray scale adjusts,
Statistic unit, for counting all adjacent images to the gray average μ (I in overlapped public domainij),
In, IijIndicate the lap with image j in image i；
Optimize unit, for the energy according to gain compensation, optimizes the gray scale gain coefficient a of every imagei,i∈[1,
N], wherein N is image quantity；
Compensating unit, for the gain compensation factor optimal according to unit gained is optimized, to the luminance channel of every image
It is as follows to do gain compensation,
Wherein, Ii(p) andGray value of any pixel p before and after gain compensation on image i is indicated respectively；
4th module, is used for the Histogram Mapping of Color Channel, including is arranged and is distinguished in Color Channel α and β with lower unit
Independently image greyscale is adjusted,
Initial cell is considered as a node for building undirected weighted graph, including by individual image, will have overlapping region
Image node between with side be connected, side right is set as the image to the gray difference value D in overlapping regionH；
Judging unit is used for given threshold TD, work as DH(Ii,Ij)≤TD, then IiAnd IjSide between image node is tone
Otherwise consistent side is the non-uniform side of tone；
Search unit, for searching for optimal reference image set, being included in search in undirected weighted graph has tone one
Cause the maximal connected subgraphs on side, the reference image set mapped using the image node set of the subgraph as color channel histograms
It closes；
Reference unit includes by the reference image in undirected weighted graph for determining that every the direct of image refers to image
Weights between node are set as 0, appoint and take one of node as root node；Each image section is determined using shortest path first
Point obtains binary tree, the father node of each image node is exactly that it carries out grey level histogram mapping to the shortest path of root node
Directly refer to image；
Traversal Unit, the grey scale mapping sequence for determining image sequence, includes the binary tree to being obtained in reference unit,
From root node, binary tree traversal, the image node sequence of actual access are exactly to carry out in the way of breadth First or depth-first
Grey scale mapping executes sequence between picture pair；
Adjustment unit, for according to the sequence that executes searched out, every image to be established and it is with reference to the histogram between image
Figure mapping relations, and adjust the gray scale of each pixel in image；
5th module, for all images to be gone to RGB color from L α β color spaces, export and preserve processing knot
Fruit.
Moreover, the calculating process of gray difference value is as follows between adjacent image pair,
For image to IiAnd Ij, I is counted respectivelyiAnd IjGray accumulation probability distribution graph CDF in overlapping region, in master
Section is wanted equally spaced to choose K probability value pk, the corresponding image greyscale values of k ∈ [1, K] are denoted as sample point
WithK is preset value；
With IiSample point and IjSample point be reference axis establish plane right-angle coordinate ovivj, K discrete point is obtained,
It is fitted to obtain solid-line curve with Quadric Spline；
Area between block curve section and datum line is calculated, A is denoted ass, then IiAnd IjGray difference value DH(Ii,Ij)=
As/(pK-p1)。
The present invention utilizes the tone correspondence of overlapping region between adjacent image, and proposition is a kind of to be effectively inhibited or eliminate
The technical solution of existing brightness and color distortion between image, the advantage of the technical program are：(1) a whole set of process flow is all
It is carried out in the higher color space of interchannel independence, can largely reduce the image brought due to Inter-channel Correlation
Distortion；(2) brightness and color are separately handled, for respective characteristic, takes different processing schemes, can effectively promotes sequence
The integral color consistency of image.(3) the technical program takes search most for color distortion problem present in sequential images
The problem of mode of big coherent image subset chooses reference color, avoids artificial selection standard reference color automatically.
Description of the drawings
Fig. 1 is the process chart of the color tone consistency bearing calibration of the carried sequence remote sensing image of the embodiment of the present invention；
Fig. 2 is CDF curves equiprobability spacing sample point schematic diagram in the embodiment of the present invention, and horizontal axis V indicates the gray scale of pixel
Value, longitudinal axis P represent less than the pixel of some gray value probability of occurrence in image；
Fig. 3 is that Histogram distance describes schematic diagram between image in the embodiment of the present invention, horizontally and vertically indicates image i and shadow
As the gray value that pixel is likely to occur in j；
Fig. 4 is the relevant classification side non-directed graph of image sequence in the embodiment of the present invention, and node number indicates the label of image, side
Indicate that there is overlapping relation between image pair；
Fig. 5 be in the embodiment of the present invention image non-directed graph using with reference to image set as the minimum spanning tree of root node.
Specific implementation mode
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, not
For limiting the present invention.
The technical solution adopted by the present invention provides a kind of method carrying out light and color homogenization processing to sequence remote sensing image.Such as figure
It is consistent with Color Channel progress to luminance channel respectively by the sequential images L α β space low in each Inter-channel Correlation shown in 1
Adjustment, finally transfers back to rgb space and obtains final process result figure again.Including following processing step：
Step 1, colored sequential images are gone into L α β color spaces from RGB color.Transfer process is as follows：
Step 1.1, the gray scale vector of pixel is gone into the spaces LMS, transition matrix T from rgb space1：
Step 1.2, the gray scale vector of pixel is gone into the spaces L α β, transition matrix T from the spaces LMS2：
Step 2, tone transmits pretreatment.It is close with area's situation is surveyed that a content can be pre-selected in or beyond sequential images
And image quality preferably remote sensing image is used as and refers to image, using the Wallis Filtering Formulas of cum rights to each image in sequence
All pixels carry out tone transmission respectively in the channels L α β.Cum rights Wallis Filtering Formulas are as follows：
μ () and σ () effects symbol indicates gray average and variance to object pixel set, Ν in formulan×n() acts on
Symbol indicates n × n neighborhood territory pixel set of object pixel, when it is implemented, the value n that those skilled in the art can be voluntarily arranged.I
(p) andThe gray value after pixel p is before treatment in pending image, I are indicated respectivelyrIndicate all pixels with reference to image
Set.Weighting parameter λ ∈ [0,1] control the intensity of the change action, and weights are bigger, and tone transmission effect is more apparent, otherwise also
So.Recommendation weights of the present invention in the channels L α β are respectively 0.8,0.1,0.1.
Step 3, the global gain compensation of luminance channel.Image greyscale is adjusted in L luminance channels, counts each phase first
The gray average of overlapping region between adjacent image, then it is excellent using the global consistency of brightness progress of the gain model to image sequence
Change.
Step 3 specific implementation of embodiment includes the following steps：
Step 3.1, all adjacent images are counted to the gray average μ (I in overlapped public domainij), wherein
IijIndicate the lap pixel set with adjacent image j in certain image i.
Step 3.2, optimize the gray scale gain coefficient a of every imagei, i ∈ [1, N], wherein N are image quantity.To Mr. Yu
Image i and adjacent image j, the energy equation of gain compensation：
eij=(aiμ(Iij)-ajμ(Iji))2+α((ai-1)2+(aj-1)2)
Wherein, eijFor the energy of gain compensation, α is a fixed weights, avoids the trivial solution of equation, IjiIndicate certain
In image j with the lap pixel set of adjacent image i.In view of all images are to the gray difference of lap, entirely
The energy equation of office：
Above formula can be by linear least square method rapid solving to get to the gain compensation factor a of each imagei。
Step 3.3, according to the optimum gain penalty coefficient acquired, gain compensation is done to the luminance channel of every image：Wherein, Ii(p) andGray value of any pixel p before and after gain compensation on image i is indicated respectively.
Step 4, the Histogram Mapping of Color Channel.Separately image greyscale is adjusted in α and β Color Channels,
Include the color distortion according to overlapping region, searches for optimal with reference to image set；Using shortest path first, every image is determined
Directly refer to image and sequence；It establishes and with reference to the Histogram Mapping relationship between image, adjusts image greyscale.
Step 4 specific implementation of embodiment is included in the channels α and β and executes following steps respectively：
Step 4.1, undirected weighted graph is built.Every image is considered as a node, and (node number is the sequence of image in Fig. 4
Label), it will be connected with side between the image node with overlapping region, side right is set as the image to the color in overlapping region
(gray scale) difference value DH.The calculating process of gray difference value is following (with image to I between adjacent image pairiAnd IjFor)：
Step 4.1.1, counts I respectivelyiAnd IjGray accumulation probability distribution graph CDF in overlapping region, in probability value
K probability value p is equally spaced chosen between 0.5%~99.5%k, the corresponding image greyscale values of k ∈ [1, K] as sample point,
It is denoted asWithAs shown in Figure 2.When it is implemented, those skilled in the art can voluntarily preset taking for K
Value, preferred value are 8.
Step 4.1.2, with IiSample point and IjSample point be reference axis establish plane right-angle coordinate ovivj, then
To K discrete point, the solid-line curve in Fig. 3 is fitted with Quadric Spline.
Step 4.1.3 calculates area between block curve section and datum line (crossing the straight line that origin and slope are 1), is denoted as
As, then IiAnd IjColor (gray scale) difference value DH(Ii,Ij)=As/(pK-p1)。
Step 4.2, given threshold TD, the side of undirected weighted graph is divided into two classes：Tone it is consistent while and tone it is non-uniform while.
Work as DH(Ii,Ij)≤TD, then IiAnd IjBetween image node while for " when tone is consistent ", such as dotted line side in Fig. 4；Otherwise, it is " color
Adjust non-uniform side ", such as solid line side in Fig. 4.When it is implemented, those skilled in the art can voluntarily predetermined threshold value, in the channels α and β
Recommendation threshold value be respectively 0.03 and 0.005.
Step 4.3, it searches for optimal with reference to image set.Maximum of the search with " the consistent side of tone " in undirected weighted graph
Connected subgraph, the reference image set mapped using the image node set of the subgraph as color channel histograms, in Fig. 4
Solid node.
Step 4.4, determine that every the direct of image refers to image.It will be between the reference image node in undirected weighted graph
Weights are set as 0, appoint and take one of node as root node.Determine each image node to root node using shortest path first
Shortest path, obtained " binary tree " be as shown in Figure 5.The father node of each image node is exactly that its progress grey level histogram reflects
That penetrates directly refers to image.
Step 4.5, the grey scale mapping sequence of image sequence is determined.To the binary tree obtained in step 4.4, pressed from root node
Breadth First or the mode binary tree traversal of depth-first, the image node sequence of actual access are exactly to carry out gray scale between picture pair
Mapping executes sequence.
Step 4.6, according to the sequence that executes searched out, every image is established and it is with reference to the Histogram Mapping between image
Relationship adjusts each pixel in image such as correspondence of the transverse and longitudinal coordinate about solid-line curve in Fig. 3, and according to this grey scale mapping relationship
Gray scale.
Step 5, all images being gone into RGB color from L α β color spaces, conversion regime is the inverse operation of step 1,
Handling result is exported and preserved, final result image sequence is obtained.
When it is implemented, method provided by the present invention, which can be based on software technology, realizes automatic running flow, mould can also be used
Block mode realizes corresponding system.The embodiment of the present invention also provides a kind of color tone consistency correction system of sequence remote sensing image,
It comprises the following modules：
First module, for colored sequential images to be gone to L α β color spaces from RGB color；
Second module transmits pretreatment for tone, including is based on referring to image, utilizes the Wallis Filtering Formulas of cum rights
Tone transmission is carried out respectively in channel L α β to all pixels of image；
Third module, the global gain for luminance channel compensates, including is arranged with lower unit in luminance channel L to image
Gray scale adjusts,
Statistic unit, for counting all adjacent images to the gray average μ (I in overlapped public domainij),
In, IijIndicate the lap with image j in image i；
Optimize unit, for the energy according to gain compensation, optimizes the gray scale gain coefficient a of every imagei,i∈[1,
N], wherein N is image quantity；
Compensating unit, for the gain compensation factor optimal according to unit gained is optimized, to the luminance channel of every image
It is as follows to do gain compensation,
Wherein, Ii(p) andGray value of any pixel p before and after gain compensation on image i is indicated respectively；
4th module, is used for the Histogram Mapping of Color Channel, including is arranged and is distinguished in Color Channel α and β with lower unit
Independently image greyscale is adjusted,
Initial cell is considered as a node for building undirected weighted graph, including by individual image, will have overlapping region
Image node between with side be connected, side right is set as the image to the gray difference value D in overlapping regionH；
Judging unit is used for given threshold TD, work as DH(Ii,Ij)≤TD, then IiAnd IjSide between image node is tone
Otherwise consistent side is the non-uniform side of tone；
Search unit, for searching for optimal reference image set, being included in search in undirected weighted graph has tone one
Cause the maximal connected subgraphs on side, the reference image set mapped using the image node set of the subgraph as color channel histograms
It closes；
Reference unit includes by the reference image in undirected weighted graph for determining that every the direct of image refers to image
Weights between node are set as 0, appoint and take one of node as root node；Each image section is determined using shortest path first
Point obtains binary tree, the father node of each image node is exactly that it carries out grey level histogram mapping to the shortest path of root node
Directly refer to image；
Traversal Unit, the grey scale mapping sequence for determining image sequence, includes the binary tree to being obtained in reference unit,
From root node, binary tree traversal, the image node sequence of actual access are exactly to carry out in the way of breadth First or depth-first
Grey scale mapping executes sequence between picture pair；
Adjustment unit, for according to the sequence that executes searched out, every image to be established and it is with reference to the histogram between image
Figure mapping relations, and adjust the gray scale of each pixel in image；
5th module, for all images to be gone to RGB color from L α β color spaces, export and preserve processing knot
Fruit.
Each module specific implementation can be found in corresponding steps, and it will not go into details by the present invention.
Specific embodiment described herein is only an example for the spirit of the invention.Technology belonging to the present invention is led
The technical staff in domain can make various modifications or additions to the described embodiments or replace by a similar method
In generation, however, it does not deviate from the spirit of the invention or beyond the scope of the appended claims.
Claims (4)
1. a kind of color tone consistency bearing calibration of sequence remote sensing image, which is characterized in that include the following steps：
Step 1, colored sequence remote sensing image is gone into L α β color spaces from RGB color；
Step 2, tone transmits pretreatment, including is based on referring to image, using the Wallis Filtering Formulas of cum rights to sequence remote sensing
The all pixels of image carry out tone transmission respectively in channel L α β；
Step 3, the global gain compensation of luminance channel, including following sub-step is executed in luminance channel L to sequence remote sensing image
Gray scale adjusts,
Step 3.1, all adjacent images are counted to the gray average μ (I in overlapped public domainij), wherein IijIt indicates
In image i with the lap of image j；
Step 3.2, according to the energy of gain compensation, optimize the gray scale gain compensation factor a of every sequence remote sensing imagei,i∈[1,
N], wherein N is image quantity；
Step 3.3, according to gain compensation factor optimal obtained by step 3.2, the luminance channel of every sequence remote sensing image is done
Gain compensation is as follows,
Wherein, Ii(p) andGray value of any pixel p before and after gain compensation on image i is indicated respectively；
Step 4, the Histogram Mapping of Color Channel, including following sub-step is executed in Color Channel α and β separately to sequence
Row remote sensing image gray scale adjusts,
Step 4.1, undirected weighted graph is built, including individual sequence remote sensing image is considered as a node, there will be overlapping region
Image node between with side be connected, side right is set as the adjacent image to the gray difference value D in overlapping regionH；
Step 4.2, given threshold TD, work as DH(Ii,Ij)≤TD, then IiAnd IjBetween image node while consistent for tone, it is no
Then, it is the non-uniform side of tone；
Step 4.3, optimal reference image set is searched for, maximum of the search with the consistent side of tone in undirected weighted graph is included in
Connected subgraph, the reference image set mapped using the image node set of the subgraph as color channel histograms；
Step 4.4, it determines that every the direct of sequence remote sensing image refers to image, includes by the reference image section in undirected weighted graph
Weights between point are set as 0, appoint and take one of node as root node；Each image node is determined using shortest path first
Father node to the shortest path of root node, each image node is exactly that it carries out the direct with reference to shadow of grey level histogram mapping
Picture；
Step 4.5, the grey scale mapping sequence for determining sequence remote sensing image, includes the binary tree to being obtained in step 4.4, from root section
Press breadth First or the mode binary tree traversal of depth-first, the image node sequence of actual access is exactly that carry out sequence distant
Sense image and its execute sequence with reference to grey scale mapping between image；
Step 4.6, according to the sequence that executes searched out, every sequence remote sensing image is established and it is with reference to the histogram between image
Mapping relations, and adjust the gray scale of each pixel in sequence remote sensing image；
Step 5, all sequences remote sensing image is gone into RGB color from L α β color spaces, exports and preserves handling result.
2. the color tone consistency bearing calibration of sequence remote sensing image according to claim 1, it is characterised in that：Adjacent image pair
It is as follows in the calculating process of the gray difference value of overlapping region,
Step 4.1.1, for image to IiAnd Ij, I is counted respectivelyiAnd IjGray accumulation probability distribution graph CDF in overlapping region,
K probability value p is equally spaced chosen between probability value 0.5%~99.5%k, the corresponding image greyscale value conducts of k ∈ [1, K]
Sample point is denoted asWithK is preset value；
Step 4.1.2, with IiSample point and IjSample point be reference axis establish plane right-angle coordinate ovivj, obtain K from
Scatterplot is fitted to obtain block curve with Quadric Spline；
Step 4.1.3 calculates area between block curve section and datum line, is denoted as As, then IiAnd IjGray difference value DH(Ii,
Ij)=As/(pK-p1)。
3. a kind of color tone consistency of sequence remote sensing image corrects system, which is characterized in that comprise the following modules：
First module, for colored sequence remote sensing image to be gone to L α β color spaces from RGB color；
Second module transmits pretreatment for tone, including is based on referring to image, using the Wallis Filtering Formulas of cum rights to sequence
The all pixels of row remote sensing image carry out tone transmission respectively in channel L α β；
Third module, the global gain for luminance channel compensates, including is arranged with lower unit in luminance channel L to sequence remote sensing
Image greyscale adjusts,
Statistic unit, for counting all adjacent images to the gray average μ (I in overlapped public domainij), wherein
IijIndicate the lap with image j in image i；
Optimize unit, for the energy according to gain compensation, optimizes the gray scale gain compensation factor a of every sequence remote sensing imagei,i
∈ [1, N], wherein N are image quantity；
Compensating unit, for the gain compensation factor optimal according to unit gained is optimized, the brightness to every sequence remote sensing image
It is as follows that gain compensation is done in channel,
Wherein, Ii(p) andGray value of any pixel p before and after gain compensation on image i is indicated respectively；
4th module, be used for Color Channel Histogram Mapping, including be arranged with lower unit Color Channel α and β independently
Ground adjusts sequence remote sensing image gray scale,
Initial cell is considered as a node for building undirected weighted graph, including by individual sequence remote sensing image, will have overlapping
It is connected with side between the image node in region, side right is set as the adjacent image to the gray difference value D in overlapping regionH；
Judging unit is used for given threshold TD, work as DH(Ii,Ij)≤TD, then IiAnd IjSide between image node is that tone is consistent
Otherwise side is the non-uniform side of tone；
Search unit, for searching for optimal reference image set, being included in search in undirected weighted graph has the consistent side of tone
Maximal connected subgraphs, the reference image set mapped using the image node set of the subgraph as color channel histograms；
Reference unit includes by the reference in undirected weighted graph for determining that every the direct of sequence remote sensing image refers to image
Weights between image node are set as 0, appoint and take one of node as root node；Each shadow is determined using shortest path first
As the shortest path of node to root node, binary tree is obtained, the father node of each image node is exactly that it carries out grey level histogram
Mapping directly refers to image；
Traversal Unit, the grey scale mapping sequence for determining sequence remote sensing image, includes the binary tree to being obtained in reference unit,
From root node, binary tree traversal, the image node sequence of actual access are exactly to carry out in the way of breadth First or depth-first
Sequence remote sensing image and its execute sequence with reference to grey scale mapping between image；
Adjustment unit, for according to the sequence that executes searched out, every sequence remote sensing image to be established and it is with reference between image
Histogram Mapping relationship, and adjust the gray scale of each pixel in sequence remote sensing image；
5th module exports for all sequences remote sensing image to be gone to RGB color from L α β color spaces and preserves place
Manage result.
4. the color tone consistency of sequence remote sensing image corrects system according to claim 3, it is characterised in that：Adjacent image pair
It is as follows in the calculating process of the gray difference value of overlapping region,
For image to IiAnd Ij, I is counted respectivelyiAnd IjGray accumulation probability distribution graph CDF in overlapping region, in probability value
K probability value p is equally spaced chosen between 0.5%~99.5%k, the corresponding image greyscale values of k ∈ [1, K] as sample point,
It is denoted asWithK is preset value；
With IiSample point and IjSample point be reference axis establish plane right-angle coordinate ovivj, K discrete point is obtained, with two
Secondary B-spline fits to obtain block curve；
Area between block curve section and datum line is calculated, A is denoted ass, then IiAnd IjGray difference value DH(Ii,Ij)=As/(pK-
p1)。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610269941.4A CN105957111B (en) | 2016-04-27 | 2016-04-27 | The color tone consistency bearing calibration of sequence remote sensing image and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610269941.4A CN105957111B (en) | 2016-04-27 | 2016-04-27 | The color tone consistency bearing calibration of sequence remote sensing image and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105957111A CN105957111A (en) | 2016-09-21 |
| CN105957111B true CN105957111B (en) | 2018-09-18 |
Family
ID=56915733
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610269941.4A Active CN105957111B (en) | 2016-04-27 | 2016-04-27 | The color tone consistency bearing calibration of sequence remote sensing image and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105957111B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109272465B (en) * | 2018-09-15 | 2021-11-05 | 武汉智觉空间信息技术有限公司 | Aviation image color consistency processing algorithm |
| CN109410136B (en) * | 2018-10-10 | 2020-10-27 | 中国人民解放军战略支援部队信息工程大学 | Color homogenizing method and processing device based on shortest transmission path |
| CN110555822B (en) * | 2019-09-05 | 2023-08-29 | 北京大视景科技有限公司 | Color consistency adjustment method for real-time video fusion |
| CN110599424B (en) * | 2019-09-16 | 2020-07-14 | 北京航天宏图信息技术股份有限公司 | Method and device for automatic image color-homogenizing processing, electronic equipment and storage medium |
| TWI741437B (en) * | 2019-12-09 | 2021-10-01 | 財團法人資訊工業策進會 | Image analysis apparatus and image analysis method |
| CN111277764B (en) * | 2020-03-10 | 2021-06-01 | 西安卓越视讯科技有限公司 | 4K real-time video panorama stitching method based on GPU acceleration |
| CN111652826B (en) * | 2020-05-18 | 2023-04-25 | 哈尔滨工业大学 | Method for homogenizing multiple/hyperspectral remote sensing images based on Wallis filtering and histogram matching |
| CN115917639A (en) * | 2020-07-21 | 2023-04-04 | 高通股份有限公司 | Method for reducing color gamut mapping brightness loss |
| CN112884676B (en) * | 2021-03-18 | 2023-04-14 | 国家海洋信息中心 | Large-range aerial remote sensing image color matching method based on space division progressive control |
| CN112907485B (en) * | 2021-03-18 | 2023-04-25 | 国家海洋信息中心 | Remote sensing image batch toning method based on lalpha beta space color mapping |
| CN112950510B (en) * | 2021-03-22 | 2024-04-02 | 南京莱斯电子设备有限公司 | Large scene spliced image chromatic aberration correction method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542528A (en) * | 2011-12-26 | 2012-07-04 | Tcl集团股份有限公司 | Image conversion processing method and system |
| CN103065334A (en) * | 2013-01-31 | 2013-04-24 | 金陵科技学院 | Color cast detection and correction method and device based on HSV (Hue, Saturation, Value) color space |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20130067917A (en) * | 2011-12-14 | 2013-06-25 | 한국전자통신연구원 | Color correcting apparatus and method for color consistency between stereoscopic display devices |
-
2016
- 2016-04-27 CN CN201610269941.4A patent/CN105957111B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542528A (en) * | 2011-12-26 | 2012-07-04 | Tcl集团股份有限公司 | Image conversion processing method and system |
| CN103065334A (en) * | 2013-01-31 | 2013-04-24 | 金陵科技学院 | Color cast detection and correction method and device based on HSV (Hue, Saturation, Value) color space |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105957111A (en) | 2016-09-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105957111B (en) | The color tone consistency bearing calibration of sequence remote sensing image and system | |
| CN110619632B (en) | Mango example confrontation segmentation method based on Mask R-CNN | |
| CN107016647B (en) | Panoramic picture color tone consistency correcting method and system | |
| CN110428371A (en) | Image defogging method, system, storage medium and electronic equipment based on super-pixel segmentation | |
| US20150304525A1 (en) | Color correction based on multiple images | |
| CN111798467A (en) | Image segmentation method, device, equipment and storage medium | |
| JP2021531571A (en) | Certificate image extraction method and terminal equipment | |
| CN109147005A (en) | It is a kind of for the adaptive colouring method of infrared image, system, storage medium, terminal | |
| CN103971134B (en) | Image classification, retrieval and bearing calibration, and related device | |
| CN107993209A (en) | Image processing method, device, computer-readable recording medium and electronic equipment | |
| US20230059499A1 (en) | Image processing system, image processing method, and non-transitory computer readable medium | |
| CN116485785A (en) | Surface defect detection method for solar cell | |
| CN114241326A (en) | Progressive intelligent production method and system for ground feature elements of remote sensing images | |
| CN107909542A (en) | Image processing method, device, computer-readable recording medium and electronic equipment | |
| CN114067222B (en) | Urban water body remote sensing classification method and device | |
| CN114187880A (en) | Image display method, image display device and display system | |
| CN114202491B (en) | Method and system for enhancing optical image | |
| CN116309110A (en) | Low-light image defogging method based on lightweight deep neural network | |
| CN113436081A (en) | Data processing method, image enhancement method and model training method thereof | |
| CN115375561A (en) | Remote sensing image dodging and color homogenizing method for optimizing brightness, contrast and chromatic aberration | |
| CN114187515A (en) | Image segmentation method and image segmentation device | |
| Su et al. | Image enhancement and brightness equalization algorithms in low illumination environment based on multiple frame sequences | |
| CN111640068A (en) | Unsupervised automatic correction method for image exposure | |
| CN109451292A (en) | Color temp bearing calibration and device | |
| CN116703744B (en) | Remote sensing image dodging and color homogenizing method and device based on convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20180515 Address after: 518063 north half floor, 27 / F, East Tower building, Baidu International Building, Shenzhen, Guangdong, Nanshan District Applicant after: SHENZHEN JIMU YIDA SCIENCE AND TECHNOLOGY Co.,Ltd. Address before: 430072 Wuhan University, Luojia mountain, Wuchang District, Wuhan, Hubei Applicant before: Wuhan University |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Method and System for Tone Consistency Correction of Sequential Remote Sensing Images Effective date of registration: 20230914 Granted publication date: 20180918 Pledgee: CITIC Bank Limited by Share Ltd. Shenzhen branch Pledgor: SHENZHEN JIMU YIDA SCIENCE AND TECHNOLOGY Co.,Ltd. Registration number: Y2023980056847 |