CN109886193A - A kind of cirrus cloud detection method, device and computer readable storage medium - Google Patents
A kind of cirrus cloud detection method, device and computer readable storage medium Download PDFInfo
- Publication number
- CN109886193A CN109886193A CN201910129005.7A CN201910129005A CN109886193A CN 109886193 A CN109886193 A CN 109886193A CN 201910129005 A CN201910129005 A CN 201910129005A CN 109886193 A CN109886193 A CN 109886193A
- Authority
- CN
- China
- Prior art keywords
- grayscale image
- primary channel
- pixel
- cirrus
- channel grayscale
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 43
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 238000004364 calculation method Methods 0.000 claims abstract description 15
- 230000006870 function Effects 0.000 claims description 49
- 230000015654 memory Effects 0.000 claims description 34
- 206010028980 Neoplasm Diseases 0.000 claims description 27
- 238000004422 calculation algorithm Methods 0.000 claims description 27
- 238000012360 testing method Methods 0.000 claims description 16
- 238000004891 communication Methods 0.000 claims description 11
- 238000002372 labelling Methods 0.000 claims description 4
- 238000000034 method Methods 0.000 description 55
- 230000008569 process Effects 0.000 description 14
- 238000012545 processing Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 10
- 238000004590 computer program Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 8
- 238000010801 machine learning Methods 0.000 description 5
- 230000001360 synchronised effect Effects 0.000 description 5
- 238000005520 cutting process Methods 0.000 description 4
- 230000003190 augmentative effect Effects 0.000 description 3
- 238000005253 cladding Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 239000003086 colorant Substances 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 230000004807 localization Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- KLDZYURQCUYZBL-UHFFFAOYSA-N 2-[3-[(2-hydroxyphenyl)methylideneamino]propyliminomethyl]phenol Chemical compound OC1=CC=CC=C1C=NCCCN=CC1=CC=CC=C1O KLDZYURQCUYZBL-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 201000001098 delayed sleep phase syndrome Diseases 0.000 description 1
- 208000033921 delayed sleep phase type circadian rhythm sleep disease Diseases 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000010410 layer Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035800 maturation Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
本发明公开了一种卷云检测方法，包括以下步骤：输入遥感图像，并获取遥感图像的原色通道灰度图；获取原色通道灰度图的分形维数特征图；基于分形维数特征图和原色通道灰度图获取权值灰度图；对原色通道灰度图的像素点进行聚类计算，获取像素点的聚类结果，基于聚类结果确定原色通道灰度图每一像素点的预分配标签；基于原色通道灰度图、权值灰度图和预分配标签建立图割模型；对图割模型进行最小割计算，输出遥感图像的检测结果。本发明还同时公开了一种卷云检测设备及计算机可读存储介质。
The invention discloses a cirrus cloud detection method, comprising the following steps: inputting a remote sensing image, and obtaining a primary color channel grayscale map of the remote sensing image; obtaining a fractal dimension feature map of the primary color channel grayscale map; based on the fractal dimension feature map and The grayscale image of the primary color channel obtains the weight grayscale image; the pixel points of the grayscale image of the primary color channel are clustered to obtain the clustering result of the pixel point, and the pre-determination of each pixel point of the grayscale image of the primary color channel is determined based on the clustering result. Assign labels; build a graph cut model based on primary color channel grayscale map, weight grayscale map and pre-assigned labels; perform minimum cut calculation on the graph cut model, and output the detection results of remote sensing images. The invention also discloses a cirrus cloud detection device and a computer-readable storage medium at the same time.
Description
Technical field
The present invention relates to technical field of remote sensing image processing more particularly to a kind of cirrus detection methods, equipment and computer
Readable storage medium storing program for executing.
Background technique
Cirrus detection in remote sensing images has extremely wide application, such as weather forecast, geographical monitoring in life.
Specifically, since in remotely sensed image, cirrus can block other regions, therefore for of different shapes in remote sensing images
Cirrus how to extract and research hotspot in recent years.And the shape of cirrus is often changeable, brightness and the high spoke of some remote sensing
Penetrate that object is such as avenged and white building is difficult to differentiate between, and there is also some translucent thin clouds, these situations can all make to roll up
The difficulty of cloud detection greatly improves.
In the prior art, the cirrus detection method in remote sensing images is mainly based upon the detection method of single-frame images, including
Utilization support vector machines (the supportvector that thresholding method and machine learning method, such as Kang et al. were proposed in 2017
Machine, SVM) training multiple features fusion model method that cirrus is detected；What Yuan et al. was proposed in 2015 adopts
Region between division cirrus and other objects is combined with SVM with bag of words (bag of words, BoW) to examine cirrus
The method of survey；What Zhan et al. was proposed in 2017 uses convolutional neural networks (Convolutional neural
Networks, CNN) carry out the method etc. that feature extraction detects cirrus.But since thresholding method can not extract well
The texture and structural characteristic of cirrus, and the contrast condition in remote sensing images between cirrus and other objects is excessively relied on, therefore
It is difficult to differentiate between cirrus and other high radiating objects；And machine learning method needs a large amount of sample data to be trained, therefore works as sample
When this is limited, cirrus can not effectively be detected.
On the other hand, figure cuts a kind of image partition method commonly used multiple fields of the method as prevalence, by using figure
The method of cutting can divide the image into problem and be converted into model optimization problem, to reduce in image detection for the dependence journey of feature
Degree, and there is the case where for finite sample good applicability.
Summary of the invention
It is an object of the invention to: a kind of cirrus detection method is provided, cuts algorithm detection cirrus by using point shape and figure,
Solve in the detection of existing cirrus when encountering high radiating object interference and finite sample data using thresholding method and
The bad problem of machine learning method detection effect.
The present invention specifically uses following technical scheme to achieve the goals above:
In a first aspect, the present invention discloses a kind of cirrus detection method, comprising the following steps:
Step 1, input remote sensing images, and obtain the primary channel grayscale image of remote sensing images；
Step 2, the Cancers Fractional Dimension Feature figure for obtaining primary channel grayscale image；
Step 3 obtains weight grayscale image based on Cancers Fractional Dimension Feature figure and primary channel grayscale image；
Step 4 carries out cluster calculation to the pixel of primary channel grayscale image, obtains the cluster result of pixel, is based on
Cluster result determines the predistribution label of each pixel of primary channel grayscale image；
Step 5 is established figure and is cut model based on primary channel grayscale image, weight grayscale image and predistribution label；
Step 6 cuts model progress minimal cut calculating to figure, exports the testing result of remote sensing images.
Further, step 2 specifically includes:
FRACTAL SURFACES surface area S (n) in step 2.1, the calculating primary channel grayscale image at each pixel,
In, n indicates unit area scale；
Step 2.2, the fractal dimension that each pixel is calculated based on FRACTAL SURFACES surface area S (n) and fractal surface formula
d；Wherein, the fractal surface formula is S (n)=n2-d；
Step 2.3, the fractal dimension d based on each pixel obtain the Cancers Fractional Dimension Feature of the primary channel grayscale image
Figure.
Further, step 3 specifically includes:
The weighting that is multiplied is carried out using the Cancers Fractional Dimension Feature figure as weight with the primary channel grayscale image, based on weighting
As a result weight grayscale image is obtained.
Further, step 4 specifically includes:
Step 4.1 randomly selects initial value of the K pixel as cluster centre in the primary channel grayscale image,
Calculate the Europe between other pixels and the cluster centre in the primary channel grayscale image in addition to the cluster centre
Formula distance；
Step 4.2 clusters all pixels point based on the Euclidean distance, and obtain in each cluster with it is European away from
From the immediate pixel of mean value be iterated calculating as new cluster centre, when cluster centre no longer changes obtain most
Whole cluster result；
Step 4.3, the predistribution that each pixel in the primary channel grayscale image is determined based on final cluster result
Label.
Further, step 5 specifically includes:
Step 5.1 determines that the figure cuts the criterion function of model；
Step 5.2, the area item that the criterion function is determined based on the primary channel grayscale image；Based on the weight ash
Degree figure determines the border item of the criterion function；The calculating initial value of the criterion function is determined based on the predistribution label；
Step 5.3 establishes the figure based on the criterion function and cuts model.
Further, the criterion function that figure cuts model specifically includes:
E (L)=a*R (L)+B (L)；Wherein, L is the calculating initial value of the criterion function, and R (L) is the criterion function
Area item, B (L) be the criterion function border item, a be preset balance parameters, E (L) be figure cut value.
Further, model is cut to figure and carries out minimal cut calculating, specifically included:
The minimal cut value that the figure cuts model is calculated using Ford-Fulkerson labeling algorithm.
Further, the primary channel grayscale image is blue channel grayscale image.
Second aspect, the present invention disclose a kind of cirrus detection device, comprising:
Processor, memory and communication bus；
Wherein, the communication bus, for realizing the communication connection between the processor and the memory；
The memory detects program for storing the cirrus that can be run on the processor；
The processor, is used for:
Remote sensing images are inputted, and obtain the primary channel grayscale image of remote sensing images；
Obtain the Cancers Fractional Dimension Feature figure of primary channel grayscale image；
Weight grayscale image is obtained based on Cancers Fractional Dimension Feature figure and primary channel grayscale image；
Cluster calculation is carried out to the pixel of primary channel grayscale image, obtains the cluster result of pixel, based on cluster knot
Fruit determines the predistribution label of pixel；
Determine that figure cuts model based on primary channel grayscale image, weight grayscale image and predistribution label；
It cuts model to figure to calculate, and the testing result based on calculated result output remote sensing images.
The third aspect, the present invention disclose a kind of computer readable storage medium, are stored with one in the readable storage medium storing program for executing
A or multiple programs, one or more of programs can be executed by one or more processor, to realize in first aspect
The step of any one cirrus detection method.
After adopting the above scheme, beneficial effects of the present invention are as follows:
The invention avoids limitations and machine that conventional threshold values method can not differentiate cirrus Yu other high radiating objects
Learn class method to sample size and extract feature dependence, using extract cirrus Cancers Fractional Dimension Feature to figure cut algorithm into
The method of row optimization realizes effective detection to cirrus, solves the cirrus test problems in the case of limited sample size, together
When improve the accuracy rate and recall rate of testing result.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is a kind of cirrus detection method flow diagram that the embodiment of the present invention 1 provides；
Fig. 2 is a kind of cirrus detection method flow diagram that the embodiment of the present invention 2 provides；
Fig. 3 is the remote sensing images in the embodiment of the present invention 2；
Fig. 4 is the primary channel grayscale image in the embodiment of the present invention 2；
Fig. 5 is the Cancers Fractional Dimension Feature figure in the embodiment of the present invention 2；
Fig. 6 is the weight grayscale image in the embodiment of the present invention 2；
Fig. 7 is the testing result of the remote sensing images in the embodiment of the present invention 2；
Fig. 8 is a kind of hardware structural diagram for cirrus detection device that the embodiment of the present invention 3 provides.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention 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, i.e., described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is logical
The component for the embodiment of the present invention being often described and illustrated herein in the accompanying drawings can be arranged and be designed with a variety of different configurations.
Therefore, the detailed description of the embodiment of the present invention provided in the accompanying drawings is not intended to limit below claimed
The scope of the present invention, but be merely representative of selected embodiment of the invention.Based on the embodiment of the present invention, those skilled in the art
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
It should be noted that the relational terms of term " first " and " second " or the like be used merely to an entity or
Operation is distinguished with another entity or operation, and without necessarily requiring or implying between these entities or operation, there are any
This actual relationship or sequence.Moreover, the terms "include", "comprise" or its any other variant be intended to it is non-exclusive
Property include so that include a series of elements process, method, article or equipment not only include those elements, but also
Further include other elements that are not explicitly listed, or further include for this process, method, article or equipment it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including element
Process, method, article or equipment in there is also other identical elements.
It elaborates with reference to the accompanying drawings and examples to the embodiment of the present invention.
Embodiment 1
Shown in referring to Fig.1, the embodiment of the present invention provides a kind of cirrus detection method, comprising the following steps:
S101, input remote sensing images, and obtain the primary channel grayscale image of remote sensing images；
S102, the Cancers Fractional Dimension Feature figure for obtaining primary channel grayscale image；
S103, weight grayscale image is obtained based on Cancers Fractional Dimension Feature figure and primary channel grayscale image；
S104, cluster calculation is carried out to the pixel of primary channel grayscale image, the cluster result of pixel is obtained, based on poly-
Class result determines the predistribution label of each pixel of channel grayscale image；
S105, it establishes figure based on primary channel grayscale image, weight grayscale image and predistribution label and cuts model；
S106, model progress minimal cut calculating is cut to figure, export the testing result of remote sensing images.
Since detection cirrus is mainly derived from the remote sensing images of satellite or spacecraft transmission, the prior art is usual
The method detected using spectral signature is extracted from remote sensing images, and the embodiment of the present invention is calculated using point shapes such as cladding process
Method is grabbed for the atypical characteristic of cirrus, and the Cancers Fractional Dimension Feature by acquiring cirrus is realized to cirrus and other height
Effective differentiation of radiating object；Further, in the surface fractal feature based on cirrus and grayscale image each pixel cluster feelings
Condition, which is established, cuts model for the figure of cirrus detection, and the detection knot for calculating available remote sensing images saliency cirrus is cut by figure
Fruit, this calculating process are not necessarily to excessive sample size, machine learning method can not be applicable in when solving limited sample size
Problem, compared to the conventional threshold values method more efficiently and accurately that can not effectively distinguish cirrus feature.
Embodiment 2
Referring to shown in Fig. 2 to Fig. 7, the embodiment of the present invention advanced optimizes on that basis of example 1, provides a kind of cirrus inspection
Survey method, includes the following steps:
S201, input remote sensing images, and obtain the primary channel grayscale image of remote sensing images.
Remote sensing images in the embodiment of the present invention as shown in Figure 2, it is possible to understand that ground, the image are really coloured picture.Due to coloured silk
Chromatic graph picture includes to contain much information, and processing speed is slower, is unfavorable for digital processing, therefore, by colour in the embodiment of the present invention
Remote sensing images are converted into the channel grayscale image of one of primary colors of red green blue tricolor (Red&Green&Blue, RGB).
Preferably, as shown in figure 3, be in the embodiment of the present invention convert remote sensing images to blue channel grayscale image be used for into
The processing of one step.Since the attribute of cirrus is white, green channel or red channel gray scale are converted into compared to by remote sensing images
Figure, blue channel grayscale image is more preferable for the display effect on cirrus boundary, and discrimination is higher between pixel.
Specifically, some softwares or function library, such as openCV, Mathematica, matlab can be used to remote sensing
Image is handled, and primary channel grayscale image is extracted, for the concrete application process of these softwares or function library, herein not
It repeats again.
Alternatively it is also possible to convert red or green channel grayscale image for remote sensing images, although can not reach with
The same display effect of blue channel, but same technical effect can also be reached based on the promotion of subsequent algorithm, for red
The acquisition methods of chrominance channel grayscale image and green channel grayscale image, details are not described herein again.
It can be seen that being conducive to carry out remote sensing images subsequent by converting primary channel grayscale image for remote sensing images
Processing, and for the identification of cirrus have good effect.
S202, the FRACTAL SURFACES surface area S (n) in primary channel grayscale image at each pixel is calculated using cladding process,
Wherein, n indicates unit area scale.
It is to be appreciated that function f (i, j) can be set based on primary channel grayscale image here, wherein f indicates gray value,
(i, j) indicates location of pixels.When using cladding process, gray level image is imagined as a FRACTAL SURFACES in three dimensions, and
Assuming that the point that the blanket for the use of thickness in monolayer being n is n to Distance surfaces all in three-dimensional space surface covers, if covering is bent
Face upper surface is Un(i, j), lower surface Dn(i, j), and initial value has U0(i, j)=D0(i, j)=f (i, j), then on blanket
The calculation formula of surface and lower surface is as follows:
Wherein, (p, q) indicates location of pixels of the distance less than 1 with pixel (i, j), and max expression is maximized, min table
Show and is minimized.
It can be obtained by formula (1) and (2), blanket volume νnCalculation formula are as follows:
It can be obtained by formula (3), the calculation formula of blanket surface product are as follows:
S203, the fractal dimension d that each pixel is calculated based on FRACTAL SURFACES surface area S (n) and fractal surface formula；Its
In, fractal surface formula is S (n)=n2-d。
It can be obtained by formula (4) and fractal surface formula, the calculation formula of imaging surface fractal dimension d are as follows:
Wherein, In expression takes natural logrithm.
Optionally, the imaging surface FRACTAL DIMENSION of each pixel in primary channel grayscale image can also be obtained by other algorithms
Number d, other specific algorithms may include: size method, slit island method, Box-counting technique, structure function method, semivariance method and transformation
Method etc..For calculating the detailed process of imaging surface fractal dimension d using other methods, details are not described herein again.
S204, the fractal dimension d based on each pixel obtain the Cancers Fractional Dimension Feature figure of primary channel grayscale image.
It is to be appreciated that fractal dimension is able to reflect the spacial validity of complex object, it is that complex object is irregular
The measure of property.Specifically, in the embodiment of the present invention, fractal dimension can embody in primary channel grayscale image and correspond to cirrus
Pixel in image space occupy rate and filling rate.
Further, specific output effect may refer to Cancers Fractional Dimension Feature figure shown in fig. 5.It is apparent that due to cirrus
Attributive character with random multi-layer, therefore the corresponding pixel fractal dimension of cirrus is higher, and background objects pixel divides shape
Dimension is very low.By in this, cirrus and background objects can be distinguished by Cancers Fractional Dimension Feature figure.
S205, the weighting that is multiplied is carried out using Cancers Fractional Dimension Feature figure as weight with primary channel grayscale image, based on weighting knot
Fruit obtains weight grayscale image.
It is to be appreciated that in this step using the Cancers Fractional Dimension Feature of each pixel as weight to primary channel grayscale image
It is weighted, calculation formula is as follows:
B (i, j)=d (i, j) × f (i, j) ... ... ... ... ... ... ... ... .. (6)
Wherein, f is gray value, and d is Cancers Fractional Dimension Feature, and B is the weight gray value being calculated.
It is to be appreciated that the corresponding pixel fractal dimension of cirrus is higher in image, by making since structure is complicated for cirrus
After being weighted with fractal dimension to primary channel grayscale image, pixel corresponding to cirrus in image can be effectively highlighted.Tool
Body, output effect may refer to weight grayscale image shown in fig. 6.
S206, initial value of the K pixel as cluster centre is randomly selected in primary channel grayscale image, calculate primary colors
Other pixels in addition to cluster centre and the Euclidean distance between cluster centre in the grayscale image of channel.
It is to be appreciated that in the embodiment of the present invention using K mean algorithm to each pixel in primary channel grayscale image into
Row cluster.The center of a cluster can be used to represent a cluster as a kind of partition clustering algorithm in K mean algorithm, and
Initial center is first chosen at random in iterative process, which is not necessarily the point in cluster, and then passing through continuous iteration can be with
Determine final cluster centre, the algorithm is mainly for the treatment of numeric type data.And the iterative process of this method cluster, it is main logical
Cross the Euclidean distance calculated between other positions and center.Specifically, the calculation formula of Euclidean distance is as follows:
Wherein, JkIndicate the distance between any pixel and k-th cluster centre, xi,jIndicate that the two dimension of location of pixels is sat
Mark, ckIndicate k-th cluster centre.After obtaining the Euclidean distance of each pixel and cluster centre, according to minimum distance standard
Then pixel is sorted out, that is, distribute pixel to apart from the nearest corresponding class of cluster centre.
S207, all pixels point is clustered based on Euclidean distance, and obtained equal with Euclidean distance in each cluster
It is worth immediate pixel as new cluster centre and is iterated calculating, final gather is obtained when cluster centre no longer changes
Class result；
It is to be appreciated that the initial center of cluster centre due to randomly selecting, is worked as it is possible that being not present in it
Preceding corresponding cluster.It should be evident that this risk can not be eliminated by only carrying out primary cluster, and by each cluster result
In choose cluster centre again and be iterated operation, until cluster centre no longer changes, by this, may be implemented to cloud or background
The corresponding pixel of object is subdivided into corresponding class.
S208, the predistribution label that each pixel in primary channel grayscale image is determined based on final cluster result.
It is to be appreciated that in embodiments of the present invention, K mean algorithm is as a kind of machine learning method, not as existing
For carrying out sample training and detection to cirrus in technology, but for pixel each in grayscale image make it is preset sort out to
Preset label value can be assigned to each pixel；Specifically all pixels in figure are divided into two class of cloud and background objects, and
Corresponding label value is set, is such as 1 by pixel setting label value corresponding with cloud, label is arranged in pixel corresponding with background objects
Value is 0.By this clustering method, the conversion by pixel to label numerical value is realized, subsequent calculation process is convenient to, thus
Improve detection efficiency.
Optionally, the embodiment of the present invention can also use other clustering algorithms to cluster the pixel in grayscale image, such as
Mean shift clustering algorithm, density clustering algorithm etc..For using the specific implementation process of other types clustering algorithm, herein not
It repeats again.
S209, it determines that figure cuts the criterion function of model, and is based on primary channel grayscale image, weight grayscale image and predistribution mark
Sign the parameters for determining criterion function.
It is to be appreciated that the embodiment of the present invention is expected that by and establishes figure and cut model realization to valid pixel in image, that is, roll up
The detection of cloud respective pixel.
Specifically, it is the criterion function foundation that algorithm (graph-cut) is cut based on figure that figure, which cuts model, and the criterion function is such as
Under:
E (L)=a*R (L)+B (L) ... ... ... ... ... ... ... ... .. (8)
Wherein, L is the calculating initial value that figure cuts model, and R (L) is the area item that figure cuts model, and B (L) is that figure cuts model
Border item, a are preset balance parameters, and E (L) is that figure cuts value.
It is to be appreciated that in criterion function, area item is a kind of priori penalty term, and border item is phase between a kind of region
Like degree penalty term；Balance parameters cut figure the disturbance degree of value for equilibrium region item and border item, for example, saying if a takes 0
Bright consideration border item factor, does not consider equilibrium region item factor.The purpose that figure cuts algorithm is to solve between source point and meeting point
Minimal cut path, i.e. realization figure cuts the maximum result of value.
It specifically, is the area item that criterion function is determined based on primary channel grayscale image in the embodiment of the present invention；Based on power
Value grayscale image determines the border item of criterion function；The calculating initial value of criterion function is determined based on predistribution label.
It is to be appreciated that the pixel based on primary channel grayscale image can do just region occupied by cloud and background objects
The region division of step, therefore it is suitable as the area item parameter of criterion function；And weight grayscale image medium cloud is corresponding with background objects
Pixel is adapted as criterion function due to weight difference, apparent division existing for the boundary of cloud and background objects
Border item parameter；And cloud and the corresponding pixel of background objects can be carried out to the assignment of different predistribution labels by clustering algorithm,
Therefore the calculating initial value being suitable as in criterion function.
Specifically, it in the embodiment of the present invention, is directly affected since area item has the calculated result that figure is cut, a can
The positive integer greater than 0 is taken, is learnt through experiment test, the preferred value of a is 100.
It is to be appreciated that can establish figure after being determined that figure cuts the parameters of model criterion function and cut model simultaneously
It is further processed.
S210, model progress minimal cut calculating is cut to figure, calculated result is exported into the testing result for remote sensing images.
It is to be appreciated that after establishing figure and cutting model, figure can be used cuts algorithm acquisition figure and cut result.
Preferably, the embodiment of the present invention calculates the minimal cut that the figure cuts model using Ford-Fulkerson labeling algorithm
Value.
It is to be appreciated that the purpose that figure cuts algorithm, which is to obtain, meets Directed Graph Model G=<V, E>max-flow.
Specifically, which has unique source point S as starting point, i.e., in-degree is 0, while having unique meeting point
T is 0 as end point, i.e. out-degree.When algorithm starts, all vertex u, v ∈ is enabled to flow when V has f (u, v)=0, i.e. original state
Value be 0, increase flow valuve by finding the augmenting path in model later.Here augmenting path can be regarded as from source point
A paths between S to meeting point T can be such that the flow of network increases, to increase the value of stream along the path.Pass through iteration reality
It tests, calculates and stop when augmenting paths all in model are all found, obtain the minimal cut (or being max-flow) of model at this time
As testing result.
It is to be appreciated that cutting the minimal cut of model by acquisition figure as a result, output picture corresponding with cirrus can be maximized
Element, to reach the testing result of output cirrus.Specifically, output effect may refer to the detection of remote sensing images shown in Fig. 7
Result figure.
Optionally, other kinds of figure can also be used to cut algorithm and obtain testing result, including Goldberg-Tarjan is calculated
Method and Edmonds-Karp algorithm etc., details are not described herein again.
It is to be appreciated that cutting the specific implementation of algorithm for figure in the embodiment of the present invention, Python, OpenCV can be used
With the compilation tools such as Matlab, details are not described herein again.
Through testing, the accuracy rate that cirrus detects can be promoted to 95% or more using cirrus detection method of the invention, called together
The rate of returning is promoted to 90%.
Embodiment 3
Referring to shown in Fig. 8, a kind of specific hardware structure for cirrus detection device that the embodiment of the present invention 3 provides, the cirrus
Detection device 8 may include: memory 82 and processor 83；Various components are coupled by communication bus 81.It is understood that
Communication bus 81 is for realizing the connection communication between these components.Communication bus 81 except include data/address bus in addition to, further include
Power bus, control bus and status signal bus in addition.But for the sake of clear explanation, various buses are all designated as in fig. 8
Communication bus 81.
Memory 82, for storing the localization method program that can be run on processor 83；
Processor 83, for executing following steps when running localization method program:
Step 1, input remote sensing images, and obtain the primary channel grayscale image of remote sensing images；
Step 2, the Cancers Fractional Dimension Feature figure for obtaining primary channel grayscale image；
Step 3 obtains weight grayscale image based on Cancers Fractional Dimension Feature figure and primary channel grayscale image；
Step 4 carries out cluster calculation to the pixel of primary channel grayscale image, obtains the cluster result of pixel, is based on
Cluster result determines the predistribution label of each pixel of primary channel grayscale image；
Step 5 is established figure and is cut model based on primary channel grayscale image, weight grayscale image and predistribution label；
Step 6 cuts model progress minimal cut calculating to figure, exports the testing result of remote sensing images.
Further, step 2 specifically includes:
FRACTAL SURFACES surface area S (n) in step 2.1, the calculating primary channel grayscale image at each pixel,
In, n indicates unit area scale；
Step 2.2, the fractal dimension that each pixel is calculated based on FRACTAL SURFACES surface area S (n) and fractal surface formula
d；Wherein, the fractal surface formula is S (n)=n2-d；
Step 2.3, the fractal dimension d based on each pixel obtain the Cancers Fractional Dimension Feature of the primary channel grayscale image
Figure.
Further, step 3 specifically includes:
The weighting that is multiplied is carried out using the Cancers Fractional Dimension Feature figure as weight with the primary channel grayscale image, based on weighting
As a result weight grayscale image is obtained.
Further, step 4 specifically includes:
Step 4.1 randomly selects initial value of the K pixel as cluster centre in the primary channel grayscale image,
Calculate the Europe between other pixels and the cluster centre in the primary channel grayscale image in addition to the cluster centre
Formula distance；
Step 4.2 clusters all pixels point based on the Euclidean distance, and obtain in each cluster with it is European away from
From the immediate pixel of mean value be iterated calculating as new cluster centre, when cluster centre no longer changes obtain most
Whole cluster result；
Step 4.3, the predistribution that each pixel in the primary channel grayscale image is determined based on final cluster result
Label.
Further, step 5 specifically includes:
Step 5.1 determines that the figure cuts the criterion function of model；
Step 5.2, the area item that the criterion function is determined based on the primary channel grayscale image；Based on the weight ash
Degree figure determines the border item of the criterion function；The calculating initial value of the criterion function is determined based on the predistribution label；
Step 5.3 establishes the figure based on the criterion function and cuts model.
Further, the criterion function that figure cuts model specifically includes:
E (L)=a*R (L)+B (L)；Wherein, L is the calculating initial value of the criterion function, and R (L) is the criterion function
Area item, B (L) be the criterion function border item, a be preset balance parameters, E (L) be figure cut value.
Further, model is cut to figure and carries out minimal cut calculating, specifically included:
The minimal cut value that the figure cuts model is calculated using Ford-Fulkerson labeling algorithm.
Further, the primary channel grayscale image is blue channel grayscale image.
It is appreciated that the memory 82 in the embodiment of the present invention can be volatile memory or nonvolatile memory,
It or may include both volatile and non-volatile memories.Wherein, nonvolatile memory can be read-only memory (Read-
Only Memory, ROM), programmable read only memory (Programmable ROM, PROM), the read-only storage of erasable programmable
Device (Erasable PROM, EPROM), electrically erasable programmable read-only memory (Electrically EPROM, EEPROM) or
Flash memory.Volatile memory can be random access memory (Random Access Memory, RAM), be used as external high
Speed caching.By exemplary but be not restricted explanation, the RAM of many forms is available, such as static random access memory
(Static RAM, SRAM), dynamic random access memory (Dynamic RAM, DRAM), Synchronous Dynamic Random Access Memory
(Synchronous DRAM, SDRAM), double data speed synchronous dynamic RAM (Double Data Rate
SDRAM, DDRSDRAM), enhanced Synchronous Dynamic Random Access Memory (Enhanced SDRAM, ESDRAM), synchronized links
Dynamic random access memory (Synchlink DRAM, SLDRAM) and direct rambus random access memory (Direct
Rambus RAM, DRRAM).The memory 82 of system and method described herein is intended to include but is not limited to these and arbitrarily its
It is suitble to the memory of type.
And processor 83 may be a kind of IC chip, the processing capacity with signal.It is above-mentioned during realization
Each step of method can be completed by the integrated logic circuit of the hardware in processor 83 or the instruction of software form.It is above-mentioned
Processor 83 can be general processor, digital signal processor (Digital Signal Processor, DSP), dedicated
Integrated circuit (Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components.It may be implemented or execute disclosed each method, step and the logic diagram in the embodiment of the present invention.It is general
Processor can be microprocessor or the processor is also possible to any conventional processor etc..In conjunction with institute of the embodiment of the present invention
The step of disclosed method, can be embodied directly in hardware decoding processor and execute completion, or with the hardware in decoding processor
And software module combination executes completion.Software module can be located at random access memory, and flash memory, read-only memory may be programmed read-only
In the storage medium of this fields such as memory or electrically erasable programmable memory, register maturation.The storage medium is located at
The step of memory 82, processor 83 reads the information in memory 82, completes the above method in conjunction with its hardware.
Based on previous embodiment, the embodiment of the present invention provides a kind of computer-readable medium, which deposits
Cirrus detection program is contained, cirrus detects when program is executed by least one processor and realizes positioning side in any of the above-described embodiment
The step of method.
It is to be appreciated that the method and step in above embodiments, can store in computer-readable storage medium, base
In such understanding, the technical solution of the embodiment of the present invention substantially the part that contributes to existing technology or should in other words
The all or part of technical solution can be embodied in the form of software products, which is stored in one and deposits
In storage media, including some instructions are used so that a computer equipment (can be personal computer, server or network
Equipment etc.) or processor (processor) execute present invention method all or part of the steps.And storage above-mentioned is situated between
Matter include: USB flash disk, mobile hard disk, read-only memory (ROM, Read Only Memory), random access memory (RAM,
Random Access Memory), the various media that can store program code such as magnetic or disk.
It is understood that embodiments described herein can with hardware, software, firmware, middleware, microcode or its
Combination is to realize.For hardware realization, processing unit be may be implemented in one or more specific integrated circuit (Application
Specific Integrated Circuits, ASIC), digital signal processor (Digital Signal Processing,
DSP), digital signal processing appts (DSP Device, DSPD), programmable logic device (Programmable Logic
Device, PLD), field programmable gate array (Field-Programmable Gate Array, FPGA), general processor,
In controller, microcontroller, microprocessor, other electronic units for executing the application function or combinations thereof.
For software implementations, the skill of this paper can be realized by executing the module (such as process, function etc.) of this paper function
Art.Software code is storable in memory and is executed by processor.Memory can in the processor or outside the processor
It realizes in portion.
Specifically, when the processor 83 in user terminal is additionally configured to operation computer program, previous embodiment is executed
In method and step, be not discussed here.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, the shape of hardware embodiment, software implementation or embodiment combining software and hardware aspects can be used in the present invention
Formula.Moreover, the present invention, which can be used, can use storage in the computer that one or more wherein includes computer usable program code
The form for the computer program product implemented on medium (including but not limited to magnetic disk storage and optical memory etc.).
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
It should be understood that between technical solution documented by the embodiment of the present invention, in the absence of conflict, Ke Yiren
Meaning combination.
Above embodiments, only presently preferred embodiments of the present invention, are not intended to limit the scope of the present invention, all
Made any modifications, equivalent replacements, and improvements etc. within the spirit and principles in the present invention should be included in guarantor of the invention
Within the scope of shield.
Claims (10)
1. a kind of cirrus detection method, which comprises the following steps:
Step 1, input remote sensing images, and obtain the primary channel grayscale image of remote sensing images；
Step 2, the Cancers Fractional Dimension Feature figure for obtaining primary channel grayscale image；
Step 3 obtains weight grayscale image based on Cancers Fractional Dimension Feature figure and primary channel grayscale image；
Step 4 carries out cluster calculation to the pixel of primary channel grayscale image, the cluster result of pixel is obtained, based on cluster
As a result the predistribution label of each pixel of primary channel grayscale image is determined；
Step 5 is established figure and is cut model based on primary channel grayscale image, weight grayscale image and predistribution label；
Step 6 cuts model progress minimal cut calculating to figure, exports the testing result of remote sensing images.
2. cirrus detection method according to claim 1, which is characterized in that the step 2 specifically includes:
FRACTAL SURFACES surface area S (n) in step 2.1, the calculating primary channel grayscale image at each pixel, wherein n table
Show unit area scale；
Step 2.2, the fractal dimension d that each pixel is calculated based on FRACTAL SURFACES surface area S (n) and fractal surface formula；Its
In, the fractal surface formula is S (n)=n2-d；
Step 2.3, the fractal dimension d based on each pixel obtain the Cancers Fractional Dimension Feature figure of the primary channel grayscale image.
3. cirrus detection method according to claim 1, which is characterized in that the step 3 specifically includes:
The weighting that is multiplied is carried out using the Cancers Fractional Dimension Feature figure as weight with the primary channel grayscale image, is based on weighted results
Obtain weight grayscale image.
4. cirrus detection method according to claim 1, which is characterized in that the step 4 specifically includes:
Step 4.1 randomly selects initial value of the K pixel as cluster centre in the primary channel grayscale image, calculates
Between other pixels and the cluster centre in the primary channel grayscale image in addition to the cluster centre it is European away from
From；
Step 4.2 clusters all pixels point based on the Euclidean distance, and obtain in each cluster with Euclidean distance
The immediate pixel of mean value is iterated calculating as new cluster centre, obtains finally when cluster centre no longer changes
Cluster result；
Step 4.3, the predistribution label that each pixel in the primary channel grayscale image is determined based on final cluster result.
5. cirrus detection method according to claim 1, which is characterized in that the step 5 specifically includes:
Step 5.1 determines that the figure cuts the criterion function of model；
Step 5.2, the area item that the criterion function is determined based on the primary channel grayscale image；Based on the weight grayscale image
Determine the border item of the criterion function；The calculating initial value of the criterion function is determined based on the predistribution label；
Step 5.3 establishes the figure based on the criterion function and cuts model.
6. cirrus detection method according to claim 5, which is characterized in that the criterion function that the figure cuts model specifically wraps
It includes:
E (L)=a*R (L)+B (L)；Wherein, L is the calculating initial value of the criterion function, and R (L) is the area of the criterion function
Domain, B (L) are the border item of the criterion function, and a is preset balance parameters, and E (L) is that figure cuts value.
7. cirrus detection method according to claim 1, which is characterized in that described pair of figure cuts model and carry out minimal cut meter
It calculates, specifically includes:
The minimal cut value that the figure cuts model is calculated using Ford-Fulkerson labeling algorithm.
8. cirrus detection method described in -7 according to claim 1, which is characterized in that the primary channel grayscale image is that blue is logical
Road grayscale image.
9. a kind of cirrus detection device characterized by comprising
Processor, memory and communication bus；
Wherein, the communication bus, for realizing the communication connection between the processor and the memory；
The memory detects program for storing the cirrus that can be run on the processor；
The processor, is used for:
Remote sensing images are inputted, and obtain the primary channel grayscale image of remote sensing images；
Obtain the Cancers Fractional Dimension Feature figure of primary channel grayscale image；
Weight grayscale image is obtained based on Cancers Fractional Dimension Feature figure and primary channel grayscale image；
Cluster calculation is carried out to the pixel of primary channel grayscale image, obtains the cluster result of pixel, it is true based on cluster result
Determine the predistribution label of pixel；
Determine that figure cuts model based on primary channel grayscale image, weight grayscale image and predistribution label；
It cuts model to figure to calculate, and the testing result based on calculated result output remote sensing images.
10. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage have one or
Multiple programs, one or more of programs can be executed by one or more processor, to realize such as claim 1 to 8
Any one of described in cirrus detection method the step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910129005.7A CN109886193B (en) | 2019-02-21 | 2019-02-21 | A kind of cirrus cloud detection method, device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910129005.7A CN109886193B (en) | 2019-02-21 | 2019-02-21 | A kind of cirrus cloud detection method, device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109886193A true CN109886193A (en) | 2019-06-14 |
| CN109886193B CN109886193B (en) | 2020-11-20 |
Family
ID=66928743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910129005.7A Active CN109886193B (en) | 2019-02-21 | 2019-02-21 | A kind of cirrus cloud detection method, device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886193B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114022790A (en) * | 2022-01-10 | 2022-02-08 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
| CN114187320A (en) * | 2021-12-14 | 2022-03-15 | 北京柏惠维康科技有限公司 | Spine CT image segmentation method and spine imaging identification method and device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7747625B2 (en) * | 2003-07-31 | 2010-06-29 | Hewlett-Packard Development Company, L.P. | Organizing a collection of objects |
| CN102831607A (en) * | 2012-08-08 | 2012-12-19 | 深圳市迈科龙生物技术有限公司 | Method for segmenting cervix uteri liquid base cell image |
| CN103886614A (en) * | 2014-04-14 | 2014-06-25 | 重庆威堪科技有限公司 | Image edge detection method based on network node fractal dimensions |
| US8885925B2 (en) * | 2013-03-12 | 2014-11-11 | Harris Corporation | Method for 3D object identification and pose detection using phase congruency and fractal analysis |
| CN105631903A (en) * | 2015-12-24 | 2016-06-01 | 河海大学 | Remote sensing image water extraction method and device based on RGBW characteristic space diagram cutting algorithm |
| CN106228553A (en) * | 2016-07-20 | 2016-12-14 | 湖南大学 | High-resolution remote sensing image shadow Detection apparatus and method |
| CN108021890A (en) * | 2017-12-05 | 2018-05-11 | 武汉大学 | A kind of high score remote sensing image harbour detection method based on PLSA and BOW |
| CN108647658A (en) * | 2018-05-16 | 2018-10-12 | 电子科技大学 | A kind of infrared imaging detection method of high-altitude cirrus |
-
2019
- 2019-02-21 CN CN201910129005.7A patent/CN109886193B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7747625B2 (en) * | 2003-07-31 | 2010-06-29 | Hewlett-Packard Development Company, L.P. | Organizing a collection of objects |
| CN102831607A (en) * | 2012-08-08 | 2012-12-19 | 深圳市迈科龙生物技术有限公司 | Method for segmenting cervix uteri liquid base cell image |
| US8885925B2 (en) * | 2013-03-12 | 2014-11-11 | Harris Corporation | Method for 3D object identification and pose detection using phase congruency and fractal analysis |
| CN103886614A (en) * | 2014-04-14 | 2014-06-25 | 重庆威堪科技有限公司 | Image edge detection method based on network node fractal dimensions |
| CN105631903A (en) * | 2015-12-24 | 2016-06-01 | 河海大学 | Remote sensing image water extraction method and device based on RGBW characteristic space diagram cutting algorithm |
| CN106228553A (en) * | 2016-07-20 | 2016-12-14 | 湖南大学 | High-resolution remote sensing image shadow Detection apparatus and method |
| CN108021890A (en) * | 2017-12-05 | 2018-05-11 | 武汉大学 | A kind of high score remote sensing image harbour detection method based on PLSA and BOW |
| CN108647658A (en) * | 2018-05-16 | 2018-10-12 | 电子科技大学 | A kind of infrared imaging detection method of high-altitude cirrus |
Non-Patent Citations (3)
| Title |
|---|
| CHEN JUN 等: "Automatic target segmentation by improved Grabcut based on fractal", 《COMPUTER ENGINEERING AND APPLICATIONS》 * |
| 张东衡 等: "一种气液两相流气相参数图像检测方法", 《计算机测量与控制》 * |
| 菅帅: "基于粗糙表面的红外低发射率涂层发射率建模研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114187320A (en) * | 2021-12-14 | 2022-03-15 | 北京柏惠维康科技有限公司 | Spine CT image segmentation method and spine imaging identification method and device |
| CN114187320B (en) * | 2021-12-14 | 2022-11-08 | 北京柏惠维康科技股份有限公司 | Spine CT image segmentation method and spine imaging identification method and device |
| CN114022790A (en) * | 2022-01-10 | 2022-02-08 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109886193B (en) | 2020-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472623B (en) | Image detection method, device and system | |
| Wei et al. | Toward automatic building footprint delineation from aerial images using CNN and regularization | |
| CN107767382B (en) | The extraction method and system of static three-dimensional map contour of building line | |
| CN106952402B (en) | A kind of data processing method and device | |
| CN105913070B (en) | A multi-cue saliency extraction method based on light field camera | |
| CN101506840B (en) | Color Classification Method Based on Color Image Code | |
| CN105809651B (en) | Image saliency detection method based on edge dissimilarity contrast | |
| CN108460389A (en) | A kind of the type prediction method, apparatus and electronic equipment of identification objects in images | |
| US9025863B2 (en) | Depth camera system with machine learning for recognition of patches within a structured light pattern | |
| CN110991435A (en) | A method and device for locating key information of express waybill based on deep learning | |
| CN109977949A (en) | Text positioning method, device, computer equipment and the storage medium of frame fine tuning | |
| CN116664954B (en) | Hyperspectral ground object classification method based on graph convolution and convolution fusion | |
| CN107622239B (en) | A Method for Detection of Specified Building Areas in Remote Sensing Images Constrained by Hierarchical Local Structure | |
| CN111274964B (en) | Detection method for analyzing water surface pollutants based on visual saliency of unmanned aerial vehicle | |
| CN107886512A (en) | A kind of method for determining training sample | |
| CN116503388A (en) | Defect detection method, device and storage medium | |
| CN107832359A (en) | A kind of picture retrieval method and system | |
| CN112766361A (en) | Target fruit detection method and detection system under homochromatic background | |
| CN109886193A (en) | A kind of cirrus cloud detection method, device and computer readable storage medium | |
| CN105205054A (en) | Method for displaying pictures and method and device for acquiring hue characteristic values of pictures | |
| Yuan et al. | Single‐image shadow detection and removal using local colour constancy computation | |
| CN114745532A (en) | White balance processing method and device for mixed color temperature scene, storage medium and terminal | |
| CN116229205A (en) | Lipstick product surface defect data augmentation method based on small sample characteristic migration | |
| CN113139983A (en) | Human image segmentation method and device based on RGBD | |
| CN107452003A (en) | A kind of method and device of the image segmentation containing depth information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |