CN108830844A - A kind of facilities vegetable extracting method based on multidate high-resolution remote sensing image - Google Patents
A kind of facilities vegetable extracting method based on multidate high-resolution remote sensing image Download PDFInfo
- Publication number
- CN108830844A CN108830844A CN201810592833.XA CN201810592833A CN108830844A CN 108830844 A CN108830844 A CN 108830844A CN 201810592833 A CN201810592833 A CN 201810592833A CN 108830844 A CN108830844 A CN 108830844A
- Authority
- CN
- China
- Prior art keywords
- image
- remote sensing
- facilities vegetable
- resolution remote
- sensing image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30128—Food products
Abstract
The present invention provides a kind of facilities vegetable extracting methods based on multidate high-resolution remote sensing image, including：High-resolution remote sensing image is pre-processed, to extract single band image；Image enhancement processing is executed to single band image, to improve the separating capacity of facilities vegetable Yu other atural object classifications；Predetermined watch window is selected, texture analysis processing is carried out to enhancing treated image, obtains characteristic value texture image to calculate；Based on characteristic value texture image, the mask image of the building and road easily obscured with facilities vegetable is created；Exposure mask is carried out to original high-resolution remote sensing image using mask image, to extract the edge detection line of atural object figure spot；Mathematical morphology operation and thin dividing processing are carried out to edge detection line, to generate extraction result；Binary conversion treatment is executed to the gray level image of facilities vegetable, the vector figure spot of facilities vegetable is extracted and carries out vector and turnstile lattice processing, to realize facilities vegetable information extraction.
Description
Technical field
The invention belongs to image intelligent processing technology fields, are related to the information extraction technology of high-resolution remote sensing image, tool
Body is related to a kind of facilities vegetable extracting method based on multidate high-resolution remote sensing image, for being based on high-definition remote sensing
Image extracts the facilities vegetable information in image.
Background technique
With quickly propelling for China's urbanization, the demand to vegetables also increases increasingly, and intensive vegetables production is (with facility
Vegetables are representative) have become the main direction of development that vegetables produce.Facilities vegetable can extend the vegetable supply period, improve
Yield of vegetables alleviates phenomenon in short supply.The common industrialized agriculture type in the country mainly includes at present：Vinyl house is (containing medium and small
Arched shed), heliogreenhouse and attached-greenhouse.Vinyl house refers to the single arched shed using plastic film as glazing material.?
Southern region of China is Winter protection, summer sunshade, rain-proof using function.And use in northern area and mainly do sth. in advance with the spring,
Autumn delays effect, generally do sth. in advance than Open air product section or prolong the latter moon or so.Since its thermal insulation property is poor, on northern ground
Area does not have to it generally and does overwintering production.
Heliogreenhouse is that China scientific worker constantly improve one that improves and developed on the basis of the greenhouse of Yimianpu
Kind greenhouse form with Chinese characteristics.It is using solar energy as main energy sources, and night is protected using moving thermal insulation in front roof
Temperature carries out the single layer face Plastic film greenhouse of overwintering production, and east, west, three face wall body of north and the rear roofing in such greenhouse are using high
Heat-insulating construction material.It is used in the northern area of China, manually heating can not keep indoor/outdoor temperature-difference to reach under normal condition
20-30 DEG C or more, such greenhouse is widely used in 30-45 ° of area of north latitude, is that the overwintering production gardening of northern area produces
The Main Greenhouse form of product.Attached-greenhouse, which refers to, produces temperature by the large area that gutter connects for the greenhouse of multiple single spans
Room is the trend and trend of the world today and China's Development of Morden industrialized agriculture.Attached-greenhouse is according to structure type and covering
The difference of material is divided into：Even glasshouse, muiti-span greenhouse and polycarbonate plate greenhouse (greenhouse PC).Above-mentioned three kinds
Greenhouse can simply be divided into again：" cold canopy " (vinyl house) and " brooder " (heliogreenhouse and attached-greenhouse).
Protected Production of Vegetables has become one of important component of modern agriculture, and facility cultivation is at by traditional agriculture
It transform a kind of important means of modern agriculture as, and significantly increases agriculture at local pillar industry in many areas
The income of the people.In addition, the appearance of facilities vegetable so that the utilization rate in soil is high, is got rid of to a certain extent to natural environment
Dependence, have the characteristics that high investment, high technology content, high-quality, high yield, high benefit, being that most great-hearted agricultural is new produces
Industry.The construction area of facilities vegetable reflects local agricultural modernization development level, also reflects local vegetable supply energy
Power, plays vital influence for the supply-demand relationship of vegetables in the market, scientific management, vegetable for local agricultural sector
Colza plants policy making etc. and all has very important significance.Wherein, important one of subsidies for growing superior grain cultivators is exactly the face of subsidies for growing superior grain cultivators
Product, the accuracy of area are related to the vital interests of peasant.
In the prior art, the information such as area output of facilities vegetable are usually to use conventional ground investigation method or normal
The statistical data in year lacks science, wastes a large amount of time and manpower, be affected by artificial subjective factor, to government
Decision, management are difficult to provide reliability foundation, are not able to satisfy its demand.Remote sensing receives atural object electricity as long-range detection atural object
The technology of magnetic wave information characteristic there is information truth not available for conventional traditional technology to enrich, Up-to-date state, broad perspectives, dynamic
The many advantages such as property.The information such as area, the land use distribution of facilities vegetable are quickly and accurately obtained by remote sensing technology, it can
Think the distribution of reasonable Arrangement vegetable-growing area, realize intensive vegetables production, stablize and promote agricultural development horizontal, realizes agricultural resource
Efficient and sustainable use establish good basis.
The fast development of remote sensing technology, high-resolution remote sensing image, which has become, obtains the main of facilities vegetable cultivated area
Approach.How area, the type etc. of facilities vegetable efficiently, are accurately extracted from magnanimity high-definition remote sensing image data
Information is high-resolution remote sensing image intelligently one of key technology urgently to be solved in interpretation.Meanwhile facilities vegetable is remote sensing shadow
Typical, very common also critically important atural object element type, effective acquisition of information, in geographic data updates, rural area as in
The fields such as development plan, agricultural sciences management have great importance.
The diversified development of remote sensing platform, the spatially and spectrally promotion of resolution ratio, application of the remote sensing technology in agriculture field
Become more and more extensive.Carrying out atural object category classification for remote sensing image is mainly differentiation circle for determining differently species other
Face and criterion, at present in facilities vegetable extracting method, being all many times by with supervised classification and non-supervisory point
Class is algorithm, is modified using visual interpretation to improve resolution ratio.Wherein, visual interpretation can comprehensively utilize the color of atural object
The knowledge of the image features such as tune or color, shape, size, shade, texture, pattern, position and layout, but visual interpretation needs
Interpretation personnel are familiar to research area, and it is very time-consuming and laborious to classify.The method of supervised classification and unsupervised classification is due to mistake
Divide and rely on object spectrum information, cannot sufficiently excavate atural object space characteristics and other auxiliary informations, it is difficult to overcome in image
" the different spectrum of jljl " and " same object different images " phenomenon, nicety of grading are subject to certain restrictions, and need to carry out to reach higher precision
A large amount of later period amendment.Furthermore existing method further includes：The sorting algorithm of object-oriented, artificial neural network method, support to
Amount machine method, extracting method based on space structure etc..The presence for having building, road in the result of facilities vegetable is extracted at present,
This largely will cause the low precision of extraction, and it is big to obscure range.
In addition, current existing method majority is using in Landsat TM, SPOT5, RapidEye etc. from data source
Low resolution remote sensing image, is not able to satisfy fine facilities vegetable information extraction requirement, and the high-definition remote sensing that part proposes is set
It applies the extracting method of vegetables and is often affected by the quality of image, scene complexity, and need a large amount of manual intervention,
It reduce the universality of method and the degree of automation.Therefore the accurate extraction of facilities vegetable, the height that must just use to be realized
Resolution procedure remote sensing image.
Summary of the invention
In order to solve the problems in the existing technology, the invention proposes a kind of setting based on high-resolution remote sensing image
Vegetables information extraction scheme is applied, can satisfy the needs of facilities vegetable information extraction in image, improves high-resolution remote sensing image
The robustness and universality that facilities vegetable extracts, realize the extraction of facilities vegetable cultivated area.
The present invention provides a kind of facilities vegetable extracting methods based on multidate high-resolution remote sensing image, for being based on
High-resolution remote sensing image extracts the facilities vegetable information in image, includes the following steps：Step 1, it is distant to high-resolution
Sense image is pre-processed, to extract single band image；Step 2 executes image enhancement processing to single band image, thus
Improve the separating capacity of facilities vegetable and other atural object classifications；Step 3 selects predetermined watch window, and to enhancing, treated
Image carries out texture analysis processing, obtains characteristic value texture image to calculate；Step 4 is based on characteristic value texture image, wound
Build the mask image of the building and road easily obscured with facilities vegetable；Step 5, using mask image to original high-resolution
Rate remote sensing image carries out exposure mask, to extract the edge detection line of atural object figure spot；Step 6 counts edge detection line
Morphological operation and thin dividing processing are learned, to generate the extraction result of the gray level image about facilities vegetable；Step 7 is right
The gray level image of facilities vegetable executes binary conversion treatment, extracts the vector figure spot of facilities vegetable and carries out vector and turnstilees at lattice
Reason, to realize the extraction to facilities vegetable information.
Preferably, in the present invention, pretreatment includes at least：Radiation calibration carries out geometric correction, figure using reference images
As fusion, image forward, rule-based progress region cutting.Texture analysis processing is the texture statistics method of gray level co-occurrence matrixes,
And the characteristic value of characteristic value texture image includes at least：Entropy, contrast, auto-correlation, energy, homogeney.
Specifically, it is executed in step 1：Using high-resolution remote sensing image carry .rpb file to remote sensing image into
The operation of row ortho-rectification；Using scaling parameter and spectral response functions, radiation calibration is carried out to the image after ortho-rectification, from
And obtain Reflectivity for Growing Season image；Using reference images, geometric correction is carried out to Reflectivity for Growing Season image；Using GS fusion method,
High-resolution, multispectral panchromatic image carry out fusion treatment；Using Administrative boundaries to image cutting is carried out, to complete shadow
The pretreatment of picture.
Extraly, step 2 can also include：The blue wave band in high-resolution remote sensing image is extracted, histogram equalization is carried out
Change image enhancement processing.
Correspondingly, it is executed in step 2：The data of high-resolution remote sensing image are converted into double type；By high score
The data of resolution remote sensing image are converted to the gray level image in [0,1] section；The data extending of high-resolution remote sensing image is arrived
[0,255] section；The blue wave band extracted in high-resolution remote sensing image is analyzed；Histogram equalization is carried out to blue wave band
Image enhancement processing, to further increase the separating capacity of facilities vegetable Yu other atural object classifications.
It is executed in step 3：Grey measurement levelization operation；Determine watch window；Set step pitch and scanning direction；Calculate line
The gray level co-occurrence matrixes of reason；Calculate the characteristic value of characteristic value texture image；Generate characteristic value texture image.
Specifically, step 4 includes：According to characteristic value texture image, by global threshold split plot design, creation is easily and facility
The mask image of building and road that vegetables are obscured, wherein mask image is used for building and road is distant from high-resolution
It is eliminated in sense image.
It is executed in step 4：Select building, road and the high texture image of other atural object discriminations；Using global threshold
Value segmentation hair creates mask image by building and road for taking out；The building in influencing is rejected using mask image
And road.
Step 5 includes：Exposure mask is carried out to original high-resolution remote sensing image using mask image；Using iteration threshold
Split plot design gets rid of background element；Using edge detection method, the edge of facilities vegetable is extracted, to obtain edge inspection
Survey line, wherein background element includes at least image soil, and edge detection method is canny edge detection method.
Specifically, it is executed in step 5：Background element is got rid of using iterative threshold segmentation method；Using Gaussian filter
Remote sensing image is smoothed；Calculate amplitude and the direction of gradient；Non-maxima suppression is carried out to magnitude image；With double
Thresholding algorithm detection and connection image border.
Step 6 includes：Mathematical morphology operation is carried out to edge detection line, connects image slices vegetarian refreshments；Use edge
Detection line divides the image after global threshold is divided, to obtain the figure spot of facilities vegetable；Using in mathematical morphology
Opening operation gets rid of the phenomenon in flakes in figure spot, to obtain removal image；Using connected domain display mode, referred to by feature
Number is finely divided removal image and cuts, to generate the extraction result of the gray level image about facilities vegetable, wherein mathematics shape
State operation is the expansive working of mathematical morphology, and characteristic index includes area, perimeter, circularity, length ratio, squareness ratio.
In step 6, selects size for 2 " disk " structural element, the swollen of mathematical morphology is executed using structural element
Swollen operation obtains edge detection line, divides the image after global threshold is divided using edge detection line, obtains facility vegetable
The figure spot of dish, using the opening operation in mathematical morphology, the phenomenon in flakes got rid of in figure spot is adopted to obtain removal image
It is shown with connected domain, calculates the index of correlation in connected domain, and be finely divided to removal image by five characteristic indexs
It cuts, generates the extraction result of the gray level image of facilities vegetable.
It is executed in step 7：Binary conversion treatment is carried out to the gray level image of facilities vegetable；It is extracted in ENVI software
The vector figure spot of facilities vegetable and carry out vector turnstile lattice processing；In ArcGIS software, the facilities vegetable extracted is added
Geographic coordinate system, thus result image to the end.
Therefore, compared with prior art, beneficial effect below may be implemented using the present invention：
1) building and road information are extracted using global threshold segmentation, using back such as iterative method Threshold segmentation removal exposed soils
Scape information can guarantee the precision that facilities vegetable extracts using edge detection partitioning algorithm；
2) it compared with traditional supervised classification, unsupervised classification, is extracted using the class object of single width image, is more focused on to light
The in-depth analysis and excavation of spectrum and texture information, achieve the purpose that specific aim is stronger, is obviously improved nicety of grading；
3) using the auxiliary of mask process, Multi-layer technology and local data and expertise, it is ensured that every width image
Extraction accuracy；
4) advantage combined ensure that the extraction advantage of facilities vegetable and the correctness of exact classification；
5) existing facilities vegetable information extracting method is compared, the present invention improves high-resolution remote sensing image facilities vegetable
The robustness and universality of extraction, realize the extraction of facilities vegetable cultivated area.
Detailed description of the invention
Fig. 1 is a kind of facility vegetable based on multidate high-resolution remote sensing image involved in the specific embodiment of the invention
The flow chart of dish extracting method；
Fig. 2 is a kind of facility vegetable based on multidate high-resolution remote sensing image involved in the specific embodiment of the invention
The flow graph of dish extraction concrete operations；
Fig. 3, which is shown, passes through pretreated high-resolution remote sensing image using a certain area；
Fig. 4 (a), Fig. 4 (b), Fig. 4 (c), Fig. 4 (d), Fig. 4 (e) respectively illustrate total using the gray scale in the step of Fig. 1 three
Five kinds of texture eigenvalue images that raw matrix obtains；
Fig. 5 is shown using the image after the mask process in the step of Fig. 1 four；
Fig. 6 shows the result shadow obtained after Threshold segmentation and canny edge detection algorithm using the step of Fig. 1 five
Picture；
Fig. 7 shows the display result figure for carrying out figure spot statistics involved in the step of Fig. 1 six using connected domain；
Fig. 8 shows the involved display result figure that figure spot processing is carried out using shape feature index of the step of Fig. 1 six；
Fig. 9 shows the extraction obtained using the facilities vegetable information approach of the invention based on high-resolution remote sensing image
Result figure.
Specific embodiment
It will be appreciated that continuing to increase for proportion is produced by the intensive vegetables of representative of facilities vegetable in agricultural, because
This need to carry out scientific and normal management to facilities vegetable information.It addresses that need, the invention proposes one kind to be based on high-resolution
The method of the facilities vegetable information extraction of rate remote sensing image, below in conjunction with attached drawing 1-9 and specific embodiment to the present invention into
Row is described in detail.
As shown in Figure 1, the facilities vegetable information extracting method based on high-resolution remote sensing image includes the following steps：
Step 1 carries out pretreatment operation to high-resolution remote sensing image；
Step 2, extract high-resolution remote sensing image in single band, carry out image enhancement processing improve facilities vegetable and
The separating capacity of other atural object classifications；
Step 3, watch window appropriate for enhanced image selection carry out texture analysis, calculate characteristic value texture
Image；
Step 4 creates the mask image of the building, road that are easy to facilities vegetable and obscure according to characteristic value texture image；
Step 5 carries out exposure mask to original image using the mask image that previous step is made, uses Threshold segmentation again
Method is by background removal.The edge of facilities vegetable is extracted using edge detection and obtains edge detection line；
Step 6 carries out mathematical morphology operation to edge detection line, " in flakes " phenomenon therein is removed as much as possible,
Then it is shown using connected domain, is finely divided and is cut by shape feature exponent pair image, generated facilities vegetable gray level image and extract
As a result；
Step 7 carries out binary conversion treatment to facilities vegetable gray level image, extracts facilities vegetable vector figure spot, carries out vector
Lattice of turnstiling processing, realizes the extraction of facilities vegetable information；
Specifically, the specific techniqueflow of step 1~step 7 is as shown in Figure 2.
Step 1, remote sensing image read after carry out radiation calibration, using reference images carry out geometric correction, image co-registration,
Image mosaic, the rule-based pretreatment operation for carrying out the processes such as research area cutting and completing image.
Step 2 extracts the blue wave band in high-resolution remote sensing image, carries out histogram-equalized image enhancing processing and mentions
The separating capacity of high facilities vegetable and other atural object classifications.
Step 3, watch window appropriate for enhanced image selection, using the texture statistics side of gray level co-occurrence matrixes
Method calculates texture eigenvalue image, and characteristic value therein includes：Entropy (Entropy, ENT), contrast (Contrast, CON),
Auto-correlation (Correlation, COR), energy (Energy, ENE), homogeney (Homogemeity, HOM) etc..
Step 4, according to the method that characteristic value texture image is divided by global threshold, to create, to be easy to facilities vegetable mixed
The mask image of the building, road that confuse, can be by building and road from high-resolution remote sensing image by mask image
It is rejected.
Step 5, the high-resolution remote sensing image after exposure mask at this time can will using the method for iterative method Threshold segmentation
The background elements such as the soil in image are removed, and then carry out canny edge detection, obtain the edge detection of atural object figure spot
Line.
Step 6 carries out the expansive working of mathematical morphology to edge detection line, connects pixel, then use side
Image after four Threshold segmentation of edge line segmentation step, obtains the figure spot of facilities vegetable.The opening operation in morphology is continued to use,
" in flakes " phenomenon therein is removed as much as possible, is then shown using connected domain, Ar (area), Perimeter (week are passed through
It is long), Metric (circularity), Pwl (length-width ratio), Pr (squareness ratio) this five characteristic indexs image is finely divided and is cut, generate
Facilities vegetable gray level image extracts result.
Step 7 carries out binary conversion treatment to facilities vegetable gray level image, facility vegetable is then extracted in ENVI software
Dish vector figure spot, carry out vector turnstile lattice processing, the facilities vegetable geography coordinate extracted is added in ArcGIS software
System, obtains result image to the end.
Further, the step 3 specifically includes following steps：
Gray-level quantization
Texture analysis is carried out on the basis of the single band gray level image obtained after step 2, using gray scale symbiosis square
Battle array carries out texture information statistics, since the gray level of image is 256 grades, causes the very big time-consuming of calculation amount long.Therefore exist
Calculate gray level co-occurrence matrixes carries out histogram equalization to image before, then under the premise of not image texture feature first
The gray level of former image is compressed, generally takes 8 grades or 16 grades, to reduce the size of co-occurrence matrix, reduce calculation amount and
Calculate the time.
Determine watch window
The setting of window is extremely important, and can this be the key that extract accurate textures information.But the selection for window
On have certain paradox.If being an area concept for texture, it is necessary to the original embodied by consistency spatially
Then, then watch window acquirement is bigger, and the ability that can detected identity is stronger, otherwise weaker, this is resulted in every class
Borderline region misclassification rate it is larger, and calculation amount can also become larger if window is big.If from the boundary of different texture it is corresponding with
The transition of zone-texture identity and in order to be accurately positioned from the point of view of boundary, it is necessary to it is required that watch window obtain it is small
A little preferably.Difficulty caused by this way is exactly that window is too small, then can occur accidentally dividing in same textured inner.In general,
After image size determines, calculation window just determines therewith.
Set step pitch and scanning direction
Gray level co-occurrence matrixes quickly change in fine textures with distance, and with distance, then variation is slow in rough grain
Slowly, it is however generally that, for the biggish distance of smooth grain, preferable effect can be obtained for the lesser distance of rough grain
Fruit.In high-resolution remote sensing image, step pitch d takes 1, scanning direction θ to take 0 °, 45 °, 90 °, 135 °.
Calculate the gray level co-occurrence matrixes of texture
Gray level co-occurrence matrixes are defined as the joint probability distribution of pixel pair, are a symmetrical matrixes, not only reflect image ash
Degree also reflects the position between identical gray-level pixels in adjacent direction, adjacent spaces, the integrated information of amplitude of variation
Distribution characteristics is set, is the basis for calculating textural characteristics.A bit (x, y) is arbitrarily taken in the picture and deviates its a bit (x+a, y+
B) (wherein a, b are integer, artificially defined) constitutes point pair.If the gray value of the point pair is (f1, f2), then enables point (x, y) whole
It is moved on width image, then can obtain different (f1, f2) values.If the maximum gray scale of image is L, then the group of f1 and f2 amounts to
There is L*L kind.For entire image, the number of every kind of (f1, f2) value appearance is counted, is then arranged in a square matrix, then use
The total degree that (f1, f2) occurs is normalized to the probability P (f1, f2) occurred, and resulting matrix is exactly gray scale symbiosis
Matrix.Wherein, the calculating of gray level co-occurrence matrixes is as follows：
Wherein, d indicates pixel separation, and (k, 1) and (m, n) is respectively the pixel coordinate after first pixel and offset, wherein k,
M is ordinate, and D is image range.
Calculate texture eigenvalue：Entropy (Entropy, ENT), contrast (Contrast, CON), auto-correlation
(Correlation, COR), energy (Energy, ENE), homogeney (Homogemeity, HOM).
Following 5 characteristic value can be calculated according to gray level co-occurrence matrixes, formula is as follows：
Entropy (Entropy, ENT)：
Contrast (Contrast, CON)：
Wherein：| i-j |=n.
Auto-correlation (Correlation, COR ')：
Wherein：μx, μyAnd δx, δyRespectively mx, myMean value and standard deviation, mxIt is the sum of every row element in matrix P； myIt is
The sum of every column element in matrix P.
Energy (Energy, ENE ')：
Homogeney (Homogemeity, HOM)：
P (i, j) indicates the element value in GLCM in 5 above-mentioned formulas.
Generate textural characteristics image
The main thought of textural characteristics video generation is：The sub-image formed with each watch window passes through texture spy
After levying the characteristic value that calculation procedure calculates watch window image greyscale co-occurrence matrix and texture, this window line then will be represented
Reason characteristic value is assigned to the central point of window, and this completes the calculating of the texture eigenvalue of a watch window.Then window is mobile
One pixel forms another watch window image, repeats and calculates new co-occurrence matrix and texture eigenvalue.And so on, this
Sample whole image just will form a textural characteristics value matrix being made by texture eigenvalue, then will in MATLAB
This texture eigenvalue matrix conversion is shown at textural characteristics image.
Further, the step 5 specifically includes following steps：
The background elements such as exposed soil in image are removed using the method for iterative method Threshold segmentation
Iterative method is using based on the thought approached, and its step are as follows：
The maximum gradation value and minimum gradation value for finding out image, are denoted as ZMAX and ZMIN respectively, enable initial threshold T=
(ZMAX+ZMIN)/2；
Foreground and background is divided the image into according to threshold value T, finds out the average gray value ZO and ZB of the two respectively；
Find out new threshold value T=(ZO+ZB)/2；
If two average gray values ZO and ZB no longer change (or T no longer changes), T is threshold value；Otherwise turn 2) iteration
It calculates.It calculates always and is then removed the background elements such as exposed soil.
High-resolution remote sensing image is smoothed using Gaussian filter
Indicate that (line number i, the pixel value of row number j) use separable filtering to pixel in original image using I [i, j]
Method asks image and Gaussian filter convolution, it is obtaining the result is that one smoothed data array, formula are as follows：
S [i, j]=G [σ] * I [i, j]
Pixel value after wherein S [i, j] indicates smooth；G [σ] is Gaussian function；σ is stroll function, it controls smooth journey
Degree.
Calculate amplitude and the direction of gradient
2*2 first difference point approximate expression can be used to calculate x and y local derviation in the gradient of smoothed data array S [i, j]
Two several array P [i, j] and Q [i, j], formula is as follows：
P [i, j] ≈ (S [i, j+1]-S [i, j]+S [i+1, j+1]-S [i+1, j])/2
Q [i, j] ≈ (S [i, j]-S [i+1, j]+S [i, j+1]-S [i+1, j+1])/2
The mean value of finite difference is sought, in this 2*2 square so that same point in the picture calculates the local derviation of x and y
Number gradient.Amplitude M [i, j] and azimuth angle theta [i, j] can be calculated with rectangular co-ordinate to polar coordinate transformation formula, formula
As follows：
θ [i, j]=arctan (Q [i, j]/P [i, j])
Wherein arctan function contains two parameters, it indicates an angle, and value range is whole circumference range.
Non-maxima suppression is carried out to magnitude image
The purpose of non-maxima suppression is the most of non-edge point rejected in the result images that previous step is calculated.It is former
Reason is exactly by the eight neighborhood of pixel to determine whether this pixel is set to marginal point or background colour.
Image border is detected and connected with dual threashold value-based algorithm
The image generated to non-maxima suppression is detected to reduce the quantity at false edge using dual threashold value-based algorithm
Edge, generally requires a setting one high threshold TH and Low threshold TL, general high threshold and Low threshold ratio 2: 1 to 3: 1 it
Between.If the gradient magnitude of a certain location of pixels is more than high threshold, pixel is left edge pixel；If a certain pixel
The gradient magnitude of position is less than Low threshold, then pixel is excluded；If the amplitude of a certain location of pixels is between two thresholds,
The pixel is only retained when being connected to a pixel for being higher than high threshold.Dual threashold value-based algorithm can spell candidate pixel point
It is connected into profile, when formation of profile uses hysteresis threshold algorithm to these pixels.
Further, the step 6 specifically includes following steps：
Select size for 4 " disk " structural element
Structural element can regard a small image as, open commonly used in the expansion in the morphology operations of image, burn into
Closed operation etc..When being filtered with mathematical morphology to image, most important is exactly the selection for being structural element, and one
As for structural element, it is thus necessary to determine that type (shape), the size (scale) of structural element.Generally for structural element
In type (shape) for, be divided into：'arbitrary','pair','line','square','rectangle',
Diamond ', ' disk ' etc..Wherein the selection of structural element above, in high resolution image class detailed information it is abundant, it is right
For facilities vegetable greenhouse, which is not present significant anisotropy feature.Disk (circle) element tool
There is the feature of isotropic, furthermore the structural element can be removed selectively and formulate the noise of scale and uncorrelated in image
Silhouette target and retain other useful information, therefore circular configuration element processing high resolution remote sensing image in have very
Big advantage.
Morphologic expansive working is executed using structural element, obtains edge line
Using canny algorithm edge detection it is complete after, some of which edge is there is being not attached to together, institute
These places to be attached using the expansion algorithm in morphology.Expand (imdilate) operation definition：
Wherein：A in formula is input picture, and B is structural element, and x is the mobile distance of operation window.Dilation operation can be with
All background dots contacted with object are merged into object, target is increased, the cavity in target can be replenished.Concrete operations
For：With each of structural element scan image pixel, the pixel covered with each pixel in structural element with it
With operation is done, if being all 0, otherwise it is 1 which, which is 0,.
Using the image after five Threshold segmentation of edge line segmentation step, facilities vegetable figure spot is obtained
The image that image and exposure mask after step 5 Threshold segmentation is crossed mutually and operation, make some with facilities vegetable phase
Object of obscuring even delete and separates the facilities vegetable that some of them are ined succession, and obtains facilities vegetable as much as possible
Figure spot.
It, will wherein phenomenon be eliminated as much as " in flakes " using opening operation in morphology
" in flakes " phenomenon of facilities vegetable can be mitigated using the opening operation in morphology.Open (imopen) operation definition：
Opening operation is to pass sequentially through corrosion and dilation operation, and the minutia to become clear in smoothed image can be than definition
The small burr of structural element filter, the boundary of smooth larger object cuts off elongated overlap joint and plays centrifugation, while simultaneously
Unobvious its area of change.
It is shown using connected domain
The image for not only containing facilities vegetable after treatment but also containing " noise " can be distinguished them by marking
Come.The simple and effective method in each region is exactly to check the connection of each pixel pixel adjacent thereto in image after label segmentation
Property.After the processing for carrying out above-mentioned steps, the background area pixel value of image is 0, and the pixel value of target area is 1.It is set in algorithm
It is set to and piece image from left to right, be scanned from the top down, the figure spot to be marked just is needed to mark currently just
The connectivity of scanned pixel and several neighbor pixels being scanned before it.Picture is carried out using the algorithm of 4 connections
Element scanning.
If current pixel value is 0, the position of next scanning is moved to.If current pixel value is 1, the left side and upper is checked
Two adjacent pixels (because the two pixels can be scanned before current pixel) on side.The two pixel values and label
Combination just will appear four kinds of situations needs and account for.
If 1) their pixel value is all 0, (opening for a new connected domain is indicated to the new label of the pixel one at this time
Begin).
If 2) only one pixel value is 1 between them, current pixel is labeled as the label of 1 pixel value at this time.
3) if their pixel value is all 1 and label is identical, the label of current pixel is equal to the label at this time.
4) if their pixel value is 1 and label is different, lesser value therein is assigned to current pixel, later from another
While each backtracking therein needs carrying out four above-mentioned judgements until the beginning pixel for tracing back to region.
It ensures that all connected domains are all labeled in this way to come out, it is then different by being assigned to different labels again
Color or label is can be completed into plus frame in it.
Calculate connected domain in the index of correlation, by Ar (area), Perimeter (perimeter), Metric (circularity),
This five characteristic indexs of Pwl (length-width ratio), Pr (squareness ratio), which are finely divided image, cuts, and generates facilities vegetable gray level image and extracts
As a result
Relevant index can be calculated after all connected domains are all marked, and utilize these correlated characteristic exponent pairs
Image, which is finely divided, to be cut, and the gray level image for generating facilities vegetable extracts result.
After morphology operations, the extraction accuracy of the facilities vegetable greenhouse after can dividing the image into is improved, still
Still the accidentally point phenomenon that the atural objects such as some streets and buildings are mutually obscured still is had.Pass through analysis, some of which building
Compared with facilities vegetable greenhouse, building normally behaves as smaller rectangular of length-width ratio, therefore can use squareness ratio
Pr, length-width ratio Pwl and circularity Metric are split；The small figure spot in irregular shape of some of which area can be with
It is removed using two parameters of area Ar and perimeter Perimeter.
1. circularity Metric：
Wherein S indicates that figure spot region area, P are figure spot area circumference.0 < I≤1, I is bigger, then region is closer to round.
2. length-width ratio Pwl：
Wherein, a is the long side length in figure spot region, and b is the bond length in figure spot region.
3. squareness ratio Pr：
Wherein, S is figure spot region area, indicates region minimum circumscribed rectangle area.
Figure spot region area and figure spot region minimum circumscribed rectangle area ratio.The similar of a region rectangle can be measured
Degree reflects full level of the figure spot in its boundary rectangle.Value range is：[0,1], the bigger expression of rectangular degree
Region is closer to rectangle.
Note that the present invention CPU be Core (TM) i5-4590 3.30GHz, the professional edition system of memory 4GB, Windows 8
Upper use is programmed realization emulation on MATLAB R2016a software.In each specific embodiment of the invention, one is selected
Pretreated high-resolution remote sensing image is passed through in a a certain area containing 4 atural object classifications, as shown in Figure 3.By step
After the processing of histogram-equalized image enhancing in two, using the texture image extraction step of step 3 by gray level co-occurrence matrixes
In characteristic value carry out statistics and generate obtained five kinds of characteristic value texture images, as shown in Figure 4.Then step 4 kind pass through by
Feature texture image is handled, and the mask image of corresponding building and road is created, wherein the image after exposure mask is such as
Shown in Fig. 5.Then using the knot obtained after Threshold segmentation and canny edge detection algorithm in image of the step 5 after exposure mask
Fruit image, as shown in Figure 6.In the step of being embodied six, the knot of connected domain figure spot statistics is carried out by the way of 4 neighborhoods
Then " noise " figure spot therein as shown in fig. 7, is removed to obtain that treated is aobvious by fruit image by shape feature index
Show result figure, as shown in Figure 8.The most red extraction knot of the last facilities vegetable information of the invention based on high-resolution remote sensing image
Fruit, as shown in Figure 9.The result of this extraction and the result for manually visualizing interpretation can achieve 90% or more precision, it was demonstrated that
Design method of the present invention has excellent extraction effect.
In conclusion the present invention extracts building and road information using global threshold segmentation, using iterative method threshold value point
It prescinds except background informations such as exposed soils, the precision that facilities vegetable extracts can be guaranteed using edge detection partitioning algorithm.
In addition, the present invention is extracted using the class object of single width image, compared with traditional supervised classification, unsupervised classification more
Focus on the in-depth analysis and excavation to spectrum and texture information, achievees the purpose that specific aim is stronger, is obviously improved nicety of grading.
Along with mask process, Multi-layer technology, and the auxiliary of local data and expertise, it is ensured that the extraction of every width image
Precision.In conjunction with advantage ensure that the extraction advantage of facilities vegetable and the correctness of exact classification.Compared to existing facility vegetable
Dish information extracting method, robustness of the invention and universality are more preferable.
The above is most basic specific embodiment of the invention, for those skilled in the art,
Without departing from the principles of the invention, several improvement and polishing can also be made, these are improved and polishing is also regarded as this hair
Bright protection scope.It is not specified in the present invention and partly belongs to techniques known.
Claims (14)
1. a kind of facilities vegetable extracting method based on multidate high-resolution remote sensing image, for being based on high-definition remote sensing shadow
Picture extracts the facilities vegetable information in image, which is characterized in that include the following steps：
Step 1 pre-processes the high-resolution remote sensing image, to extract single band image；
Step 2 executes image enhancement processing to the single band image, to improve the facilities vegetable and other ground species
Other separating capacity；
Step 3 selects predetermined watch window, texture analysis processing is carried out to enhancing treated image, so that it is special to calculate acquisition
Value indicative texture image；
Step 4 is based on the characteristic value texture image, creates covering for the building and road easily obscured with the facilities vegetable
Film image；
Step 5 carries out exposure mask to the original high-resolution remote sensing image using the mask image, to extract ground
The edge detection line of object figure spot；
Step 6 carries out mathematical morphology operation and thin dividing processing to the edge detection line, sets to generate about described
Apply the extraction result of the gray level image of vegetables；
Step 7 executes binary conversion treatment to the gray level image of the facilities vegetable, extracts the polar plot of the facilities vegetable
Spot simultaneously carries out vector and turnstilees lattice processing, to realize extraction to facilities vegetable information.
2. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, the pretreatment includes at least：Radiation calibration, using reference images carry out geometric correction, image co-registration, image forward,
Rule-based progress region cutting.
3. the facilities vegetable extracting method according to claim 2 based on multidate high-resolution remote sensing image, feature
It is, executes in said step 1：
Ortho-rectification operation is carried out to remote sensing image using the .rpb file that the high-resolution remote sensing image carries；
Using scaling parameter and spectral response functions, radiation calibration is carried out to the image after ortho-rectification, so that it is anti-to obtain earth's surface
Penetrate rate image；
Using reference images, geometric correction is carried out to the Reflectivity for Growing Season image；
Using GS fusion method, high-resolution, multispectral panchromatic image carry out fusion treatment；
Using Administrative boundaries to image cutting is carried out, to complete the pretreatment of image.
4. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, the step 2 further includes：
The blue wave band in the high-resolution remote sensing image is extracted, histogram-equalized image enhancing processing is carried out.
5. the facilities vegetable extracting method according to claim 4 based on multidate high-resolution remote sensing image, feature
It is, executes in the step 2：
The data of the high-resolution remote sensing image are converted into double type；
The data of the high-resolution remote sensing image are converted to the gray level image in [0,1] section；
By the data extending of the high-resolution remote sensing image to [0,255] section；
The blue wave band extracted in the high-resolution remote sensing image is analyzed；
Histogram-equalized image enhancing processing is carried out to the blue wave band, to further increase the facilities vegetable and other
The separating capacity of atural object classification.
6. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, the texture statistics method that the texture analysis is handled as gray level co-occurrence matrixes, and the feature of the characteristic value texture image
Value includes at least：Entropy, contrast, auto-correlation, energy, homogeney.
7. the facilities vegetable extracting method according to claim 6 based on multidate high-resolution remote sensing image, feature
It is, is executed in the step 3：
Grey measurement levelization operation；Determine the watch window；Set step pitch and scanning direction；Calculate the gray scale symbiosis square of texture
Battle array；Calculate the characteristic value of the characteristic value texture image；Generate the characteristic value texture image.
8. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, the step 4 includes：
The building easily obscured with the facilities vegetable is created by global threshold split plot design according to the characteristic value texture image
The mask image of object and road,
Wherein, the mask image from the high-resolution remote sensing image for eliminating building and road.
9. the facilities vegetable extracting method according to claim 8 based on multidate high-resolution remote sensing image, feature
It is, is executed in the step 4：
Select building, road and the high texture image of other atural object discriminations；
Divided using global threshold and sent out, by the building and road for taking out, creates the mask image；
Building and road in influencing are rejected using the mask image.
10. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, the step 5 includes：
Exposure mask is carried out to the original high-resolution remote sensing image using the mask image；
Background element is got rid of using iterative threshold segmentation method；
Using edge detection method, the edge of the facilities vegetable is extracted, so that the edge detection line is obtained,
Wherein, the background element includes at least image soil, and the edge detection method is canny edge detection method.
11. the facilities vegetable extracting method according to claim 10 based on multidate high-resolution remote sensing image, special
Sign is, executes in the step 5：
Background element is got rid of using iterative threshold segmentation method；
The remote sensing image is smoothed using Gaussian filter；
Calculate amplitude and the direction of gradient；
Non-maxima suppression is carried out to magnitude image；
Image border is detected and connected with dual threashold value-based algorithm.
12. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, the step 6 includes：
Mathematical morphology operation is carried out to the edge detection line, connects image slices vegetarian refreshments；
Divide the image after global threshold is divided with the edge detection line, to obtain the figure spot of the facilities vegetable；
Using the opening operation in the mathematical morphology, the phenomenon in flakes in figure spot is got rid of, to obtain removal image；
The removal image is finely divided and is cut by characteristic index using connected domain display mode, to generate about described
The extraction of the gray level image of facilities vegetable as a result,
Wherein, mathematical morphology operation is the expansive working of mathematical morphology, the characteristic index include area, perimeter,
Circularity, length ratio, squareness ratio.
13. the facilities vegetable extracting method according to claim 12 based on multidate high-resolution remote sensing image, special
Sign is, executes in the step 6：
Select size for 2 " disk " structural element；
The expansive working that the mathematical morphology is executed using structural element obtains the edge detection line；
Divide the image after global threshold is divided using the edge detection line, obtains the figure spot of the facilities vegetable；
Using the opening operation in the mathematical morphology, the phenomenon in flakes in figure spot is got rid of, to obtain removal image；
It is shown using connected domain；
The index of correlation in the connected domain is calculated, and the removal image is finely divided by five characteristic indexs and is cut, it is raw
At the extraction result of the gray level image of the facilities vegetable.
14. the facilities vegetable extracting method according to claim 1 based on multidate high-resolution remote sensing image, feature
It is, is executed in the step 7：
Binary conversion treatment is carried out to the gray level image of the facilities vegetable；
The vector figure spot of the facilities vegetable is extracted in ENVI software and carry out vector turnstile lattice processing；
In ArcGIS software, the geographic coordinate system of the facilities vegetable extracted is added, to obtain result to the end
Image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810592833.XA CN108830844B (en) | 2018-06-11 | 2018-06-11 | Facility vegetable extraction method based on multi-temporal high-resolution remote sensing image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810592833.XA CN108830844B (en) | 2018-06-11 | 2018-06-11 | Facility vegetable extraction method based on multi-temporal high-resolution remote sensing image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108830844A true CN108830844A (en) | 2018-11-16 |
| CN108830844B CN108830844B (en) | 2021-09-10 |
Family
ID=64144933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810592833.XA Active CN108830844B (en) | 2018-06-11 | 2018-06-11 | Facility vegetable extraction method based on multi-temporal high-resolution remote sensing image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108830844B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109685081A (en) * | 2018-12-27 | 2019-04-26 | 中国土地勘测规划院 | A kind of joint change detecting method of Remotely sensed acquisition black fallow |
| CN109934122A (en) * | 2019-02-21 | 2019-06-25 | 北京以萨技术股份有限公司 | A kind of remote sensing image ship detecting method based on deep learning |
| CN110276797A (en) * | 2019-07-01 | 2019-09-24 | 河海大学 | A kind of area of lake extracting method |
| CN111259797A (en) * | 2020-01-16 | 2020-06-09 | 南开大学 | Iterative remote sensing image road extraction method based on points |
| CN113610013A (en) * | 2021-08-10 | 2021-11-05 | 四川易方智慧科技有限公司 | Method for extracting building outline based on RGB (Red Green blue) wave bands of high-definition remote sensing image |
| CN115346162A (en) * | 2022-10-19 | 2022-11-15 | 南京优佳建筑设计有限公司 | Indoor monitoring-based real-time monitoring method for water seepage of underground building wall |
| CN116385472A (en) * | 2023-06-07 | 2023-07-04 | 深圳市锦红兴科技有限公司 | Hardware stamping part deburring effect evaluation method |
| CN117237384A (en) * | 2023-11-16 | 2023-12-15 | 潍坊科技学院 | Visual detection method and system for intelligent agricultural planted crops |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8369622B1 (en) * | 2009-10-29 | 2013-02-05 | Hsu Shin-Yi | Multi-figure system for object feature extraction tracking and recognition |
| CN105184251A (en) * | 2015-08-31 | 2015-12-23 | 中国科学院遥感与数字地球研究所 | Water bloom area identification method and device based on high-resolution-I satellite image |
| CN106650812A (en) * | 2016-12-27 | 2017-05-10 | 辽宁工程技术大学 | City water body extraction method for satellite remote sensing image |
| CN107154044A (en) * | 2017-03-27 | 2017-09-12 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of dividing method of Chinese meal food image |
| CN107862667A (en) * | 2017-11-23 | 2018-03-30 | 武汉大学 | A kind of city shadow Detection and minimizing technology based on high-resolution remote sensing image |
-
2018
- 2018-06-11 CN CN201810592833.XA patent/CN108830844B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8369622B1 (en) * | 2009-10-29 | 2013-02-05 | Hsu Shin-Yi | Multi-figure system for object feature extraction tracking and recognition |
| CN105184251A (en) * | 2015-08-31 | 2015-12-23 | 中国科学院遥感与数字地球研究所 | Water bloom area identification method and device based on high-resolution-I satellite image |
| CN106650812A (en) * | 2016-12-27 | 2017-05-10 | 辽宁工程技术大学 | City water body extraction method for satellite remote sensing image |
| CN107154044A (en) * | 2017-03-27 | 2017-09-12 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of dividing method of Chinese meal food image |
| CN107862667A (en) * | 2017-11-23 | 2018-03-30 | 武汉大学 | A kind of city shadow Detection and minimizing technology based on high-resolution remote sensing image |
Non-Patent Citations (2)
| Title |
|---|
| ELAHE KHESALI ET AL: "Semi Automatic Road Extraction by Fusion of High Resolution Optical and Radar Images", 《J INDIAN SOC REMOTE SENS》 * |
| 李黔湘: "基于高分辨遥感数据的农业大棚面积提取及分析——以北京市大兴区为例", 《北京水务》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109685081A (en) * | 2018-12-27 | 2019-04-26 | 中国土地勘测规划院 | A kind of joint change detecting method of Remotely sensed acquisition black fallow |
| CN109685081B (en) * | 2018-12-27 | 2020-07-24 | 中国土地勘测规划院 | Combined change detection method for remote sensing extraction of abandoned land |
| CN109934122A (en) * | 2019-02-21 | 2019-06-25 | 北京以萨技术股份有限公司 | A kind of remote sensing image ship detecting method based on deep learning |
| CN110276797B (en) * | 2019-07-01 | 2022-02-11 | 河海大学 | Lake area extraction method |
| CN110276797A (en) * | 2019-07-01 | 2019-09-24 | 河海大学 | A kind of area of lake extracting method |
| CN111259797A (en) * | 2020-01-16 | 2020-06-09 | 南开大学 | Iterative remote sensing image road extraction method based on points |
| CN113610013A (en) * | 2021-08-10 | 2021-11-05 | 四川易方智慧科技有限公司 | Method for extracting building outline based on RGB (Red Green blue) wave bands of high-definition remote sensing image |
| CN115346162A (en) * | 2022-10-19 | 2022-11-15 | 南京优佳建筑设计有限公司 | Indoor monitoring-based real-time monitoring method for water seepage of underground building wall |
| CN115346162B (en) * | 2022-10-19 | 2022-12-13 | 南京优佳建筑设计有限公司 | Indoor monitoring-based real-time monitoring method for water seepage of underground building wall |
| CN116385472A (en) * | 2023-06-07 | 2023-07-04 | 深圳市锦红兴科技有限公司 | Hardware stamping part deburring effect evaluation method |
| CN116385472B (en) * | 2023-06-07 | 2023-08-08 | 深圳市锦红兴科技有限公司 | Hardware stamping part deburring effect evaluation method |
| CN117237384A (en) * | 2023-11-16 | 2023-12-15 | 潍坊科技学院 | Visual detection method and system for intelligent agricultural planted crops |
| CN117237384B (en) * | 2023-11-16 | 2024-02-02 | 潍坊科技学院 | Visual detection method and system for intelligent agricultural planted crops |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108830844B (en) | 2021-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108830844A (en) | A kind of facilities vegetable extracting method based on multidate high-resolution remote sensing image | |
| Kuang et al. | A comparative analysis of megacity expansions in China and the US: Patterns, rates and driving forces | |
| Li et al. | An index and approach for water extraction using Landsat–OLI data | |
| Pu et al. | Object-based urban detailed land cover classification with high spatial resolution IKONOS imagery | |
| CN103824077B (en) | Urban impervious layer rate information extraction method based on multi-source remote sensing data | |
| CN112381013B (en) | Urban vegetation inversion method and system based on high-resolution remote sensing image | |
| CN107392133B (en) | Desert plant remote sensing identification method using object-oriented multi-source information fusion | |
| CN103093233A (en) | Forest classification method based on object-oriented high-resolution remote sensing image | |
| González-Yebra et al. | Methodological proposal to assess plastic greenhouses land cover change from the combination of archival aerial orthoimages and Landsat data | |
| CN106548141A (en) | A kind of object-oriented farmland information extraction method based on the triangulation network | |
| Sun et al. | Coupling spectral unmixing and multiseasonal remote sensing for temperate dryland land-use/land-cover mapping in Minqin County, China | |
| Dai et al. | Spatio-temporal pattern of urban land cover evolvement with urban renewal and expansion in Shanghai based on mixed-pixel classification for remote sensing imagery | |
| CN107564016A (en) | A kind of Multi-Band Remote Sensing Images segmentation of integrally object light spectrum information and labeling method | |
| CN110070545A (en) | A kind of method that textural characteristics density in cities and towns automatically extracts cities and towns built-up areas | |
| CN113780307A (en) | Method for extracting blue-green space information with maximum regional year | |
| CN107121681B (en) | Residential area extraction system based on high score satellite remote sensing date | |
| CN106780586A (en) | A kind of solar energy potential evaluation method based on ground laser point cloud | |
| Mostafa et al. | Corresponding regions for shadow restoration in satellite high-resolution images | |
| Patidar et al. | A rule-based spectral unmixing algorithm for extracting annual time series of sub-pixel impervious surface fraction | |
| Van de Voorde et al. | Extraction of land use/land cover related information from very high resolution data in urban and suburban areas | |
| CN110705449A (en) | Land utilization change remote sensing monitoring analysis method | |
| CN109635828A (en) | A kind of typical geographical national conditions elements recognition system and method in ecological protection red line area | |
| Zhang | Modeling net primary productivity of wetland with a satellite-based light use efficiency model | |
| Yadav et al. | Urban tree canopy detection using object-based image analysis for very high resolution satellite images: A literature review | |
| CN110210376A (en) | A kind of high score remote sensing image the Northwest agricultural film extracting method based on deep learning |
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 |